{"id": "2406.07775", "title": "Self-attention-based non-linear basis transformations for compact latent space modelling of dynamic optical fibre transmission matrices", "abstract": "Multimode optical fibres are hair-thin strands of glass that efficiently transport light. They promise next-generation medical endoscopes that provide unprecedented sub-cellular image resolution deep inside the body. However, confining light to such fibres means that images are inherently scrambled in transit. Conventionally, this scrambling has been compensated by pre-calibrating how a specific fibre scrambles light and solving a stationary linear matrix equation that represents a physical model of the fibre. However, as the technology develops towards real-world deployment, the unscrambling process must account for dynamic changes in the matrix representing the fibre's effect on light, due to factors such as movement and temperature shifts, and non-linearities resulting from the inaccessibility of the fibre tip when inside the body. Such complex, dynamic and nonlinear behaviour is well-suited to approximation by neural networks, but most leading image reconstruction networks rely on convolutional layers, which assume strong correlations between adjacent pixels, a strong inductive bias that is inappropriate for fibre matrices which may be expressed in a range of arbitrary coordinate representations with long-range correlations. We introduce a new concept that uses self-attention layers to dynamically transform the coordinate representations of varying fibre matrices to a basis that admits compact, low-dimensional representations suitable for further processing. We demonstrate the effectiveness of this approach on diverse fibre matrix datasets. We show our models significantly improve the sparsity of fibre bases in their transformed bases with a participation ratio, p, as a measure of sparsity, of between 0.01 and 0.11. Further, we show that these transformed representations admit reconstruction of the original matrices with < 10% reconstruction error, demonstrating the invertibility.", "date": "2024-06-11", "categories": []}
{"id": "2606.15648", "title": "Fusing Transferred Priors and Physics-based Decomposition for Underwater Image Enhancement", "abstract": "The underwater images are captured within diverse water-medium conditions, leading to complex degradation, including color bias, low contrast, and blur effect. Recently, learning-based methods have demonstrated their potential for underwater image enhancement (UIE). However, most of the previous work focus on the training strategy or network design to make the enhanced result aligned well with the labels in datasets, ignoring that the labels are selected from the enhanced results of previous UIE methods and these pseudo-labels are noisy. Consequently, the performance of their models is not satisfactory to a certain extent. However, collecting the true labels of the underwater images is challenging. In this work, we propose a transfer learning-based UIE that does not require underwater images to have paired noisy or true labels for learning. Instead, the UIE task is first divided into global color correction, haze removal, and background noise suppression following the underwater physics. Then multiple types of prior from other vision tasks are leveraged as cross-domain supervision in each step. In this way, a novel UIE is available via transfer learning, and the physics-aligned UIE decomposition provides theoretical soundness. Qualitative and quantitative experiments demonstrate that our proposal based on physics and priors fusion achieves SOTA performance in the UIE task and effectively boosts downstream vision tasks, significantly outperforming benchmark methods. Project repo: https://github.com/Haru2022/P2-UIE.", "date": "2026-06-14", "categories": []}
{"id": "2606.19617", "title": "GB-LSR: A Fast Local Spectral Image Representation with a Single Global Bandwidth for Continuous Reconstruction and Super-Resolution", "abstract": "We present GB-LSR (Global-Bandwidth Local Spectral Representation), a fixed-grid local spectral representation for continuous image reconstruction. The image domain is partitioned into non-overlapping square patches, each carrying coefficients for a truncated Fourier basis predicted from shared convolutional-encoder features. A single trainable scalar bandwidth is shared globally across all patches and images, and reconstruction at any continuous coordinate is a fixed-size basis contraction whose cost is independent of image size. We study three bandwidth-handling variants: a trainable global scalar (main), a fixed global scalar, and a per-patch bandwidth field. On a standardized native-reconstruction benchmark across Kodak, Set14, and Urban100, the main variant outperforms matched-budget amortized LIIF / LTE / WIRE re-implementations by 2.8-3.6 dB PSNR and 0.11-0.15 LPIPS, while running at roughly one-quarter of the slowest baseline's inference cost. The single global scalar suffices empirically: per-patch adaptive-bandwidth alternatives do not improve over it on either a closed-form locality diagnostic or an end-to-end ablation. In a separate arbitrary-scale super-resolution (ASR) extension, GB-LSR achieves competitive PSNR-Y under a canonical-style SR protocol and runs 1.44x faster than LIIF-RDN and 3.25x faster than LTE-SwinIR at x4; within the same extension, a variant trained and evaluated without 4-corner local-ensemble averaging gives a 1.77x speedup with 35% lower peak memory and negligible PSNR change, while additionally widening the RDN encoder from 64 to 96 channels gives a small positive PSNR shift with a 1.58x speedup and 31% lower peak memory. Native-reconstruction claims are scoped to the matched-budget amortized protocol, and ASR claims are scoped to a separate canonical-style SR protocol.", "date": "2026-06-17", "categories": []}
{"id": "2606.19802", "title": "Flow Map Denoisers: Traversing the Distortion-Perception Plane for Inverse Problems", "abstract": "Image restoration faces a fundamental tradeoff: methods that minimize error produce blurry reconstructions, while those that maximize perceptual quality yield sharp but less faithful images. Existing approaches either commit to a single operating point on this distortion perception (DP) frontier or require paired-data supervision, auxiliary models, or hyperparameter tuning of the sampler to access different points. We show that flow map models, a recent extension of flow matching for few-step sampling that learns an average field, implicitly define a one-parameter family of denoisers that continuously spans the DP frontier. The lookahead parameter t acts as a control knob between the MMSE and perceptual regimes. For Gaussian targets, we prove that varying t exactly recovers the optimal DP frontier; for natural images, we observe similar behavior empirically. Within a Plug-and-Play solver, the same mechanism extends to general inverse problems, where it controls a tradeoff between perceptual alignment and data consistency. Despite the lack of exact optimality guarantees in this setting, a single trained flow map spans the DP tradeoff, matching or exceeding specialized baselines at both extremes. Extensive experiments on CelebA ($128\\times 128$) and AFHQ ($256\\times 256$) across several linear and nonlinear inverse tasks validate our findings.", "date": "2026-06-18", "categories": []}
{"id": "2606.19985", "title": "Vision-Reasoning-Guided Occlusion Removal from Light Fields", "abstract": "Occlusion-robust scene recovery remains a major challenge in computational imaging, particularly in natural environments where dense foreground vegetation severely limits visibility. We propose a vision-reasoning-guided light field occlusion removal framework that combines the visibility recovery capability of light field integration (LFI) with the semantic reasoning capacity of vision-language models (VLMs). Multi-view observations are first integrated via LFI to suppress foreground occlusions and produce an initial visibility-enhanced representation. A VLM is then incorporated as a conditional semantic prior to restore degraded structures and recover fine details, guided by the observed measurements. To improve recovery consistency and reduce hallucination artifacts, we introduce a multi-sample fusion strategy that aggregates multiple generated hypotheses into a unified estimate. Experimental results on synthetic and real-world datasets demonstrate state-of-the-art performance, achieving the highest average SSIM across four synthetic light field benchmark scenes (4-Syn) and strong generalization across structured and unstructured acquisition settings. These results highlight the effectiveness of combining physical imaging constraints with vision-language reasoning for robust perception under severe occlusion, with applicability to search-and-rescue and exploratory robotic navigation.", "date": "2026-06-18", "categories": []}
{"id": "2508.11211", "title": "Efficient Image-to-Image Schr\\\"odinger Bridge for CT Field of View Extension", "abstract": "Computed tomography (CT) is a cornerstone imaging modality for non-invasive, high-resolution visualization of internal anatomical structures. However, when the scanned object exceeds the scanner's field of view (FOV), projection data are truncated, resulting in incomplete reconstructions and pronounced artifacts near FOV boundaries. Conventional reconstruction algorithms struggle to recover accurate anatomy from such data, limiting clinical reliability. Deep learning approaches have been explored for FOV extension, with diffusion generative models representing the latest advances in image synthesis. Yet, conventional diffusion models are computationally demanding and slow at inference due to their iterative sampling process. To address these limitations, we propose an efficient CT FOV extension framework based on the image-to-image Schr\\\"odinger Bridge (I$^2$SB) diffusion model. Unlike traditional diffusion models that synthesize images from pure Gaussian noise, I$^2$SB learns a direct stochastic mapping between paired limited-FOV and extended-FOV images. This direct correspondence yields a more interpretable and traceable generative process, enhancing anatomical consistency and structural fidelity in reconstructions. I$^2$SB achieves superior quantitative performance, with root-mean-square error (RMSE) values of 49.8 HU on simulated noisy data and 152.0 HU on real data, outperforming state-of-the-art diffusion models such as conditional denoising diffusion probabilistic models (cDDPM) and patch-based diffusion methods. Moreover, its one-step inference enables reconstruction in just 0.19 s per 2D slice, representing over a 700-fold speedup compared to cDDPM (135 s) and surpassing DiffusionGAN (0.58 s), the second fastest. This combination of accuracy and efficiency indicates that I$^2$SB has potential for real-time or clinical deployment.", "date": "2025-08-15", "categories": []}
{"id": "2510.13562", "title": "An efficient approach with theoretical guarantees to simultaneously reconstruct activity and attenuation sinogram for TOF-PET", "abstract": "In positron emission tomography (PET), it is indispensable to perform attenuation correction in order to obtain the quantitatively accurate activity map (tracer distribution) in the body. Generally, this is carried out based on the estimated attenuation map obtained from computed tomography or magnetic resonance imaging. However, except for errors in the attenuation correction factors obtained, the additional scan not only brings in new radiation doses and/or increases the scanning time but also leads to severe misalignment induced by various motions during and between the two sequential scans. To address these issues, based on maximum likelihood estimation, we propose a new mathematical model for simultaneously reconstructing the activity and attenuation sinogram from the time-of-flight (TOF)-PET emission data only. Particularly, we make full use of the exclusively exponential form for the attenuation correction factors, and consider the constraint of a total amount of the activity in some mask region in the proposed model. Furthermore, we prove its well-posedness, including the existence, uniqueness and stability of the solution. We propose an alternating update algorithm to solve the model, and also analyze its convergence. Finally, numerical experiments with various TOF-PET emission data demonstrate that the proposed method is of numerical convergence and robust to noise, and outperforms some state-of-the-art methods in terms of accuracy and efficiency, and has the capability of autonomous attenuation correction.", "date": "2025-10-15", "categories": []}
{"id": "2606.18826", "title": "EDoF-NeRF: extended depth-of-field neural radiance fields using a coded aperture camera", "abstract": "We propose a method for extending the depth-of-field (DoF) to construct high-fidelity neural radiance fields (NeRF) -- an emerging technique for rendering photorealistic novel views from a dataset of images captured at different viewpoints, based on implicit neural representations. The trade-off between DoF and light quantity is inherent not only in conventional cameras but also in NeRF, since the datasets used by NeRF are captured by these cameras. To address this issue, we introduce a coded aperture placed at the camera pupil, preserving spatial frequency components under defocused conditions. We develop a camera model incorporating coded apertures into NeRF, allowing direct input of coded images and enabling the generation of novel views with an extended DoF. We validate the proposed method, termed extended DoF-NeRF (EDoF-NeRF), through simulations and experiments, demonstrating its superior performance compared to conventional aperture cameras.", "date": "2026-06-17", "categories": []}
{"id": "2404.09790", "title": "NTIRE 2024 Challenge on Image Super-Resolution (x4): Methods and Results", "abstract": "This paper reviews the NTIRE 2024 challenge on image super-resolution ($\\times$4), highlighting the solutions proposed and the outcomes obtained. The challenge involves generating corresponding high-resolution (HR) images, magnified by a factor of four, from low-resolution (LR) inputs using prior information. The LR images originate from bicubic downsampling degradation. The aim of the challenge is to obtain designs/solutions with the most advanced SR performance, with no constraints on computational resources (e.g., model size and FLOPs) or training data. The track of this challenge assesses performance with the PSNR metric on the DIV2K testing dataset. The competition attracted 199 registrants, with 20 teams submitting valid entries. This collective endeavour not only pushes the boundaries of performance in single-image SR but also offers a comprehensive overview of current trends in this field.", "date": "2024-04-15", "categories": []}
{"id": "2406.07435", "title": "Beware of Aliases -- Signal Preservation is Crucial for Robust Image Restoration", "abstract": "Image restoration networks are usually comprised of an encoder and a decoder, responsible for aggregating image content from noisy, distorted data and to restore clean, undistorted images, respectively. Data aggregation as well as high-resolution image generation both usually come at the risk of involving aliases, i.e.~standard architectures put their ability to reconstruct the model input in jeopardy to reach high PSNR values on validation data. The price to be paid is low model robustness. In this work, we show that simply providing alias-free paths in state-of-the-art reconstruction transformers supports improved model robustness at low costs on the restoration performance. We do so by proposing BOA-Restormer, a transformer-based image restoration model that executes downsampling and upsampling operations partly in the frequency domain to ensure alias-free paths along the entire model while potentially preserving all relevant high-frequency information.", "date": "2024-06-11", "categories": []}
{"id": "2602.22159", "title": "CASR: A Robust Cyclic Framework for Arbitrary Large-Scale Super-Resolution with Distribution Alignment and Self-Similarity Awareness", "abstract": "Arbitrary-Scale SR (ASISR) remains fundamentally limited by cross-scale distribution shift: once the inference scale leaves the training range, noise, blur, and artifacts accumulate sharply. We revisit this challenge from a cross-scale distribution transition perspective and propose CASR, a simple yet highly efficient cyclic SR framework that reformulates ultra-magnification as a sequence of in-distribution scale transitions. This design ensures stable inference at arbitrary scales while requiring only a single model. CASR tackles two major bottlenecks: distribution drift across iterations and patch-wise diffusion inconsistencies. The proposed SSAM module aligns structural distributions via superpixel aggregation, preventing error accumulation, while SARM module restores high-frequency textures by enforcing correlation-guided consistency and preserving self-similarity structure through correlation alignment. Despite using only a single model, our approach significantly reduces distribution drift, preserves long-range texture consistency, and achieves superior generalization even at extreme magnification.", "date": "2026-02-25", "categories": []}
{"id": "2606.17093", "title": "Diagnosing and Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry", "abstract": "Learning-based single-shot fringe projection profilometry (FPP) has been studied mostly at close range. The long-range regime (standoff beyond 1 m) remains largely unaddressed: inverse-square intensity falloff lowers fringe signal-to-noise ratio and degrades physical ground truth, the single-shot problem is ill-posed because fringe-order information is absent from one image, and these architectures have not been studied mechanistically. We present a diagnose-repair-verify study using mechanistic interpretability (MI) and conformal uncertainty quantification (UQ) as convergent diagnostics: they agree on one physical failure locus, driving and verifying an architectural repair. On a photorealistic synthetic benchmark (15,600 fringe images, 50 objects at 1.5-2.1 m), a best UNet baseline reaches 14.54 mm object mean absolute error (MAE). Three probes (linear probing, Grad-CAM, flat-plane out-of-distribution test) converge: the baseline solves the task via object-boundary shape priors rather than fringe-phase decoding. We repair this with PhiCalNet, which outputs wrapped phase rather than depth and applies a fixed differentiable calibration layer mapping phase to depth, removing the shape-prior solution from the hypothesis space architecturally rather than by a loss penalty. A physics-informed loss that enforces the same physics as a soft penalty on a depth-regressing network yields no measurable gain, isolating the architecture as the operative factor. PhiCalNet reduces object MAE 3.3x to 4.46 mm; the residual is carried by 0.103% of pixels at the +/-pi wrap discontinuity. Pixel-wise conformal UQ confirms the diagnosis: rejecting the top 5% of object pixels by snapshot disagreement cuts PhiCalNet RMSE by 64% (20.6->7.4 mm) versus 3.5% for the baseline. MI and UQ converge on the same failure locus.", "date": "2026-06-13", "categories": []}
{"id": "2606.17343", "title": "Bayesian Magnetic Resonance Joint Image Reconstruction and Uncertainty Quantification using Sparsity Prior Models and Markov Chain Monte Carlo Sampling", "abstract": "We propose a novel framework for uncertainty quantification using compressed sensing magnetic resonance image reconstruction. The problem is formulated within a Bayesian framework as a linear inverse problem, with prior distributions assigned to the unknown model parameters. Specifically, the image to be reconstructed is assumed to be sparse in a given basis. We develop a general framework applicable to any basis and as examples, we test the sparsity of the image in its (1) spatial gradients using a total variation prior model, and in its (2) wavelet transform. A Markov chain Monte Carlo (MCMC) method, based on a split-and-augmented Gibbs sampler, is then employed to sample from the posterior distribution of the unknown parameters. The non-differentiable conditional distributions are efficiently sampled using a proximal MCMC method. The proposed algorithms are validated on both single-coil and multi-coil datasets using various k-space sub-sampling patterns and ratios. The results demonstrate the superior performance of each proposed approach in reconstructing images compared to its counterpart optimisation-based method. Moreover, our framework effectively quantifies uncertainty, showing a notable correlation between estimated uncertainty maps and error maps computed using ground truth and reconstructed images, compared with existing deep learning-based methods.", "date": "2026-06-15", "categories": []}
{"id": "2606.17557", "title": "Universal Image Restoration via Internalized Chain-of-Thought Reasoning", "abstract": "Image restoration seeks to recover high-quality images from degraded inputs but becomes highly ill-posed under complex, mixed degradations. While unified all-in-one models are common, their performance declines as degradation complexity increases. Recent works adopt Chain-of-Thought (CoT) reasoning for multi-round restoration using specialized modules. However, this approach faces two key limitations: (i) increased computational cost due to multi-step processing, and (ii) weak modeling of interactions between degradations during stepwise inference. We introduce CoTIR, a universal image restoration framework that internalizes CoT reasoning within a single model. Concretely, we view image restoration as a specialized subtask of image editing, which implies that a large-scale pre-trained editing model provides a more favorable optimization starting point. Building on this, we fine-tune the model for restoration and further encode structured CoT-style reasoning into the learning objective via a differentiable formulation inspired by Lagrangian optimization, enabling holistic restoration without chaining specialized restorers. To facilitate training and evaluation, we further present CoTIR-Bench, a large-scale benchmark comprising 5.2 million samples with CoT-style reasoning traces. Extensive experiments on CoTIR-Bench and broad real composite degradation scenes show that CoTIR achieves stronger perceptual quality and more competitive fidelity than both all-in-one models and multi-round restoration methods. The source code is available at https://github.com/gy65896/CoTIR.", "date": "2026-06-16", "categories": []}
{"id": "2511.12024", "title": "Null-Space Diffusion Distillation Unlocks Speed, Fidelity and Realism in Lensless Imaging", "abstract": "Lensless imaging reconstructs scenes from highly multiplexed measurements, resulting in a severely ill-posed inverse problem. In this work, we identify a fundamental trade-off between measurement consistency, perceptual quality, and inference speed across lensless reconstruction paradigms. Traditional methods favor consistency but produce perceptually degraded results, supervised approaches achieve high-quality reconstructions with fast inference but may violate physical constraints, and diffusion-prior methods achieve high perceptual quality and consistency--particularly when structured constraints such as range-null decomposition are used--but remain slow due to iterative sampling. Motivated by this observation, we propose Null-Space Diffusion Distillation (NSDD), a single-pass reconstruction model that distills structured diffusion-prior inference into an efficient feed-forward network. NSDD learns to produce high-quality reconstructions that preserve measurement consistency while avoiding costly iterative sampling. Experimental results demonstrate that NSDD achieves perceptual quality and consistency competitive with diffusion-prior methods, while providing significantly faster inference and offering a favorable balance across all three objectives. Furthermore, ablation experiments show that distilling the range--null decomposition improves reconstruction quality and robustness over unstructured full-reconstruction distillation, including on unseen real scenes. These results highlight the potential of structure-aware distillation for efficient lensless imaging. Code is available at github.com/JRCSAVSN/NullSpaceDiffusionDistillation.", "date": "2025-11-15", "categories": []}
{"id": "2606.01110", "title": "Accelerating physics-informed neural networks for full waveform inversion using a hybrid quantum-classical finite-basis architecture", "abstract": "Full waveform inversion (FWI) reconstructs heterogeneous material properties from receiver data but remains computationally demanding. Physics-informed neural networks (PINNs) and their domain-decomposed variants (FBPINNs) offer a mesh-free alternative but face convergence challenges when representing complex velocity fields. We present a hybrid quantum-classical FBPINN for acoustic FWI, bringing together quantum computing and classical machine learning, in which the decomposed wavefield network and the global velocity network are implemented as classical-to-quantum pipelines terminating in parameterized quantum circuits (PQCs). The PQCs are realized as differentiable JAX statevector simulators, enabling end-to-end automatic differentiation through the classical PINN, the quantum circuit, and the physics-informed loss. On a geophysical anomaly benchmark, the quantum hybrid reaches a lower L1 velocity error than the primary classical FBPINN baseline in approximately 8x fewer training iterations, despite using approximately 33% fewer trainable parameters, and it outperforms all 15 classical hyperparameter variants tested. A second benchmark (checkerboard) demonstrates the generality of the inversion pipeline, confirming that the quantum hybrid architecture can recover structured spatial variations beyond the localized anomaly benchmark. Our framework is broadly applicable to wave-based inverse problems beyond geophysics, including medical ultrasound tomography and non-destructive evaluation.", "date": "2026-05-31", "categories": []}
{"id": "2606.15967", "title": "CRIS: Cross-Plane Self-Supervised Isotropic Restoration for Anisotropic Volumetric Imaging Across Modalities", "abstract": "Anisotropic volumetric acquisitions are common in clinical MRI and volume electron microscopy (vEM), where sparse through-plane sampling creates thick slices or sections that degrade orthogonal reformats and downstream analysis. We present CRIS, a cross-plane self-supervised framework for isotropic restoration without paired isotropic ground truth. CRIS casts 3D restoration as 2D stripe completion on orthogonal reformats of an isotropic grid: high-resolution in-plane slices are synthetically degraded and periodically masked for training, while at inference blank slices define the isotropic grid, two orthogonal reformats are restored, and predictions are fused by multi-view averaging. We evaluate CRIS on two MRI cohorts and two microscopy benchmarks up to 8x anisotropy. On brain MRI, CRIS achieves 32.921 +/- 0.436 dB PSNR and 0.9631 +/- 0.0027 SSIM, outperforming interpolation, SMORE4, SIMPLE, SA-INR, and ATME, and gives the best segmentation consistency (Dice 0.940 +/- 0.004, ASSD 0.245 +/- 0.014 mm, HD99 1.275 +/- 0.061 mm). On reference-free abdominal MRI, CRIS reduces FID/KID to 48.714/0.023. On vEM, CRIS outperforms interpolation, NIIV, and vEMINR, reaching 29.133 dB/0.834 3D PSNR/SSIM at 4x, 27.123 dB/0.734 on EPFL at 8x, and 21.915 dB/0.699 on noisy hemibrain data. In a robustness experiment, one variable-gap CRIS model evaluated across gap factors 3--7 and coronal, axial, and sagittal degradations maintained higher PSNR/SSIM than interpolation (36.36--31.14 dB and 0.977--0.932 vs. 33.07--27.85 dB and 0.951--0.853). These results support CRIS as a modality-flexible route to isotropic restoration without paired isotropic targets or configuration-specific retraining. Code is available at https://github.com/adi-hatav/CRIS.", "date": "2026-06-14", "categories": []}
{"id": "2606.16031", "title": "The Third Challenge on Image Denoising at NTIRE 2026: Methods and Results", "abstract": "This paper reports on the NTIRE 2026 Challenge on Image Denoising, specifically focusing on the high-noise regime ($\\sigma = 50$). The competition investigates advanced neural architectures designed to restore high-fidelity details from images corrupted by additive white Gaussian noise (AWGN). Unlike constrained benchmarks, this track emphasizes peak quantitative performance, measured by Peak Signal-to-Noise Ratio (PSNR), without limitations on parameter count or computational overhead. By synthesizing contributions from 20 finalist teams out of 116 registrants, this report benchmarks the latest technical innovations and provides a comprehensive snapshot of the current state-of-the-art in unconstrained image restoration.", "date": "2026-06-14", "categories": []}
{"id": "2606.16212", "title": "LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction", "abstract": "Sparse-view CT reduces radiation dose and scanning time by acquiring fewer projection views, but angular undersampling makes reconstruction severely ill-posed, causing streak artifacts, structural blurring, and loss of fine details. Existing supervised methods are often tied to specific sampling settings, whereas generative methods may introduce anatomically inconsistent hallucination-like structures under severe undersampling. We propose Lucid, a sparsity-adaptive, consistency-guided reconstruction framework based on a Flow Matching generative prior for sparse-view CT. Lucid is trained only on high-quality CT images to learn a continuous transport between a Gaussian distribution and the high-quality CT image distribution, independent of view sampling. During inference, the sampling sparsity level is explicitly incorporated to adapt the generative trajectory of a single pretrained model. Specifically, Lucid constructs a degradation-matched initial state by sparsity-weighted fusion of the sparse-view FBP image and Gaussian noise, performs sparsity-modulated Flow Matching updates, and applies projection-domain data-consistency correction after each prior update. Experiments under multiple sparse-view settings show that Lucid achieves stable reconstruction performance across different sampling densities, improves image quality and structural fidelity, and reduces the risk of hallucination-like structures in generative sparse-view CT reconstruction.", "date": "2026-06-15", "categories": []}
{"id": "2606.16392", "title": "Towards UAV Image Dehazing: A UAV Atmospheric Scattering Model, Benchmark, and Geometry-Aware Deep Unfolding Network", "abstract": "In UAV applications, haze significantly obscures distant details and weaken structural information, hindering the recovery of details. Current UAV scenarios still face two key challenges: (i) paired hazy/clean images from the real world are unobtainable, while the classical atmospheric scattering model is inadequate for modeling the spatially non-uniform haze in UAV imagery; (ii) existing dehazing methods struggle to remove the heavy haze accumulated in the upper regions of UAV images. To address these issues, we first propose a UAV Atmospheric Scattering Model (UASM), which explicitly incorporates flight altitude, viewing pitch, and extinction to characterize the non-uniform haze distribution in UAV imaging. Based on UASM, we develop a physics-driven dehazing framework, termed Geometry-aware Proximal Deep Unfolding Network (GP-DUN). Specifically, GP-DUN consists of three key modules: a Latent Geometry Estimator (LGE) that infers transmittance consistent with UAV imaging geometry, a Geometry-aware Gradient Descent Module (GeoGDM) that embeds UASM into the data-fidelity term and performs physics-consistent closed-form updates, and an Pooling-Expert Proximal Mapping Module (PE-PMM) that learns an implicit prior to restore textures and structures beyond the capability of explicit physical modeling. In addition, we further construct UASM-HazeSet, which provides controllable paired synthetic data together with 2,285 real UAV haze images for testing. Extensive experiments show that GP-DUN consistently outperforms existing methods on both UASM-HazeSet and real UAV haze benchmarks.", "date": "2026-06-15", "categories": []}
{"id": "2606.16396", "title": "SP$^3$: Spherical Priors for Plug-and-Play Restoration", "abstract": "In this paper, we introduce SP$^3$, a novel Plug-and-Play algorithm that accelerates maximum a posteriori image restoration by replacing denoisers with Spherical Encoders (SE) as generative priors. SP$^3$ approximates the intractable proximal prior step by utilizing the SE tightly structured latent space as a robust projection onto the natural image manifold. Alternating this projection with a closed-form data-consistency step, via Half-Quadratic Splitting, achieves stable convergence without requiring gradient computation during inference. This unique formulation unlocks \"anytime\" restoration capabilities, producing sharp, plausible images from the first iteration. Evaluations across a variety of image restoration tasks demonstrate that SP$^3$ achieves perceptual quality comparable to state-of-the-art zero-shot diffusion and flow methods while being $3$-$630\\times$ faster.", "date": "2026-06-15", "categories": []}
{"id": "2606.05102", "title": "ZipSplat: Fewer Gaussians, Better Splats", "abstract": "Feed-forward 3D Gaussian Splatting methods reconstruct a scene from posed or pose-free images in a single forward pass, yet current approaches predict one Gaussian per input pixel, tying the representation budget to camera resolution rather than scene complexity. A flat wall and a richly textured object thus produce equally many Gaussians despite very different geometric needs. We propose ZipSplat, a token-based feed-forward model that decouples Gaussian placement from the pixel grid. A multi-view backbone extracts dense visual tokens, and k-means clustering compresses them into a compact set of scene tokens. Cross- and self-attention refine these tokens, and a lightweight MLP decodes each into a group of Gaussians with unconstrained 3D positions. Because clustering is applied at inference, a single trained model spans the quality-efficiency curve without retraining. ZipSplat operates without ground-truth poses or intrinsics, yet sets a new state of the art on DL3DV and RealEstate10K with ${\\sim}6{\\times}$ fewer Gaussians than pixel-aligned methods, surpassing the best pose-free baseline by 2.1dB and 1.2dB PSNR, respectively. It further generalizes zero-shot to Mip-NeRF360 and ScanNet++, outperforming all comparable baselines. Our project page is at https://veichta.com/zipsplat.", "date": "2026-06-03", "categories": []}
{"id": "2606.14139", "title": "Decoupled Latent Optimization of Diffusion Models for Full Waveform Inversion", "abstract": "Full waveform inversion (FWI) recovers subsurface velocity from seismic recordings by solving a severely ill-posed, nonconvex PDE-constrained optimization. Classical regularizers stabilize the inversion but fail to reproduce realistic geological structures; recent diffusion-prior methods improve realism at the cost of a fragile trade-off between data fidelity and prior consistency. We propose Decoupled Latent Optimization (DLO), which relaxes the standard latent-optimization formulation into a quadratic-penalty objective over an auxiliary physical variable and a latent variable. The data-fidelity gradient acts in physical space, the diffusion sampler contributes only through a decoded prior sample, and the standard smoothed-velocity initialization of classical FWI is preserved. On the OpenFWI benchmark, DLO outperforms classical regularizers and existing diffusion-based methods under clean, noisy, and missing-trace acquisitions. The prior, trained on 70*70 OpenFWI models, transfers directly to the Marmousi and Overthrust benchmarks, where DLO recovers intricate fault structures and remains robust to initialization smoothing and measurement noise.", "date": "2026-06-12", "categories": []}
{"id": "2606.14344", "title": "More with LESS -- Local Scene Representations for Tactile Imaging", "abstract": "Tactile imaging seeks to reconstruct the internal structure of soft objects through touch sensing, with applications in medical diagnosis and robotic manipulation. Recent self-supervised learning approaches have shown promising results, but rely on global, unstructured representations and robot-controlled sensing, limiting generalization and practical use. We propose Local Encoder for Spatial Sensing (LESS), an object-centric tactile representation that exploits the local nature of touch. The tactile scene is modeled as a grid of recurrent encoders with local receptive fields, whose states are fused to reconstruct 2D or 3D images of internal structure. This compositional design enables strong generalization: models trained on single-inclusion phantoms accurately image objects with multiple inclusions and varying sizes. The local structure further supports spatial uncertainty estimation. In addition, we enable hand-held tactile imaging via external pose tracking and human-like palpation data, and extend tactile imaging to full 3D reconstruction.", "date": "2026-06-12", "categories": []}
{"id": "2606.14638", "title": "Improving Lunar Topography with Deep Learning Schr\\\"odinger Bridges", "abstract": "Increasing the resolution of planetary topography models can enable a better understanding of surface processes and geomorphology; however, existing analytical super-resolution methods are expensive and difficult to apply at large scales. Generative models provide the tools to learn complex relationships within data and can be applied at scale due to hardware accelerators and parallelization. We present a diffusion-based Schr\\\"odinger Bridge (SB) generative modeling approach for lunar topography super-resolution, connecting the distribution of low-resolution topography to that of high-resolution topography, incorporating physically-constraining optical imagery. Our approach is inspired by existing Shape-from-Shading methods, which improve a priori low-resolution topography by using optical images at the target resolution. We train SBs on a novel dataset of rendered lunar topography, emulating optical imagery from the Lunar Reconnaissance Orbiter Narrow Angle Camera. The result is a flexible approach for topography super-resolution which can provide pixel-level uncertainties in the reconstruction.", "date": "2026-06-12", "categories": []}
{"id": "2606.12368", "title": "DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images", "abstract": "While monocular depth estimation has achieved significant progress, achieving generalized metric depth estimation for both narrow field-of-view (FoV) perspectives and $360^\\circ$ panoramas remains an unsolved challenge. Existing methods are often tailored to specific camera types and struggle to produce accurate metric depth that generalizes across diverse settings. This limitation stems from two key challenges: the inherent geometric discrepancy between perspective and panoramic cameras, and the scarcity of panoramic training data with metric annotations. In this work, we introduce DepthMaster, a unified metric depth estimation framework. Rather than employing specialized networks to learn spherical distortions, we reformulate the problem by decomposing panoramic images into overlapping perspective patches. Crucially, distinct from prior projection-based methods that rely on ad-hoc architectural modifications to handle boundaries, we introduce a novel Correspondence Consistency Loss (CCL) and inject virtual projection cameras as geometric priors, allowing us to seamlessly stitch the patches while avoiding specialized operators and keeping the backbone largely compatible with standard Transformer designs. This strategy also resolves the geometric differences by unifying all inputs into a canonical perspective representation, and effectively circumvents data scarcity by directly unlocking powerful metric priors from vast perspective datasets. Trained on a mixed dataset that contains only one panorama dataset, DepthMaster achieves state-of-the-art zero-shot performance on 13 diverse datasets, outperforming not only universal methods but also leading specialist models in both perspective and panoramic domains.", "date": "2026-06-10", "categories": []}
{"id": "2606.13136", "title": "An Extensible and Lightweight Unified Architecture for Demosaicing Pixel-bin Image Sensors", "abstract": "Pixel-bin image sensors are becoming the default choice for smartphone cameras due to their resolution vs light-gathering trade-off. However, their larger inter-color separation compared to the Bayer color filter array (CFA) makes them challenging to demosaic. Furthermore, existing deep learning-based demosaicing methods are CFA-specific, requiring multiple individual models that take up precious onboard resources and demand larger development and maintenance efforts. In this work, we propose a modular unified architecture for demosaicing various pixel-bin sensors that provides higher image quality while being extensible and lightweight. Additionally, to enable plug-and-play operation, we introduce a learning-free CFA-identification module to detect the CFA type of raw data accurately.", "date": "2026-06-11", "categories": []}
{"id": "2606.13644", "title": "Surflo: Consistent 3D Surface Flow Model with Global State", "abstract": "Geometry is invariant to viewpoint, which makes any collection of images a redundant encoding of a single 3D state. Existing feed-forward reconstruction models fail to exploit this: per-view methods emit overlapping, unaligned pointmaps that grow linearly with input count, while global-latent methods commit to a fixed, low-resolution output. We introduce Surflo, which compresses a variable number of unposed RGB views into K latent tokens-one global state-and decodes oriented 3D surface points by independently transporting them from noise onto the surface via flow matching. This frees the output from any fixed grid or token budget: the same latent yields from a few thousand to a million points in a single forward pass. To suppress the local inconsistencies inherent to independent per-point decoding, an inference-time guidance term correlates nearby points by injecting a photometric gradient during ODE integration. Surflo matches or surpasses feed-forward baselines on surface metrics, runs an order of magnitude faster than optimization-based methods that require hundreds of views, and is the only feed-forward approach to combine a global latent with arbitrary-resolution decoding.", "date": "2026-06-11", "categories": []}
{"id": "2512.20464", "title": "Snapshot 3D image projection using a diffractive decoder", "abstract": "3D image display is essential for next-generation volumetric imaging; however, dense depth multiplexing for 3D image projection remains challenging because diffraction-induced cross-talk rapidly increases as the axial image planes get closer. Here, we introduce a 3D display system comprising a digital encoder and a diffractive optical decoder, which simultaneously projects different images onto multiple target axial planes with high axial resolution. By leveraging multi-layer diffractive wavefront decoding and deep learning-based end-to-end optimization, the system achieves high-fidelity depth-resolved 3D image projection in a snapshot, enabling axial plane separations on the order of a wavelength. The digital encoder leverages a Fourier encoder network to capture multi-scale spatial and frequency-domain features from input images, integrates axial position encoding, and generates a unified phase representation that simultaneously encodes all images to be axially projected in a single snapshot through a jointly-optimized diffractive decoder. We characterized the impact of diffractive decoder depth, output diffraction efficiency, spatial light modulator resolution, and axial encoding density, revealing trade-offs that govern axial separation and 3D image projection quality. We further demonstrated the capability to display volumetric images containing 28 axial slices, as well as the ability to dynamically reconfigure the axial locations of the image planes, performed on demand. Finally, we experimentally validated the presented approach, demonstrating close agreement between the measured results and the target images. These results establish the diffractive 3D display system as a compact and scalable framework for depth-resolved snapshot 3D image projection, with potential applications in holographic displays, AR/VR interfaces, and volumetric optical computing.", "date": "2025-12-23", "categories": []}
{"id": "2606.11601", "title": "Spatially Coupled Phase-to-Depth Calibration for Fringe Projection Profilometry", "abstract": "In fringe projection profilometry (FPP), depth is commonly recovered by fitting a phase-to-depth relation independently at each camera pixel. Although such pixel-wise calibration achieves high local accuracy, neighboring pixels can acquire markedly different calibration functions even when they observe the same smooth surface, producing spatially inconsistent geometry and structured surface artifacts. We propose a spatially coupled phase-depth transformation in which all pixels share a single low-dimensional mapping-global phase scalars combined with affine spatial terms on the undistorted reference-camera grid-rather than independent per-pixel fits, optionally augmented by a bounded, spatially smooth correction field. We further introduce a native-grid pairing scheme that constructs phase-depth calibration pairs directly on the reference-camera grid: when depth supervision comes from a rectified active-stereo pipeline, planes are fitted in stereo 3D and sampled back onto the camera grid along native rays, so the phase maps are never rectified. On a dental target with high-resolution scanner ground truth, the proposed model attains point-to-surface RMSE comparable to an active-stereo reference (about 12{\\mu}m aggregate) while substantially improving spatial coherence over pixel-wise polynomial and rational calibration, and reduces the runtime mapping to a few element-wise operations per pixel with negligible parameter storage.", "date": "2026-06-10", "categories": []}
{"id": "2501.01481", "title": "Unleashing Correlation and Continuity for Hyperspectral Reconstruction from RGB Images", "abstract": "Reconstructing Hyperspectral Images (HSI) from RGB images can yield high spatial resolution HSI at a lower cost, demonstrating significant application potential. This paper reveals that local correlation and global continuity of the spectral characteristics are crucial for HSI reconstruction tasks. Therefore, we fully explore these inter-spectral relationships and propose a Correlation and Continuity Network (CCNet) for HSI reconstruction from RGB images. For the correlation of local spectrum, we introduce the Group-wise Spectral Correlation Modeling (GrSCM) module, which efficiently establishes spectral band similarity within a localized range. For the continuity of global spectrum, we design the Neighborhood-wise Spectral Continuity Modeling (NeSCM) module, which employs memory units to recursively model the progressive variation characteristics at the global level. In order to explore the inherent complementarity of these two modules, we design the Patch-wise Adaptive Fusion (PAF) module to efficiently integrate global continuity features into the spectral features in a patch-wise adaptive manner. These innovations enhance the quality of reconstructed HSI. We perform comprehensive comparison and ablation experiments on the mainstream datasets NTIRE2022 and NTIRE2020 for the spectral reconstruction task. Compared to the current advanced spectral reconstruction algorithms, our designed algorithm achieves State-Of-The-Art (SOTA) performance.", "date": "2025-01-02", "categories": []}
{"id": "2601.04776", "title": "Segmentation-Driven Monocular Shape from Polarization based on Physical Model", "abstract": "Monocular shape-from-polarization (SfP) leverages the intrinsic relationship between light polarization properties and surface geometry to recover surface normals from single-view polarized images, providing a compact and robust approach for three-dimensional (3D) reconstruction. Despite its potential, existing monocular SfP methods suffer from azimuth angle ambiguity, an inherent limitation of polarization analysis, that severely compromises reconstruction accuracy and stability. This paper introduces a novel segmentation-driven monocular SfP (SMSfP) framework that reformulates global shape recovery into a set of local reconstructions over adaptively segmented convex sub-regions. Specifically, a polarization-aided adaptive region growing (PARG) segmentation strategy is proposed to decompose the global convexity assumption into locally convex regions, effectively suppressing azimuth ambiguities and preserving surface continuity. Furthermore, a multi-scale fusion convexity prior (MFCP) constraint is developed to ensure local surface consistency and enhance the recovery of fine textural and structural details. Extensive experiments on both synthetic and real-world datasets validate the proposed approach, showing significant improvements in disambiguation accuracy and geometric fidelity compared with existing physics-based monocular SfP techniques.", "date": "2026-01-08", "categories": []}
{"id": "2606.10280", "title": "Overlapped Wavelet Diffusion for Low-Light Image Enhancement", "abstract": "In this study, we propose an overlapped wavelet diffusion framework for Low-Light Image Enhancement (LLIE), which incorporates two complementary components to achieve blocking artifact-free and detail-preserving enhancement. Although recent diffusion-based LLIE methods have demonstrated remarkable performance compared with traditional approaches, DiffLL still suffers from blocking artifacts caused by the Haar Wavelet Transform (WT) and blurred edges or over-smoothed textures due to the limitations of its High-Frequency Restoration Module (HFRM). To overcome these issues, we introduce an Overlapped WT (OWT) that incorporates correlations across neighboring regions, thereby structurally preventing blocking artifacts. Furthermore, we integrate a low-frequency-guided High-Frequency Enhance Block (HFEBlock) to strengthen detail recovery, yielding sharper edges and more reliable textures. Extensive experiments on the LOLv1 and LOLv2-real datasets demonstrate that our framework, termed OWDiff, consistently outperforms existing LLIE methods both qualitatively and quantitatively, achieving superior visual quality while maintaining computational efficiency. OWDiff effectively addresses the structural limitations of the Haar WT and the HFRM, achieving an average PSNR gain of 0.58 dB, along with a 1.64% relative improvement in SSIM and a 5.9% relative reduction in LPIPS, compared to DiffLL across both the LOLv1 and LOLv2-real datasets.", "date": "2026-06-09", "categories": []}
{"id": "2606.10373", "title": "PF-Trans: Physics-Embedded Frequency-Aware Transformer for Spectral Reconstruction", "abstract": "Snapshot Broadband Filter Array (BFA) imaging provides high light throughput for spectral reconstruction but introduces severe spectral aliasing due to complex modulation. Current deep learning approaches, limited to spatial denoising, often fail to address the global frequency-specific degradations caused by the mask structure. To address this, we propose a Physics-embedded Frequency-aware Transformer (PF-Trans) for high-fidelity remote sensing spectral reconstruction. Our method explicitly integrates the physical sensing model through mask injection and a gray-scale consistency loss to ensure physical fidelity. Furthermore, we introduce a Dual-domain Block with a parallel Fast Fourier Transform (FFT) branch, enabling the network to perceive and suppress aliasing artifacts in the frequency domain. Extensive experiments on multiple datasets demonstrate that PF-Trans achieves state-of-the-art performance, achieving a Peak Signal-to-Noise Ratio (PSNR) of up to 48.50 dB on the GF-5 Shanghai dataset, significantly outperforming comparison methods.", "date": "2026-06-09", "categories": []}
{"id": "2606.07675", "title": "The Need for Neural ISP in the Small-Pixel Era: How Shrinking Pixels Push Optics to the Limit and Neural Restoration Pushes Back", "abstract": "Smartphone telephoto cameras are approaching a \"telephoto physics wall\": as pixel pitches shrink toward sub-0.5 micron, the optics remain limited by geometric aberrations, leading to diminishing returns on resolution. Traditional Image Signal Processors (ISPs) cannot eliminate these aberrations, because they operate through local, stage-wise processing with no explicit model of the underlying point spread function (PSF). We demonstrate how a learning-based Neural ISP for image restoration, trained on the underlying degradations, inverts what stage-wise pipelines cannot, turning small-pixel designs into a net advantage.\n  We investigate this through a controlled simulation of a representative telephoto module, evaluating five configurations (0.35--0.75 micron pixel pitch). The aperture is scaled proportionally to keep per-pixel SNR and diffraction spot size fixed, thereby isolating geometric aberration and spatial sampling. While the traditional ISP improves only modestly with smaller pixels, the Neural ISP scales substantially: at 0.35 micron} it reaches 745 cycles/mm MTF50 (vertical), a 2.5--3x resolution improvement over the traditional ISP, and LPIPS improves significantly from 0.244 to 0.151 while traditional results stay comparatively flat. In a low-SNR extension (15 dB per-frame bursts at 0.35 micron), a multi-frame Neural ISP recovers performance close to the bright-light single-frame baseline, whereas a multi-frame traditional ISP shows no meaningful improvement -- indicating that traditional pipelines at small pixels are bottlenecked by uncorrected PSF blur rather than by noise. These results point to a design philosophy in which Neural ISPs enable high-resolution telephoto modules by correcting residual optical aberrations rather than requiring increasingly complex optics.", "date": "2026-06-04", "categories": []}
{"id": "2606.08324", "title": "Set-Based Transformer for Atmospheric Compensation in Standoff LWIR Hyperspectral Imaging", "abstract": "Passive long-wave infrared (LWIR) hyperspectral imaging under a standoff geometry depends on atmospheric absorption and emission, as well as reflected radiance, thus making atmospheric compensation essential to get knowledge of a target of interest. Despite its importance, this compensation has been largely overlooked due to its practical and modeling difficulty. In this paper, we present a lightweight set-based deep learning framework that takes multiple radiance measurements, collected at different standoff ranges, as input and jointly estimates transmittance, atmospheric path radiance, and a shared downwelling spectrum. We analyze the learned representation with a sparse autoencoder and observe that several latent features do activate on geographically coherent subsets of the test data despite the absence of location supervision. Experiments on a MODTRAN generated standoff LWIR dataset demonstrate low spectral distortion across all estimated products. The dataset and code is publicly available at: https://factral.co/SAE-LWIR/", "date": "2026-06-06", "categories": []}
{"id": "2606.09110", "title": "HDRAgent: An Agentic Framework for Multi-Exposure HDR Imaging", "abstract": "Most existing multi-exposure HDR methods follow a fixed feed-forward reconstruction paradigm, making them prone to ghosting artifacts in complex dynamic scenes. To address this issue, we propose HDRAgent, the first agent-driven framework for HDR imaging, which adaptively selects reconstruction strategies according to the current scene conditions. Specifically, to provide scene-specific prior knowledge, we introduce a fine-grained contextual knowledge matching (FCM) module. This module leverages multimodal large language model (MLLM)-derived scene perception to retrieve relevant historical cases and tool knowledge, organizing them into structured evidence for MLLM-based adaptive tool scheduling. In addition, we propose a perception--distortion feedback mechanism that transforms post-execution quality assessment and artifact diagnosis into structured feedback, which is accumulated in historical memory to help subsequent contextual knowledge refinement and strategy selection. Furthermore, considering that extreme motion can invalidate alignment methods, we design an agent-guided generative alignment strategy that uses MLLM-based dynamic-region parsing to reconstruct unreliable contents in non-reference frames under reference-frame guidance. Experiments demonstrate that HDRAgent effectively reduces ghosting and local artifacts while achieving competitive or superior objective performance and visual quality.", "date": "2026-06-08", "categories": []}
{"id": "2606.09792", "title": "End-to-End Optimization of Incoherent Imaging for Classification Under Detector-Limited Readout", "abstract": "End-to-end co-optimization of optical front-ends (e.g. metasurfaces) and neural network back-ends has been widely applied to imaging tasks, yet a formalism characterizing when and why such systems outperform conventional lens-based imaging is largely lacking. This paper focuses on object classification, a central imaging task, and asks when end-to-end optimization of a phase mask for incoherent imaging improves performance over a conventional focusing lens. We find that these gains arise primarily under constrained detector readout and are limited under full detector readout. In the latter setting, we prove that no incoherent phase mask exceeds the ideal-channel mutual information between detector measurements and class labels; a conventional focusing lens approaches this ceiling, and joint optimization yields no empirical gain. When detector readout is constrained -- by coarse spatial sampling or a limited number of measurements -- optimized optics can substantially improve classification by increasing class separability in the detector measurements. These gains are largest under low detector noise and shrink as noise grows, because the optics shape the signal before it reaches the detector but cannot remove noise added afterward. The advantage also depends on the spectral structure of the task: co-design helps most when class-discriminative content is concentrated at lower spatial frequencies than within-class variation. We develop a theoretical framework formalizing these distinctions and test its predictions on synthetic data and standard benchmarks (MNIST, FashionMNIST, SVHN).", "date": "2026-06-08", "categories": []}
{"id": "2409.13477", "title": "A Plug-and-Play Method for Guided Multi-contrast MRI Reconstruction based on Content/Style Modeling", "abstract": "Since the various MR contrasts of a given anatomy contain redundant information, one contrast can be used to guide the reconstruction of another undersampled contrast acquired subsequently in the same session. To solve this reconstruction problem leveraging multi-contrast side information, several end-to-end learning-based methods have been proposed. However, a key challenge is the requirement for large paired training datasets comprising raw k-space data and aligned reference images. We propose a modular plug-and-play method, which requires no k-space training data and relies solely on partially paired image-domain datasets. A content/style model of two-contrast MR image data is first learned and subsequently applied as a plug-and-play operator in iterative reconstruction. The disentanglement of content and style allows explicit representation of contrast-independent and contrast-specific factors. Consequently, incorporating prior information into the reconstruction reduces to a simple replacement operation on the aliased content of the estimated image using high-quality content derived from the reference scan. Combining this operation with an MR data consistency step, followed by a corrective procedure for the content estimate, yields an iterative scheme. We name this novel approach PnP-CoSMo. It offers, by design, cross-contrast generalizability and provides an explanatory framework based on the shared and non-shared generative factors underlying the two given contrasts. We explore various aspects, including interpretability and convergence, via simulations. Furthermore, its practicality is demonstrated on the public NYU fastMRI DICOM dataset, showing equivalent or superior quality and greater generalizability compared to end-to-end methods. On two in-house multi-coil datasets, PnP-CoSMo enabled up to 32.6% greater acceleration over non-guided reconstruction at given SSIM.", "date": "2024-09-20", "categories": []}
{"id": "2603.21510", "title": "Unregistered Spectral Image Fusion: Unmixing, Adversarial Learning, and Recoverability", "abstract": "This paper addresses the fusion of a pair of spatially unregistered hyperspectral image (HSI) and multispectral image (MSI) covering roughly overlapping regions. HSIs offer high spectral but low spatial resolution, while MSIs provide the opposite. The goal is to integrate their complementary information to enhance both HSI spatial resolution and MSI spectral resolution. While hyperspectral-multispectral fusion (HMF) has been widely studied, the unregistered setting remains challenging. Many existing methods focus solely on MSI super-resolution, leaving HSI unchanged. Supervised deep learning approaches were proposed for HSI super-resolution, but rely on accurate training data, which is often unavailable. Moreover, theoretical analyses largely address the co-registered case, leaving unregistered HMF poorly understood. In this work, an unsupervised framework is proposed to simultaneously super-resolve both MSI and HSI. The method integrates coupled spectral unmixing for MSI super-resolution with latent-space adversarial learning for HSI super-resolution. Theoretical guarantees on the recoverability of the super-resolution MSI and HSI are established under reasonable generative models -- providing, to our best knowledge, the first such insights for unregistered HMF. The approach is validated on semi-real and real HSI-MSI pairs across diverse conditions.", "date": "2026-03-23", "categories": []}
{"id": "2605.25757", "title": "Broadband Hyperspectral 3D Imaging using Dispersed Structured Light", "abstract": "Hyperspectral 3D imaging enables the capture of dense spectral information and scene geometry but has traditionally been confined to narrow spectral windows, typically the visible range. In this work, we introduce a broadband hyperspectral 3D imaging (BH3D) method to extend this capability across the full visible-near-infrared and short-wavelength infrared (SWIR) spectrum (450-1500 nm). This broad coverage is critical as it captures complementary physical cues: visible wavelengths reveal surface appearance, while SWIR bands provide insight into subsurface properties and material composition. However, realizing BH3D is challenging due to fundamental sensor constraints between visible-spectrum silicon and SWIR-spectrum InGaAs sensors, which necessitate complex multi-spectrograph designs. Here we propose a single-spectrograph BH3D system, using a stereo setup comprising visible and SWIR cameras, that reconstructs dense broadband hyperspectral reflectance together with accurate 3D geometry. Our key idea is to extend dispersed structured light to the broadband regime using a single spectrograph. We model the image formation of broadband dispersed structured light, and estimate hyperspectral reflectance and depth. We validate our approach on diverse real-world scenes, demonstrating accurate reconstruction with a mean spectral angle mapper of 0.13 rad, root mean square error of 0.03, and mean depth error of 4.5 mm. We further demonstrate identifying metameric materials, performing imaging through opaque layers, uncovering hidden features on banknotes, and revealing blood vessels.", "date": "2026-05-25", "categories": []}
{"id": "2606.06540", "title": "ErA: Error-Aware Deep Unrolling Network for Single Image Defocus Deblurring", "abstract": "We introduce ErA (Error-Aware Deep Unrolling Network), an end-to-end frame work for single-image defocus deblurring. ErA jointly learns a compact kerne basis and per-pixel weights, while an error-aware term in Augmented Lagrangian unrolling corrects kernel estimation errors via alternating updates and ResUNet denoisers. It achieves state-of-the-art PSNR/SSIM on DPDD, RealDOF, and RTF, and shows strong generalization on CUHK without ground truth.", "date": "2026-06-04", "categories": []}
{"id": "2606.06690", "title": "RPC-GS: Gaussian Splatting with native RPC Rendering for Satellite Imagery", "abstract": "We present RPC-GS, the first Gaussian Splatting framework for satellite imagery that operates natively with Rational Polynomial Camera (RPC) models. The RPC model is the de facto standard for representing the complex imaging geometry of modern pushbroom satellite sensors. To simplify rendering, prior satellite Gaussian Splatting methods replace the RPC model with perspective or affine camera approximations, leading to geometric errors during reconstruction. RPC-GS avoids these approximations by projecting Gaussian means and covariances directly through the RPC model during the splatting process. We embed the RPC model in a chain of carefully selected geo-coordinate transformations representing a mapping from splatting-suitable scene coordinates to image coordinates. To map the Gaussian covariance matrices, we derive a numerically robust Jacobian-based covariance projection for the (partially nonlinear) coordinate transformations. Since RPCs lack an explicit notion of camera depth, we integrate a metric ray-based depth formulation. We benchmark RPC, perspective, and affine camera models in a unified framework, with our native RPC renderer consistently achieving the lowest reconstruction error on leading satellite benchmark datasets, improving mean altitude error over perspective and affine approximations by 29.6% and 63.8% on DFC2019, and by 9.9% and 37.9% on IARPA2016. We release our code to support future research of Gaussian Splatting in the satellite imaging domain.", "date": "2026-06-04", "categories": []}
{"id": "2606.06901", "title": "LUCID: Learning Unified Control for Image Deflaring and Exposure Mastery in Nighttime Photography", "abstract": "Photography is the art of painting with light, yet nighttime scenes are shaped by competing degradations: intense flares obscure scene structure, while photon-limited regions collapse into noise. Conventional approaches address these factors in isolation, overlooking the fact that these degradations are fundamentally entangled. To bridge this gap, we introduce LUCID, a unified framework that reframes nighttime restoration as a continuous and controllable process rather than a fixed correction. We decompose nighttime restoration into two cooperative components: a flare disentanglement module that lifts the 'curtain' of optical artifacts to provide reliable structural guidance, and a diffusion-driven module that leverages generative priors to reconstruct clean and well-exposed imagery. Crucially, LUCID introduces explicit controllability through a novel four-mode training strategy, enabling users to steer the restoration process via classifier-free guidance (CFG) and allowing selective control over light sources and their associated flare and ghosting artifacts, while also supporting high dynamic range (HDR) reconstruction through continuous exposure control. Extensive experiments demonstrate that LUCID consistently outperforms state-of-the-art methods across diverse real-world nighttime scenarios.", "date": "2026-06-05", "categories": []}
{"id": "2606.07257", "title": "On the conditional equivalence of phase retrieval algorithms", "abstract": "Phase retrieval - recovering a complex-valued field from intensity measurements - is typically solved using variants of the Gerchberg-Saxton (GS) algorithm, understood as alternating projections between measurement planes. Meanwhile, modern computational imaging increasingly relies on gradient-based optimization and automatic differentiation. Here we show that these two approaches are mathematically identical: the GS magnitude replacement step is exactly a unit gradient descent step on an amplitude least-squares loss. This equivalence enables seamless integration of classical phase retrieval with differentiable physics pipelines. We further identify two complementary probabilistic interpretations of this equivalence: globally, the amplitude loss is the negative log-likelihood under Gaussian amplitude noise; locally, each projection step arises as a Bayesian update with the propagated field as prior. The local view provides qualitative guidance for relaxation in iterative phase retrieval.", "date": "2026-06-05", "categories": []}
{"id": "2506.08809", "title": "Training-Free Inference for High-Resolution Sinogram Completion", "abstract": "High-resolution sinogram completion is critical for computed tomography reconstruction, as missing projections can introduce severe artifacts. While diffusion models provide strong generative priors for this task, their inference cost grows prohibitively with resolution. We propose HRSino, a training-free and efficient diffusion inference approach for high-resolution sinogram completion. By explicitly accounting for spatial heterogeneity in signal characteristics, such as spectral sparsity and local complexity, HRSino allocates inference effort adaptively across spatial regions and resolutions, rather than applying uniform high-resolution diffusion steps. This enables global consistency to be captured at coarse scales while refining local details only where necessary. Experimental results show that HRSino reduces peak memory usage by up to 30.81% and inference time by up to 17.58% compared to the state-of-the-art framework, and maintains completion accuracy across datasets and resolutions.", "date": "2025-06-10", "categories": []}
{"id": "2606.06278", "title": "Geodesic Flow Matching on a Riemannian Degradation Manifold for Blind Image Restoration", "abstract": "Blind image restoration requires recovering clean images from observations corrupted by unknown and potentially mixed degradations. While recent deterministic flow-based methods model restoration as transport processes that map degraded images to clean ones, they typically rely on Euclidean interpolation, implicitly assuming linear degradation geometry. In this paper, we explicitly model degradations as points on a low-dimensional Riemannian manifold and formulate restoration as geodesic transport on the joint image-manifold space. Using a geodesic flow matching objective, we learn intrinsic transport dynamics that respect the curvature of degradation space. This framework generalizes linear flow matching, provides a principled treatment of mixed degradations as geodesic compositions, and yields a clean theoretical interpretation for generalization beyond observed degradations.", "date": "2026-06-04", "categories": []}
{"id": "2606.04419", "title": "L-TGVN: Leveraging Longitudinal Priors for Personalized Rapid MRI", "abstract": "MRI provides excellent soft-tissue contrast without ionizing radiation, but long acquisition times increase patient discomfort while also raising exam costs and limiting scanner throughput. A common approach to reduce scan time is to acquire fewer measurements, which yields an ill-posed linear inverse problem; recovering diagnostic-quality images therefore requires incorporating prior knowledge beyond the measured data. In follow-up exams, the most recent prior scan of a patient can provide a highly informative subject-specific context, but practical use is complicated by temporal changes (including pathology progression), misalignment between scans, and protocol drift across acquisitions. In this work, we introduce L-TGVN, a Longitudinal Trust-Guided Variational Network that leverages prior scans as side information to reconstruct the current scan from heavily undersampled measurements. Crucially, L-TGVN constrains the influence of prior scans to be consistent with the acquired measurements. Unlike many existing longitudinal reconstruction methods, it does not require explicit pre-registration between prior and current scans. It further accommodates differences in acquisition protocols across visits (e.g., changes in sequence parameters). We evaluate L-TGVN against matched-capacity baselines, including prior-guided methods and methods that do not use longitudinal priors, and observe consistent improvements in standard quantitative metrics together with better preservation of fine structures at challenging accelerations. Source code is available at github.com/sodicksonlab/L-TGVN.", "date": "2026-06-03", "categories": []}
{"id": "2606.04593", "title": "4D Reconstruction from Sparse Dynamic Cameras", "abstract": "Although dynamic 3D (i.e., 4D) reconstruction from a monocular dynamic camera has recently advanced, it remains fundamentally limited by depth ambiguity. In this paper, we focus on an alternative practical way, i.e., sparse dynamic camera setup, where a handful of independently moving cameras capture the same subjects. While keeping capture costs low, this setup introduces multi-view constraints and remains practical for real-world video production such as sports, concerts, and TV shows. Despite its potential, our experiments show that naive extensions of existing monocular or dense-fixed camera-based methods are insufficient since they fail to resolve the complex spatiotemporal inconsistencies across views and time. To fill this gap, we propose a simple yet effective 3D track initialization method designed to ensure spatiotemporal consistency by integrating inter-camera feature matching with intra-camera point tracking. Additionally, we incorporate a noise-robust depth-ordering regularization loss and a spatiotemporally diverse batch sampling strategy to enhance optimization stability and cross-view generalization. Furthermore, to address the lack of standardized benchmarks for this task, we introduce LetCamsGo, a new real-world video dataset with 5 sequences across 4 diverse environments, recorded by three independently moving cameras and one fixed camera. Comprehensive benchmarking on LetCamsGo demonstrated that our proposed framework improves 4D reconstruction quality in dynamic regions compared with baselines, paving the way for a low-cost 4D reconstruction paradigm in the wild.", "date": "2026-06-03", "categories": []}
{"id": "2511.13020", "title": "PHASE: Physiology-Aware Hyperspectral Reconstruction via Object-to-Human Domain Adaptation", "abstract": "Although hyperspectral imaging offers unparalleled non-invasive physiological insight, its bulky hardware, slow acquisition, and regulatory burden severely limit its clinical availability. A natural workaround is to reconstruct hyperspectral information from ubiquitous RGB or CASSI measurements. However, existing paradigms, developed for object-centric scenes, rely on reflectance-based feature alignment, assuming that spectral similarity preserves semantic meaning. This assumption breaks down in physiological imaging, where visually similar RGB responses may arise from distinct and entangled physiological states. This mismatch motivates a shift from reflectance alignment to physiology-aware representation learning, grounded in shared light-matter interaction principles -- a shift that introduces fundamental challenges from cross-channel semantic shifts (C1) and irreversible information loss in RGB-based acquisition (C2). We therefore design PHASE, a physiology-aware hyperspectral reconstruction paradigm that fundamentally redefines object-to-human transfer by disentangling cross-channel physiological semantics via Physiological Channel Reinterpretation and restricting reconstruction to physiologically plausible solutions through Physiologically Constrained Alignment. Under two source-to-target transfer protocols, PHASE consistently outperforms state-of-the-art methods by up to +2.20 SSIM and -3.06 in SAM with merely 1.5% labeled supervision.", "date": "2025-11-17", "categories": []}
{"id": "2511.17126", "title": "Towards Blind Lens Aberration Correction via Large LensLib Pre-training and Discrete Degradation Priors", "abstract": "Emerging deep-learning-based lens library pre-training (LensLib-PT) pipeline offers a new avenue for blind lens aberration correction by training a universal neural network, demonstrating strong capability in handling diverse unknown optical degradations. This work proposes FoundCAC, a universal foundational framework that resolves two challenges hindering the generalization of existing pipelines: the difficulty of scaling training data and the absence of prior guidance characterizing optical degradation. To improve data scalability, we expand the design specifications to increase degradation diversity and construct AODLibpro, a large-scale, unbiased lens library based on a uniform sampling strategy that quantifies spatial-variation patterns and severity. In terms of model design, to leverage Point Spread Functions (PSFs) as guidance while maintaining the blind paradigm, we propose a multi-stage vector-quantized representation learning scheme. This paradigm is specifically designed to construct a Latent PSF Representation (LPR), explicitly encoding complex continuous PSFs into a discrete degradation prior to regularize the highly ill-posed restoration process. Through a simple yet effective codebook-freezing strategy, our framework leverages the discrete prior to elevate full-shot restoration performance and unlock highly efficient few-shot adaptation for unseen lenses. Experiments on diverse aberrations of synthetic LensLib and real-world lenses demonstrate that our framework achieves state-of-the-art zero-shot generalization while enabling highly efficient few-shot adaptation for specific lenses. The source code and datasets will be made publicly available at https://github.com/zju-jiangqi/FoundCAC.", "date": "2025-11-21", "categories": []}
{"id": "2601.22443", "title": "Weak Diffusion Priors Can Still Achieve Strong Inverse-Problem Performance", "abstract": "Can a diffusion model trained on bedrooms recover human faces? Diffusion models are widely used as priors for inverse problems, but standard approaches usually assume a high-fidelity model trained on data that closely match the unknown signal. In practice, one often must use a mismatched or low-fidelity diffusion prior. Surprisingly, these weak priors often perform nearly as well as full-strength, in-domain baselines. We study when and why inverse solvers are robust to weak diffusion priors. Through extensive experiments, we find that weak priors succeed when measurements are highly informative (e.g., many observed pixels), and we identify regimes where they fail. To explain this behavior, we combine Bayesian-consistency theory with local-correlation analysis: the theory gives conditions under which high-dimensional measurements make the posterior concentrate near the true signal, while the correlation analysis shows that weak and stronger natural-image priors can share similar local spatial structure. These results provide a principled justification on when weak diffusion priors can be used reliably. Code is available at https://github.com/jjia131/weak-diffusion-priors-inverse-problem.", "date": "2026-01-30", "categories": []}
{"id": "2604.07048", "title": "PRISM: Rethinking Atmospheric Scattering Reconstruction as a Unified Understanding and Restoration Model for Real-world Dehazing", "abstract": "Real-world image dehazing (RID) aims to remove haze-induced degradation from real scenes. This task remains challenging due to non-uniform haze distribution, spatially varying color shifts, and the scarcity of paired real hazy-clean data. In PRISM, we propose Proximal Scattering Atmosphere Reconstruction (PSAR), a physically structured framework that jointly reconstructs the clear scene and scattering variables under the atmospheric scattering model, making the restoration process more interpretable in complex real-world conditions. To bridge the synthetic-to-real gap, we design an online non-uniform haze synthesis pipeline and a Selective Self-Distillation Adaptation (SSDA) scheme for unpaired real-world scenarios, which enables the model to selectively learn from high-quality perceptual targets while leveraging its intrinsic scattering understanding to audit residual haze and guide self-refinement. Experiments on real-world benchmarks demonstrate that PRISM achieves competitive performance on RID tasks.", "date": "2026-04-08", "categories": []}
{"id": "2606.03666", "title": "Beyond Single Solution: Multi-Hypothesis Collaborative Deep Unfolding Network for Image Compressive Sensing", "abstract": "Recent deep unfolding networks (DUNs) have advanced Compressive Sensing (CS) by effectively integrating iterative optimization with deep learning architectures. However, most CS approaches predominantly confine their inference to a single solution space, neglecting the inherent ill-posedness of CS problems that intrinsically permits multiple plausible candidate hypotheses. In this paper, a novel Multi-Hypothesis Collaborative Deep Unfolding CS Network (MHC-DUN) is proposed, which explicitly models and leverages multiple hypotheses by jointly optimizing across diverse solution spaces. Specifically, following the Proximal Gradient Descent algorithm, MHC-DUN jointly performs gradient descent and proximal mapping within this multi-hypothesis paradigm. i) For gradient descent, a well-designed AlphaNet is introduced to dynamically predict spatially varying step sizes for all hypotheses, enabling collaborative gradient updates across multiple solutions. ii) For proximal operator, a sophisticated multi-hypothesis collaborative proximal mapping module is designed, which leverages both intra-hypothesis and inter-hypothesis correlation priors to jointly refine multiple solutions. To enable end-to-end training, a novel composite loss function is designed, which balances measurement fidelity, hypothesis diversity, and reconstruction accuracy, encouraging exploration of complementary solutions while maintaining reconstruction fidelity. Experimental results reveal that the proposed CS method outperforms existing CS networks.", "date": "2026-06-02", "categories": []}
{"id": "2606.03806", "title": "TeX-1500: A Paired Real-World LWIR Hyperspectral Dataset and Benchmark for Temperature-Emissivity-Texture Decomposition", "abstract": "Temperature-emissivity-texture (TeX) decomposition seeks to recover object heat state, material spectral response, and visible-like geometric texture from long-wave infrared hyperspectral imaging (LWIR HSI). Existing TeX pipelines are mainly scene-specific inverse solvers, and the lack of paired LWIR HSI-TeX supervision has limited learning-based decomposition. To address this gap, we introduce TeX-1500, a large-scale paired LWIR HSI-TeX dataset and benchmark for supervised HSI-to-TeX decomposition. TeX-1500 contains 1,522 calibrated real-scene pairs from DARPA Invisible Headlights (DARPA IH) pushbroom imagery and our FTIR acquisitions, covering five locations, four seasons, diverse acquisition times, heterogeneous wavelength layouts, and two sensor families. Each sample stores a calibrated valid-band radiance cube, calibrated wavelength positions, and aligned temperature, emissivity, and texture supervision constructed through a consistent restoration and TeX-construction protocol. We further provide TeX-UNet, a simple wavelength-aware baseline that maps calibrated HSI bands and wavelength positions to TeX fields. Experiments on the held-out DARPA IH pushbroom scenes and zero-/few-shot transfer to FTIR scenes show that TeX-1500 provides usable paired supervision and a measurable benchmark for data-driven physical-property-centered thermal perception.", "date": "2026-06-02", "categories": []}
{"id": "2510.03938", "title": "Super-resolution image projection over an extended depth of field using a diffractive decoder", "abstract": "Image projection systems must be efficient in data storage, computation and transmission while maintaining a large space-bandwidth-product (SBP) at their output. Here, we introduce a hybrid image projection system that achieves extended depth-of-field (DOF) with improved resolution, combining a convolutional neural network (CNN)-based digital encoder with an all-optical diffractive decoder. A CNN-based encoder compresses input images into compact phase representations, which are subsequently displayed by a low-resolution (LR) projector and processed by an analog diffractive decoder for all-optical image reconstruction. This optical decoder is completely passive, designed to synthesize pixel super-resolved image projections that feature an extended DOF while eliminating the need for additional power consumption for super-resolved image reconstruction. Our pixel super-resolution (PSR) image projection system demonstrates high-fidelity image synthesis over an extended DOF of ~267xW, where W is the illumination wavelength, concurrently offering up to ~16-fold SBP improvement at each lateral plane. The proof of concept of this approach is validated through an experiment conducted in the THz spectrum, and the system is scalable across different parts of the electromagnetic spectrum. This image projection architecture can reduce data storage and transmission requirements for display systems without imposing additional power constraints on the optical decoder. Beyond extended DOF PSR image projection, the underlying principles of this approach can be extended to various applications, including optical metrology and microscopy.", "date": "2025-10-04", "categories": []}
{"id": "2511.07438", "title": "Two Datasets Are Better Than One: Method of Double Moments for 3-D Reconstruction in Cryo-EM", "abstract": "Cryo-electron microscopy (cryo-EM) is a powerful imaging technique for reconstructing three-dimensional molecular structures from noisy tomographic projection images of randomly oriented particles. We introduce a new data fusion framework, termed the method of double moments (MoDM), which reconstructs molecular structures from two instances of the second-order moment of projection images obtained under distinct orientation distributions: one uniform, the other non-uniform and unknown. We prove that these moments generically uniquely determine the underlying structure, up to a global rotation and reflection, and we develop a convex-relaxation-based algorithm that achieves accurate recovery using only second-order statistics. Our results demonstrate the advantage of collecting and modeling multiple datasets under different experimental conditions, illustrating that leveraging dataset diversity can substantially enhance reconstruction quality in computational imaging tasks.", "date": "2025-11-02", "categories": []}
{"id": "2512.20251", "title": "Degradation-Aware Metric Prompting for Hyperspectral Image Restoration", "abstract": "Unified hyperspectral image (HSI) restoration aims to recover diverse degradations within a single model. However, current methods often rely on impractical explicit priors or opaque black-box representations that overfit to training distributions, hampering generalization to unseen scenarios. To bridge this gap, we propose Degradation-Aware Metric Prompting (DAMP), a novel framework that characterizes multi-dimensional degradations through interpretable spatial-spectral metrics. These metrics serve as Degradation Prompts (DP), enabling the model to capture shared characteristics across tasks and adapt to unknown corruptions. Central to our framework is the Degradation-Adaptive Mixture-of-Experts (DAMoE), where Spatial-Spectral Adaptive Modules (SSAMs) serve as experts that utilize learnable fusion coefficients to specialize in distinct degradation degrees. By using DP as a gating router, DAMoE dynamically activates specialized experts tailored to the specific degradation profile. Extensive experiments on natural and remote sensing HSI datasets demonstrate that DAMP achieves state-of-the-art performance and exhibits exceptional zero-shot generalization on unseen restoration tasks. Code is publicly available at \\href{DAMP}{https://github.com/MiliLab/DAMP}.", "date": "2025-12-23", "categories": []}
{"id": "2603.23647", "title": "{\\lambda}Split: Self-Supervised Content-Aware Spectral Unmixing for Fluorescence Microscopy", "abstract": "In fluorescence microscopy, spectral unmixing aims to recover individual fluorophore concentrations from spectral images that capture mixed fluorophore emissions. Since classical methods operate pixel-wise and rely on least-squares fitting, their performance degrades with increasingly overlapping emission spectra and higher levels of noise, suggesting that a data-driven approach that can learn and utilize a structural prior might lead to improved results. Learning-based approaches for spectral imaging do exist, but they are either not optimized for microscopy data or are developed for very specific cases that are not applicable to fluorescence microscopy settings. To address this, we propose {\\lambda}Split, a physics-informed deep generative model that learns a conditional distribution over concentration maps using a hierarchical Variational Autoencoder. A fully differentiable Spectral Mixer enforces consistency with the image formation process, while the learned structural priors enable state-of-the-art unmixing and implicit noise removal. We demonstrate {\\lambda}Split on 3 real-world datasets that we synthetically cast into a total of 66 challenging spectral unmixing benchmarks. We compare our results against a total of 10 baseline methods, including classical methods and a range of learning-based methods. Our results consistently show competitive performance and improved robustness in high noise regimes, when spectra overlap considerably, or when the spectral dimensionality is lowered, making {\\lambda}Split a new state-of-the-art for spectral unmixing of fluorescent microscopy data. Importantly, {\\lambda}Split is compatible with spectral data produced by standard confocal microscopes, enabling immediate adoption without specialized hardware modifications.", "date": "2026-03-24", "categories": []}
{"id": "2605.21964", "title": "Dual-Integrated Low-Latency Single-Lens Infrared Computational Imaging for Object Detection", "abstract": "Computational imaging enables compact infrared systems, but deep-learning pipelines that combine image reconstruction and object detection often introduce substantial inference latency. Most existing acceleration strategies compress the reconstruction network while overlooking physical priors from the optical path, leaving a trade-off between accuracy and speed. We present Physics-aware Dual-Integrated Network (PDI-Net), a low-latency framework that integrates infrared reconstruction with object detection and further embeds optical priors into the learning process. PDI-Net uses a supervised U-Net during training, while a semi-U-Net encoder shares features directly with a YOLO-based detector during inference, avoiding full image reconstruction. To bridge the gap between fidelity-oriented reconstruction features and detection-oriented semantics, we introduce a physics-aware large-small bridge (PALS-Bridge), which uses field-dependent point spread function priors to adaptively modulate multiscale convolutional branches. A physics-informed optical degradation simulation pipeline is also developed for training and validation. The method is deployed on a single-lens infrared camera, reducing system weight by about 50% compared with traditional multi-lens designs. On the M3FD benchmark under low-SNR conditions, PDI-Net reduces inference time by 84.06% compared with the Rec+Det with pruning strategy while improving mAP@0.5:0.95 by 5.07%. These results demonstrate compact, low-latency computational infrared imaging for real-time object detection on resource-constrained platforms.", "date": "2026-05-21", "categories": []}
{"id": "2606.00228", "title": "LithoGRPO: Fast Inverse Lithography via GRPO Reinforced Flow Matching", "abstract": "In semiconductor manufacturing, lithography projects circuit layouts onto silicon wafers through an optical mask. As circuit features shrink below the wavelength of light, optical diffraction causes the printed patterns to deviate from their intended layouts. Inverse Lithography Technology (ILT) addresses this challenge by generating optimized masks that enhance the fidelity of pattern transfer onto wafers. While ILT resembles an image synthesis task, its reliance on explicit physical metrics for mask evaluation limits the applicability of existing generative models. We introduce LithoGRPO, an ILT framework that integrates the flow-matching paradigm with GRPO-based reinforcement learning (RL) fine-tuning, enabling efficient exploration of diverse masks for a given target layout. Unlike purely generative or optimization-based approaches, RL in LithoGRPO exploits the explicitly defined, physics-based reward function of ILT, enabling optimization under complex, process-aware constraints. To the best of our knowledge, this is the first framework that unifies flow matching and RL for mask optimization. To improve RL sampling efficiency, we propose a fast shot-counting algorithm for manufacturability evaluation, achieving over 130x speedup while preserving the mask ranking of the traditional shot-count metric. Extensive experiments demonstrate that LithoGRPO achieves state-of-the-art performance over both optimization-based and learning-based methods, while maintaining efficient mask generation.", "date": "2026-05-29", "categories": []}
{"id": "2606.00509", "title": "Structure-Aware Consistency Priors for Shape from Polarization in Complex Media", "abstract": "Recovering surface normals from single view polarization images in complex media remains challenging. This paper focuses on ice as a representative complex medium, where intricate light matter interactions lead to a nonlinear mapping between polarization observations and surface normals. To address this, a structure-aware polarization prior based on autocorrelation functions is proposed to capture the local spatial consistency of AoLP. Building on this, a dual-branch network (IceSfP) is designed to integrate raw polarization features with priors via cross modal attention and multi-scale feature fusion, enabling accurate surface normal estimation under complex media conditions. To evaluate the method, the first real-world ice SfP dataset is constructed. Experimental results show that the method outperforms existing approaches across all metrics, achieving a MAE of 16.01 deg, which is 2.74 deg lower than the second-best method. The framework provides a generalizable solution for high-precision geometric perception in complex media.", "date": "2026-05-30", "categories": []}
{"id": "2606.00704", "title": "VICR: Visual In-Context Restoration for Real-World Image Super-Resolution", "abstract": "Real-world image super-resolution (Real-ISR) requires balancing structural fidelity to degraded observations with realistic detail synthesis. However, existing generative Real-ISR methods often rely on entangled conditioning mechanisms, leading to structural drift or semantically inconsistent details. To address this issue, we propose Visual In-Context Restoration (VICR), a Diffusion Transformer (DiT)-based framework that formulates Real-ISR as image completion. Specifically, we introduce a decoupled visual prior injection mechanism that derives local and global cues from the low-quality (LQ) image: local cues help recover image structures and support high-frequency detail synthesis, while global cues guide overall generation and promote semantic consistency. For ambiguous regions under severe degradation, VICR employs an inference-time agent to refine semantic prompts using visual evidence from the LQ input while keeping model parameters fixed. Experiments show that VICR achieves state-of-the-art performance across multiple Real-ISR benchmarks with only 127M trainable parameters.", "date": "2026-05-30", "categories": []}
{"id": "2606.02309", "title": "Measurement Geometry and Design for Trustworthy Generative Inverse Problems", "abstract": "Generative models are increasingly used as priors for inverse problems, but their ability to produce realistic images creates a basic trust problem: a plausible reconstruction may be supported by the measurements, or it may be filled in by the prior along unobserved directions. This distinction is especially important in medical imaging, where acquisition operators are designed under scan-time, dose, and calibration constraints. We study generative inverse problems from a measurement-geometry perspective. The central question is whether a fixed measurement operator can distinguish nearby images that are plausible under the generative prior, and whether this relationship can guide better measurements. We introduce a local measurement-manifold compatibility measure that quantifies how well the operator observes prior-relevant tangent directions. Under local regularity assumptions, we prove that this quantity controls the stable part of the reconstruction error, while the generative prior controls off-manifold drift. This worst-direction certificate motivates practical fixed and sequential acquisition rules based on overall local volume preservation, including a posterior-cloud design that adapts measurements at test time without training a sampling policy. Across row-sampling, tomographic, and MR acquisition settings, the proposed scores predict failure modes, explain measurement-induced hallucinations, and guide better sampling. In fastMRI Cartesian sampling, posterior-cloud measurement design improves over strong non-learned ACS-preserving baselines, including variable-density and Poisson-like masks.", "date": "2026-06-01", "categories": []}
{"id": "2606.02331", "title": "Hallucination-Aware Diffusion Sampling for Inverse Problems via Robust Prior Updates", "abstract": "Diffusion-based inverse problem solvers can produce realistic reconstructions, but realism alone does not ensure that the recovered details are supported by the measurement. We study this failure as measurement-conditioned hallucination: visually meaningful content that is either implausible or inconsistent with the measured instance. Our analysis separates Bayes-rule-based diffusion inverse solvers into a prior update and a measurement-conditioning step, showing that hallucinated content can enter through the prior-side proposal before the measurement correction is applied. Motivated by this view, we propose Robust Prior Update (RPU), a solver-level module that probes the local stability of the diffusion prior update, re-anchors the resulting displacement at the current iterate, and leaves the measurement update unchanged. We instantiate RPU in DPS and evaluate it on FFHQ and ImageNet inverse problems using automatic metrics and human faithfulness studies. On FFHQ, RPU improves PSNR and LPIPS over DPS across box inpainting, Gaussian deblurring, and motion deblurring. In human judgments, RPU receives 91.9% of blind non-tie majority preferences and 91.1% of ground-truth-assisted non-tie preferences on FFHQ box inpainting, while the ImageNet Gaussian reader study is tie-heavy but favors RPU among non-tie cases. These results support a targeted claim: robustifying the prior update can improve instance faithfulness in diffusion inverse solvers, especially when the prior shapes weakly constrained content.", "date": "2026-06-01", "categories": []}
{"id": "2606.02565", "title": "Policy-based Foveated Imaging and Perception", "abstract": "Ultra-high-resolution image sensors offer the potential to capture fine spatial details critical for many visual perception tasks, but acquiring and processing all pixels at full resolution is often infeasible under realistic bandwidth, latency, and power constraints. Existing approaches address this challenge through acquisition strategies such as spatial or temporal downsampling, which irrevocably discard information before task relevance can be assessed. In this work, we introduce a real-time, predictive, and task-aware foveated imaging system that operates directly at image acquisition time. Leveraging emerging dual-stream sensor architectures, our method dynamically allocates limited pixel bandwidth to task-relevant regions of interest while maintaining a low-resolution global context. We formulate foveated acquisition as a sensor attention policy-learning problem, in which past observations guide actions that determine future measurements, closing the perception-acquisition loop. Through extensive simulation across multiple perception tasks, we demonstrate that our approach achieves high task performance under strict pixel budgets and significantly outperforms relevant baselines operating at the same bandwidth. We further validate our system on a 200-megapixel dual-stream sensor, capturing real-world videos under realistic bandwidth and latency constraints, demonstrating the practical feasibility of task-driven, acquisition-time foveated imaging.", "date": "2026-06-01", "categories": []}
{"id": "2509.25269", "title": "Position-Blind Ptychography: Viability of image reconstruction via data-driven variational inference", "abstract": "In this work, we present and investigate the novel blind inverse problem of position-blind ptychography, i.e., ptychographic phase retrieval without any knowledge of scan positions, which then must be recovered jointly with the image. The motivation for this problem comes from single-particle diffractive X-ray imaging, where particles in random orientations are illuminated and a set of diffraction patterns is collected. If one uses a highly focused X-ray beam, the measurements would also become sensitive to the beam positions relative to each particle and therefore ptychographic, but these positions are also unknown. We investigate the viability of image reconstruction in a simulated, simplified 2-D variant of this difficult problem, using variational inference with modern data-driven image priors in the form of score-based diffusion models. We find that, with the right illumination structure and a strong prior, one can achieve reliable and successful image reconstructions even under measurement noise, in all except the most difficult evaluated imaging scenario.", "date": "2025-09-28", "categories": []}
{"id": "2604.10273", "title": "Dual-Exposure Imaging with Events", "abstract": "By combining complementary benefits of short- and long-exposure images, Dual-Exposure Imaging (DEI) enhances image quality in low-light scenarios. However, existing DEI approaches inevitably suffer from producing artifacts due to spatial displacement from scene motion and image feature discrepancies from different exposure times. To tackle this problem, we propose a novel Event-based DEI (E-DEI) algorithm, which reconstructs high-quality images from dual-exposure image pairs and events, leveraging high temporal resolution of event cameras to provide accurate inter-/intra-frame dynamic information. Specifically, we decompose this complex task into an integration of two sub-tasks, i.e., event-based motion deblurring and low-light image enhancement tasks, which guides us to design E-DEI network as a dual-path parallel feature propagation architecture. We propose a Dual-path Feature Alignment and Fusion (DFAF) module to effectively align and fuse features extracted from dual-exposure images with assistance of events. Furthermore, we build a real-world Dataset containing Paired low-/normal-light Images and Events (PIED). Experiments on multiple datasets show the superiority of our method. The code and dataset are available at github.", "date": "2026-04-11", "categories": []}
{"id": "2605.30541", "title": "SubsurfaceGen: Procedural Generation of Field-Scale Earth Models and Seismic Data", "abstract": "Full waveform inversion (FWI) is the gold standard for subsurface imaging, with applications from carbon sequestration to energy and mineral exploration to earthquake hazard assessment. Machine learning approaches to FWI need field-scale, geologically diverse, and physically realistic training data, but existing resources such as Marmousi, SEAM, and OpenFWI fall short on spatial extent, temporal extent, geological diversity, and physical realism. We address these limitations with SubsurfaceGen, a GPU-accelerated generator for 3D velocity models and seismic data. Along with SubsurfaceGen, we release a paired dataset of 4,276 2D velocity slices, 5 s wavefields, and 8 s shot gathers drawn from 42 realistic, field-scale 3D velocity models, each spanning 10 km x 10 km laterally and 6.19 km deep at 10 m resolution. The dataset spans six geological settings -- four built with SubsurfaceGen and two drawn from prior sources -- relevant for carbon sequestration and hydrocarbon exploration. We use this dataset to evaluate neural operators on wavefield prediction and encoder-decoders on end-to-end velocity inversion, holding out one geological setting for out-of-distribution testing. These experiments surface failure modes at field-scale and demonstrate how SubsurfaceGen and the associated dataset can impact ML-based FWI.", "date": "2026-05-28", "categories": []}
{"id": "2605.31302", "title": "MoE-dqINR: A Unified Mixture-of-Experts Implicit Neural Representation Framework for Scan-Specific Dynamic and Quantitative MRI Reconstruction", "abstract": "Undersampled magnetic resonance imaging (MRI) reconstruction seeks to recover temporally or contrast-varying image series from incomplete multicoil k-space data while preserving state-dependent fidelity for dynamic and quantitative MRI (qMRI). Existing scan-specific implicit neural representations (INRs) often use monolithic spatiotemporal coordinate fields, explicit subspaces, motion or deformation models, calibration variables, or sequence-specific quantitative signal models. These design choices can limit flexibility in sharing spatial information while adapting image synthesis across acquisition states. Moreover, many INR-based baselines remain computationally demanding, typically requiring per-scan optimization times on the order of hundreds to thousands of seconds. We propose MoE-dqINR, a scan-specific multicoil MRI reconstruction framework that factorizes the image-domain representation into shared spatial experts and a state-conditioned routing pathway. Spatial experts encode reusable coordinate-dependent image content, whereas routing weights, conditioned on ordered acquisition states, synthesize each dynamic frame or contrast state from a common expert bank. The representation is coupled to a multicoil MRI forward model, uses the normalized state index to drive routing in both dynamic and quantitative MRI. By separating shared spatial representation from state-dependent synthesis, the framework provides an image-first architecture for dynamic and quantitative MRI while reducing scan-specific INR optimization to approximately 30 s per scan in our experiments. The proposed formulation establishes state-conditioned mixture-of-experts INR as a scan-specific multicoil MRI reconstruction prior that unifies shared spatial representation, dynamic- and qMRI-specific synthesis, and practical per-scan efficiency.", "date": "2026-05-29", "categories": []}
{"id": "2605.31400", "title": "FSM-Net: An Efficient Frequency-Spatial Network for Real-World Deblurring", "abstract": "Real-world image deblurring demands both high-fidelity restoration and computational efficiency, a balance existing methods often struggle to achieve. In this paper, we propose FSM-Net (Frequency-Spatial Multi-branch Network), a highly efficient solution that secured 2nd place in the NTIRE 2026 Challenge on Efficient Real-World Deblurring. FSM-Net pioneers a dual-domain approach: a novel Frequency Attention module explicitly recovers high-frequency structural details via FFT, while a Cross-Gated Vision E-Branchformer at the bottleneck captures global dependencies with linear complexity. To ensure robust convergence, we employ a progressive curriculum training strategy guided by a composite loss function (Multi-Scale Charbonnier, Structural Edge, and Frequency). Evaluated on the RSBlur benchmark, FSM-Net achieves an outstanding 33.144 dB PSNR with only 4.94M parameters and 159.35 GMACs (at 1920x1200 resolution). By effectively pushing the Pareto frontier of efficiency and quality, FSM-Net establishes a strong baseline for resource-constrained image restoration.", "date": "2026-05-29", "categories": []}
{"id": "2605.31426", "title": "Self-Tuning Regularization for Image Scanning Microscopy", "abstract": "Image Scanning Microscopy (ISM) is a fluorescence imaging technique that combines detector-array acquisition and computational reconstruction to achieve the theoretical resolution of an ideal confocal microscope, i.e., one operating with an infinitesimally small pinhole, while maintaining high signal-to-noise ratio. Among the reconstruction methods for obtaining the super-resolved image, multi-image deconvolution (MID) and its extension aimed at preserving the optical sectioning capability of confocal microscopy, known as super-resolution sectioning ISM (s$^2$ISM), are among the most widely used approaches. Both methods rely on Richardson--Lucy-type iterative schemes, whose semi-convergent behavior requires early stopping and often leads to noise amplification and reconstruction artifacts. In this work, we introduce a self-tuning explicit regularization framework for both MID and s$^2$ISM reconstruction. Within a Bayesian maximum a posteriori formulation, we combine a multi-frame Poisson data fidelity term with explicit regularization, considering $\\ell_1$ and smoothed total variation penalties as representative examples. We further develop an automatic and ground-truth-free strategy for regularization parameter selection by adapting the residual whiteness principle to the multi-frame Poisson setting and introducing a spectral high-pass extension tailored to s$^2$ISM. The resulting framework enables stable reconstructions without empirical stopping rules. To demonstrate the proposed framework, we consider first-order optimization schemes based on proximal gradient and mirror descent methods with adaptive backtracking strategies. Experiments on simulated and real fluorescence ISM datasets demonstrate improved reconstruction stability and image quality with respect to unregularized approaches, while enabling robust super-resolution and optical sectioning in low-photon conditions.", "date": "2026-05-29", "categories": []}
{"id": "2605.31596", "title": "KLIP: localized distribution shift detection via KL-divergence with diffusion priors in Inverse Problems", "abstract": "Diffusion models have shown promising performance as data-driven priors for computational imaging, as well as some capacity to detect out-of-distribution (OOD) images. However, existing approaches to OOD detection often require some knowledge of the shifted distribution, fail to detect subtle or localized distribution shifts, and operate on full images, rather than the indirect measurements available in inverse problems. We propose an OOD detection metric based on the Kullback-Leibler divergence between the diffusion prior and the posterior distribution, that (i) does not require any calibration data or knowledge of the shifted distribution, and (ii) can detect whole images as OOD as well as localize OOD patches within an image. Experimentally, we show that this metric can detect subtle yet semantically meaningful distribution shifts, such as the shift from healthy liver CT scans to those with tumors, and generalizes across different types of diffusion models, datasets, and inverse problems. Our code can be found at https://github.com/voilalab/KLIP.", "date": "2026-05-29", "categories": []}
{"id": "2508.15151", "title": "Zero-shot CT Super-Resolution using Diffusion-based 2D Projection Priors and Signed 3D Gaussians", "abstract": "Computed tomography (CT) is important in clinical diagnosis, but acquiring high-resolution (HR) CT is constrained by radiation exposure risks. While deep learning-based super-resolution (SR) methods have shown promise for reconstructing HR CT from low-resolution (LR) inputs, supervised approaches require paired datasets that are often unavailable. Zero-shot methods address this limitation by operating on single LR inputs; however, they frequently fail to recover fine structural details due to limited LR information within individual volumes. To overcome these limitations, we propose a novel zero-shot 3D CT SR framework that integrates diffusion-based upsampled 2D projection priors into the 3D reconstruction process. Specifically, our framework consists of two stages: (1) LR CT projection SR, training a diffusion model on abundant X-ray data to upsample LR projections, thereby enhancing the scarce information inherent in the LR inputs. (2) 3D CT volume reconstruction, using 3D Gaussian splatting with our novel Negative Alpha Blending (NAB-GS), which models positive and negative Gaussian densities to learn signed residuals between diffusion-generated HR and upsampled LR projections. Our framework demonstrates superior quantitative and qualitative performance on two public datasets, and expert evaluations present the framework's clinical potential at 4x.", "date": "2025-08-21", "categories": []}
{"id": "2510.27663", "title": "Bayesian model selection and misspecification testing in imaging inverse problems only from noisy and partial measurements", "abstract": "Modern imaging techniques heavily rely on Bayesian statistical models to address difficult image reconstruction and restoration tasks. This paper addresses the objective evaluation of such models in settings where ground truth is unavailable, with a focus on model selection and misspecification diagnosis. Existing unsupervised model evaluation methods are often unsuitable for computational imaging due to their high computational cost and incompatibility with modern image priors defined implicitly via machine learning models. We herein propose a general methodology for unsupervised model selection and misspecification detection in Bayesian imaging sciences, based on a novel combination of Bayesian cross-validation and data fission, a randomized measurement splitting technique. The approach is compatible with any Bayesian imaging sampler, including diffusion and plug-and-play samplers. We demonstrate the methodology through experiments involving various scoring rules and types of model misspecification, where we achieve excellent selection and detection accuracy with a low computational cost.", "date": "2025-10-31", "categories": []}
{"id": "2603.00377", "title": "Improving Full Waveform Inversion in Large Model Era", "abstract": "Full Waveform Inversion (FWI) is a highly nonlinear and ill-posed problem that aims to recover subsurface velocity maps from surface-recorded seismic waveforms data. Existing data-driven FWI typically uses small models, as available datasets have limited volume, geological diversity, and spatial extent, leading to substantial concerns about overfitting. Although they perform well on synthetic datasets, current methods fail to generalize to more realistic geological structures. In this work, we show that a model trained entirely on simulated and relatively simple data can generalize remarkably well to challenging and unseen geological benchmarks. We provide a working recipe that tames a billion-parameter model for FWI through coordinated scaling across three axes: model capacity, data diversity, and training strategy. Our model achieves state-of-the-art performance on OpenFWI and significantly narrows the generalization gap in data-driven FWI. Across six challenging geophysical benchmarks, including Marmousi, 2D SEG/EAGE Salt and Overthrust, 2004 BP, Sigsbee, and SEAM Phase I, it infers complex structures absent from the training set and delivers significant performance improvements (SSIM from 0.5844 to 0.7669). Overall, our results demonstrate that with an appropriate scaling strategy, large models trained on simple synthetic data can achieve substantial generalization to more complex and realistic geological structures.", "date": "2026-02-27", "categories": []}
{"id": "2605.23531", "title": "PixIE: Prompted Pixel-Space Low-Light Image Enhancement", "abstract": "Low-light images suffer from severe noise, contrast loss, and semantic ambiguity, making enhancement a joint problem of denoising and detail recovery. We propose PixIE, a feed-forward pixel-space LLIE framework semantically prompted by a vision foundation model. PixIE first performs cross-scale denoising to suppress noise and preserve structure, then refines details using DINO-Prompted Pixel Blocks (DPPBs), which inject intermediate DINOv3 features through patch-conditioned, spatially continuous per-pixel modulation. To make pixel-space attention efficient across scales, we introduce Spatial-Channel Compaction (SCC), which jointly reduces the spatial token grid and channel dimension. We further propose Multi-Receptive-Field Pixel Embedding (MRPE) to provide neighborhood-aware pixel representations before semantic prompting, improving robustness to signal-dependent noise beyond point-wise embeddings. Experiments on LLIE benchmarks show that PixIE improves average PSNR by 1.9-15.0% over recent state-of-the-art methods and reduces LPIPS by 8.5-44.4%. Qualitative comparisons further show sharper details and more stable textures, improving both reconstruction fidelity and perceptual quality.", "date": "2026-05-22", "categories": []}
{"id": "2605.28711", "title": "Stage-wise Distortion-Perception Traversal in Zero-shot Inverse Problems with Diffusion Models", "abstract": "The distortion-perception (D-P) tradeoff is a fundamental phenomenon of Bayesian inverse problems, which characterizes the inherent tension between distortion performance and perceptual quality. Enabling flexible traversal of the D-P tradeoff at inference time is crucial for practical applications. Despite the recent success of diffusion models in zero-shot inverse problem solving, efficient and principled strategies for D-P traversal in diffusion-based inverse algorithms remain inadequately characterized. In this paper, we propose a stage-wise framework for realizing D-P traversal using a single diffusion model in zero-shot inverse problems. Our proposed method, termed MAP-RPS, starts with an MAP estimation stage that approximates the MMSE solution and provides a low-distortion initialization, followed by a re-noised posterior sampling stage that progressively improves perceptual quality. We provide theoretical analyses for both stages, establishing the validity and effectiveness of the proposed design. Furthermore, we extend MAP-RPS to the latent space, yielding LMAP-RPS, which enjoys broader applicability by leveraging large-scale pre-trained latent diffusion backbones. Extensive experiments demonstrate that MAP-RPS and LMAP-RPS enable more effective D-P traversal on various tasks, while also exhibiting strong performance as efficient solvers for real-world inverse problems.", "date": "2026-05-27", "categories": []}
{"id": "2605.29012", "title": "Trajectory Constraints for Imaging Inverse Problems", "abstract": "Diffusion-based and iterative methods have become effective tools for solving imaging inverse problems. Their reconstruction process naturally forms a trajectory of intermediate estimates. Although these intermediate estimates define a reconstruction trajectory, most methods do not explicitly regularize the transitions between consecutive states. To address this limitation, we introduce TRACE, a training-free TRAjectory-Constrained rEconstruction framework that stabilizes the reconstruction path by coupling adjacent states along the trajectory. This gives a trajectory-level model that can be interpreted as a sequence of proximal updates. Since the exact proximal update is generally intractable, we approximate it with a neural mapping. This yields a diffusion-like reconstruction process with an explicit coupling between neighboring states. We provide a stability analysis showing that temporal coupling bounds trajectory variation and that this control is preserved under untrained network updates. Experiments on linear and nonlinear image reconstruction tasks show that TRACE improves reconstruction quality. Trajectory-level analyses and ablations confirm that temporal coupling directly affects state transitions along the reconstruction path.", "date": "2026-05-27", "categories": []}
{"id": "2504.04924", "title": "Inter-event Interval Microscopy for Event Cameras", "abstract": "Event cameras, an innovative bio-inspired sensor, differ from traditional cameras by sensing changes in intensity rather than directly perceiving intensity and recording these variations as a continuous stream of \"events\". The intensity reconstruction from these sparse events has long been a challenging problem. Previous approaches mainly focused on transforming motion-induced events into videos or achieving intensity imaging for static scenes by integrating modulation devices at the event camera acquisition end. In this paper, for the first time, we achieve event-to-intensity conversion using a static event camera for both static and dynamic scenes in fluorescence microscopy. Unlike conventional methods that primarily rely on event integration, the proposed Inter-event Interval Microscopy (IEIM) quantifies the time interval between consecutive events at each pixel. With a fixed threshold in the event camera, the time interval can precisely represent the intensity. At the hardware level, the proposed IEIM integrates a pulse light modulation device within a microscope equipped with an event camera, termed Pulse Modulation-based Event-driven Fluorescence Microscopy. Additionally, we have collected IEIMat dataset under various scenes including high dynamic range and high-speed scenarios. Experimental results on the IEIMat dataset demonstrate that the proposed IEIM achieves superior spatial and temporal resolution, as well as a higher dynamic range, with lower bandwidth compared to other methods. The code and the IEIMat dataset will be made publicly available.", "date": "2025-04-07", "categories": []}
{"id": "2512.00814", "title": "IRPO: Boosting Image Restoration via Post-training GRPO", "abstract": "Post-training has become effective for high-level generation, but its role in low-level vision remains underexplored. Existing image restoration methods often rely on fixed pixel-wise fitting to ground-truth images, which can lead to over-smoothing and weak generalization. We propose IRPO, a GRPO-based post-training framework for deterministic restoration models. IRPO is built around two axes: data formulation and reward modeling. For data formulation, we select the 30% underperforming samples from the pre-training stage, which improves both accuracy and training efficiency. For reward modeling, we combine fidelity-oriented and quality-aware feedback with three components: a General Reward for structural fidelity, an Expert Reward that uses a Vision-Language Model as a coarse visual-quality judge, and a Restoration Reward for task-specific low-level cues. Experiments on six in-domain and five out-of-domain (OOD) benchmarks show that IRPO improves the AdaIR baseline by 0.93 dB on in-domain tasks and 3.43 dB on OOD settings. Our code can be shown in https://github.com/HaoxuanXU1024/IRPO.", "date": "2025-11-30", "categories": []}
{"id": "2601.10334", "title": "An analytic theory of convolutional neural network inverse problems solvers", "abstract": "Supervised convolutional neural networks (CNNs) are widely used to solve imaging inverse problems, achieving state-of-the-art performance in numerous applications. However, despite their empirical success, these methods are poorly understood from a theoretical perspective and often treated as black boxes. To bridge this gap, we analyze trained neural networks through the lens of the Minimum Mean Square Error (MMSE) estimator, incorporating functional constraints that capture two fundamental inductive biases of CNNs: translation equivariance and locality via finite receptive fields. Under the empirical training distribution, we derive an analytic, interpretable, and tractable formula for this constrained variant, termed Local-Equivariant MMSE (LE-MMSE). Through extensive numerical experiments across various inverse problems (denoising, inpainting, deconvolution), datasets (FFHQ, CIFAR-10, FashionMNIST), and architectures (U-Net, ResNet, PatchMLP), we demonstrate that our theory matches the neural networks outputs (PSNR $\\gtrsim25$dB). Furthermore, we provide insights into the differences between \\emph{physics-aware} and \\emph{physics-agnostic} estimators, the impact of high-density regions in the training (patch) distribution, and the influence of other factors (dataset size, patch size, etc).", "date": "2026-01-15", "categories": []}
{"id": "2605.27460", "title": "D$^2$Turb: Depth-Aware Simulation and Decoupled Learning for Single-Frame Atmospheric Turbulence Mitigation", "abstract": "Single-frame atmospheric turbulence mitigation is inherently ill-posed due to spatially varying blur coupled with non-rigid geometric distortion. Existing end-to-end approaches trained on flat-field simulations often struggle to balance texture recovery with geometric rectification. To overcome this limitation, we propose D$^2$Turb, a unified framework that bridges physics-grounded simulation with explicitly decoupled restoration. First, we introduce a Depth-Aware Turbulence Synthesis protocol that incorporates scene depth into the phase-to-space formulation. This generates physically consistent, depth-dependent degradations and provides a crucial intermediate tilt supervision signal for disentangled learning. Building upon this simulation engine, D$^2$Turb decomposes restoration into two interactive stages: texture deblurring and geometric rectification. The texture deblurring stage employs a deblurring backbone to recover fine-grained details while preserving geometric distortion for the subsequent rectification stage. To mitigate the information fragmentation commonly observed in cascaded designs, we further propose an Adaptive Structural Prior Injection (ASPI) mechanism that dynamically transfers deep structural representations from the deblurring module to guide dense flow prediction for spatial unwarping. Extensive experiments demonstrate that D$^2$Turb achieves state-of-the-art performance on both synthetic and real-world datasets, with consistent improvements in both texture recovery and geometric fidelity. Our code and pre-trained models are publicly available at https://github.com/HertzDot222/D2Turb.", "date": "2026-05-25", "categories": []}
{"id": "2605.28124", "title": "Gradient Step Plug-and-Play Model for Dental Cone-Beam CT Reconstruction", "abstract": "The goal of this work is to reduce the effect of photon noise in dental cone-beam CT reconstruction. We consider an inverse problem formulation and develop a databased prior. To this end, we simulate fan-beam acquisitions and add photon noise to the projection data. The prior is obtained by training a gradient-step denoiser using reconstructed simulated acquisitions. The trained model is integrated into a plug-and-play gradient-step algorithm to reconstruct images from simulated projections. Experiments on synthetic data demonstrate the denoising capabilities of the trained model, while qualitative evaluations on real images showcase the algorithm's performance and generalization ability.", "date": "2026-05-27", "categories": []}
{"id": "2605.28392", "title": "Bound-Constrained Sparse Representation for Electrical Impedance Tomography", "abstract": "This study proposes a bound-constrained sparse representation (BC-SR) framework for electrical impedance tomography (EIT), aimed at improving conductivity estimation without explicit regularization. BC-SR adopts a representation-driven strategy, generating conductivity from low-dimensional latent variables via an implicit composite parameterization. Structural priors are embedded using a truncated graph-Laplacian basis, while a bound-preserving nonlinear mapping enforces admissible conductivity ranges and improves conditioning through implicit gradient modulation. The approach ensures robust convergence, even under noisy or incomplete data. Extensive validation on 2D/3D simulations, tank experiments, and in-vivo lung data shows that BC-SR improves physical consistency and structural fidelity, offering enhanced robustness compared to traditional methods. Additionally, BC-SR enables 3D time-difference EIT reconstruction, offering improved spatial resolution and a more coherent representation of 3D conductivity distributions, particularly for in-vivo lung data. This suggests potential for improved performance in EIT, particularly in clinical applications for respiratory monitoring.", "date": "2026-05-27", "categories": []}
{"id": "2504.07853", "title": "V2V3D: View-to-View Denoised 3D Reconstruction for Light-Field Microscopy", "abstract": "Light field microscopy (LFM) has gained significant attention due to its ability to capture snapshot-based, large-scale 3D fluorescence images. However, existing LFM reconstruction algorithms are highly sensitive to sensor noise or require hard-to-get ground-truth annotated data for training. To address these challenges, this paper introduces V2V3D, an unsupervised view2view-based framework that establishes a new paradigm for joint optimization of image denoising and 3D reconstruction in a unified architecture. We assume that the LF images are derived from a consistent 3D signal, with the noise in each view being independent. This enables V2V3D to incorporate the principle of noise2noise for effective denoising. To enhance the recovery of high-frequency details, we propose a novel wave-optics-based feature alignment technique, which transforms the point spread function, used for forward propagation in wave optics, into convolution kernels specifically designed for feature alignment. Moreover, we introduce an LFM dataset containing LF images and their corresponding 3D intensity volumes. Extensive experiments demonstrate that our approach achieves high computational efficiency and outperforms the other state-of-the-art methods. These advancements position V2V3D as a promising solution for 3D imaging under challenging conditions.", "date": "2025-04-10", "categories": []}
{"id": "2512.01572", "title": "Reconstructing Multi-Scale Physical Fields from Extremely Sparse Measurements with an Autoencoder-Diffusion Cascade", "abstract": "Extreme sensor sparsity makes full-field reconstruction a fundamentally ill-posed problem in scientific sensing,where the goal is to infer physical fields from sparse measurements.In this regime,the posterior is severely underconstrained and inherently multimodal,making its approximation highly ill-conditioned.Specifically,deterministic mappings collapse uncertainty,direct conditional learning cannot cover the space of possible observation-conditioned solutions,and likelihood-guided sampling becomes highly sensitive to noise and sensor configurations.These limitations result in unstable posterior estimates and highlight the need for modeling uncertainty in a structural manner.To this end,we propose Cascaded Sensing,a hierarchical framework that restructures posterior inference across scales.Rather than modeling the full-field posterior directly,Cas-Sensing first resolves global structural ambiguity through a deterministic coarse-stage estimator.A neural-operator-based functional autoencoder,trained with masked inputs,maps sparse observations to a coarse-scale structural field,acting analogously to a maximum a posteriori estimator that selects the dominant global configuration.This structural anchor fixes the principal degrees of freedom of the posterior and transforms the problem into a better-conditioned residual inference task.A conditional diffusion model then learns only the refined-scale residual distribution,confining sampling to a stable neighborhood of plausible solutions and suppressing competition among observation-consistent modes.To enhance robustness under varying sensing conditions,we introduce mask-cascade training,which exposes the model to diverse sparse observation patterns through intermediate coarse reconstructions.During inference,manifold-constrained guidance enforces observation consistency as a refinement mechanism rather than a global mode-selection process.", "date": "2025-12-01", "categories": []}
{"id": "2605.13455", "title": "Bayesian In Vivo Tracking of Synapses using Joint Poisson Deconvolution and Diffeomorphic Registration", "abstract": "Synapses are densely packed submicron structures that dynamically reorganize during learning and memory formation. Longitudinal \\textit{in vivo} imaging of fluorescently tagged synaptic receptors offers a promising opportunity to study large-scale synaptic dynamics and how these processes are disrupted in neurological disease. However, in vivo imaging with 2-photon microscopy uses low laser power and therefore suffers from low signal-to-noise ratio (SNR) and high shot noise, nonlinear tissue motion between days, nonstationary fluctuations in synaptic fluorescence, and significant blur induced by the microscope point spread function (PSF). Together, these factors make it challenging to detect and track synapses, especially in regions with high synaptic density. This paper presents a novel template-based framework for modeling synapses as varying luminance point sources that move under a nonlinear tissue deformation. Taking a unified Bayesian approach, we apply this model to microscopy data by deriving a posterior that incorporates a diffeomorphic mapping for domain warping, a Gaussian point spread function for the imaging process, and a Poisson observation model for raw photon counts. The Bayesian solution simultaneously: (1) Constructs a probabilistic template of synapse locations, (2) denoises and deconvolves the image data, (3) infers fluorescence intensities, (4) performs diffeomorphic image registration to correct for tissue motion, and (5) provides confidence regions for these parameter estimates. We demonstrate the framework on both a 2D+t simulated dataset and a 3D+t longitudinal \\textit{in vivo} microscopy dataset of fluorescent synapses imaged in a mouse over two weeks.", "date": "2026-05-13", "categories": []}
{"id": "2605.26127", "title": "Rapid online deep artifact suppression for real-time spiral bSSFP CMR with blipped-CAIPI simultaneous multi-slice imaging at 1.5 T", "abstract": "Purpose: Real-time (RT) bSSFP MRI enables fast free-breathing cardiovascular imaging but requires 10-16 slices for functional assessment, resulting in prolonged scan times. Simultaneous multi-slice (SMS) imaging can reduce acquisition time but when combined with non-Cartesian trajectories, it relies on iterative reconstructions that preclude online use. This study investigates deep artifact suppression to facilitate rapid, online reconstruction of RT-SMS.\n  Methods: A spiral bSSFP SMS RT sequence with two simultaneously acquired slices was implemented at 1.5 T. Reconstruction used slice separation in k-space, followed by deep artifact suppression in image space using a 3D U-Net. Ten healthy volunteers were imaged. RT-SMS image quality and reconstruction time were compared between deep artifact suppression and compressed sensing (CS) reconstructions. Left (LV) and right (RV) ventricular volumes at end diastole (EDV) and end systole (ESV) and LV mass (LVM) were compared between RT-SMS with deep artifact suppression and reference-standard breath-hold (BH) imaging.\n  Results: The RT-SMS acquisition was ~13x faster than BH imaging (15 s vs 3 min 15 s). RT-SMS reconstruction using deep artifact suppression was ~50x faster than CS (30 s vs 24 min 55 s). Deep artifact suppression consistently outperformed CS in quantitative and qualitative image quality (p<0.001). Functional agreement between BH and RT-SMS with deep artifact suppression was good (LVEDV: -7.5 +/- 6.8 ml, LVESV: -0.9 +/- 4.2 ml, RVEDV: -6.4 +/- 8.4 ml, RVESV: 0.2 +/- 10.7 ml, LVM: -10.3 +/- 11.0 g).\n  Conclusion: Online deep artifact suppression reconstruction for RT-SMS bSSFP CMR enables free-breathing short-axis coverage with a substantial reduction in acquisition and reconstruction time while maintaining diagnostic image quality.", "date": "2026-05-18", "categories": []}
{"id": "2605.26470", "title": "Triadic Dynamics Aware Diffusion Posterior Sampling for Inverse Problems: Optimizing Guidance and Stochasticity Schedules", "abstract": "Generative posterior sampling using diffusion models has emerged as a dominant paradigm for solving inverse problems in imaging, which usually consists of three main components: data consistency (DC) guidance, classifier-free guidance (CFG) and stochasticity. While prior arts have focused on how to develop each or all components, less attention has given to how to schedule them, leading to heuristically fixed or partially adjusted suboptimal schedules. In this work, we argue that the interactions among all three components in terms of scheduling are crucial for significantly improved performance in solving inverse problems in imaging. Our analysis shows that aggressive CFG early in sampling conflict with DC guidance, while stochasticity brings the trajectory back to higher-probability regions. Based on these findings, we propose Triadic Dynamics Aware Posterior Sampling (TriPS), which reformulates posterior sampling as a time-varying control problem and optimizes schedules following a triadic trend of decreasing DC and stochasticity scales alongside increasing CFG scale. TriPS achieves this through two strategies: template-based search over functional priors for reliable baseline schedules, and Group Relative Policy Optimization (GRPO)-based reinforcement learning for more flexible temporal curves. Experiments demonstrate TriPS outperforms state-of-the-art baselines in data fidelity and perceptual realism.", "date": "2026-05-26", "categories": []}
{"id": "2605.26729", "title": "Learning Reference-Guided Exposure Correction with Hybrid Illumination Characteristics", "abstract": "We present HICNet, a reference-guided exposure correction framework. A lightweight, content-agnostic encoder distills each image into a compact illumination embedding capturing regional brightness, edge contrast, and higher-order luminance moments. The embedding difference between a source and its reference drives a multi-scale modulation network that combines FiLM-based global adjustment with Photometric Channel Rebalancing for fine-grained, illumination-aware spectral gating, producing exposure-matched outputs while faithfully preserving scene details. A cross-batch contrastive loss orders the illumination manifold, bolstering robustness to diverse lighting conditions. Trained without ground truth or intrinsic decomposition, HICNet attains better accuracy on public benchmarks and generalizes well to entirely unseen scenes.", "date": "2026-05-26", "categories": []}
{"id": "2507.14760", "title": "QUTCC: Quantile Uncertainty Training and Conformal Calibration for Imaging Inverse Problems", "abstract": "While deep learning offers tremendous promise for scientific and medical imaging, any failures and hallucinations (predictions that do not coincide with reality) are hard to pinpoint and can have serious downstream consequences. Uncertainty estimation techniques, such as conformal prediction, can help by predicting statistically valid error bars for a model's prediction. However, popular conformal prediction methods were not designed for high-dimensional image-valued problems and do not take into account spatial correlations within an image during conformal calibration, resulting in larger-than-necessary uncertainty intervals. We propose a practical simultaneous quantile regression method that enables non-linear, spatially-adaptive scaling during conformal calibration. Our method, QUTCC uses a U-Net architecture with a quantile embedding to learn a full conditional quantile distribution during training, and then leverages this non-linear, learned function for spatially-adaptive conformal calibration. At test time, our method can efficiently estimate uncertainty intervals with pixel-marginal coverage guarantees. In addition, QUTCC can also predict pixel-wise conditional probability density estimates without any built-in distributional assumptions. We evaluate our method on several denoising problems, accelerated magnetic resonance imaging, and quantitative phase microscopy. Our method consistently produces tighter uncertainty intervals than prior conformal methods at the same coverage level, can predict plausible conditional distributions for different tasks, and in some cases, high-uncertainty regions can help us locate hallucinations in a model's prediction.", "date": "2025-07-19", "categories": []}
{"id": "2510.07343", "title": "Local MAP Sampling for Diffusion Models", "abstract": "Diffusion Posterior Sampling (DPS) provides a principled Bayesian approach to inverse problems by sampling from $p(x_0 \\mid y)$. While posterior sampling is valuable for capturing uncertainty and multi-modality, many classical and practical inverse problem settings ultimately prioritize accurate point estimation -- most notably the MAP estimator, which has long served as a standard reconstruction objective in imaging and scientific applications. We introduce Local MAP Sampling (LMAPS), a new inference framework that iteratively solves local MAP subproblems along the diffusion trajectory. This perspective clarifies their connection to global MAP and DPS, offering a unified probabilistic interpretation for optimization-based methods. Building on this foundation, we develop practical algorithms with a covariance approximation motivated by a Gaussian prior assumption, and a reformulated objective for stability and interpretability. Across a broad set of image restoration and scientific tasks, LMAPS achieves state-of-the-art performance.", "date": "2025-10-07", "categories": []}
{"id": "2512.13597", "title": "Lighting in Motion: Spatiotemporal HDR Lighting Estimation", "abstract": "We present Lighting in Motion (LiMo), a diffusion-based approach to spatiotemporal lighting estimation. LiMo targets both realistic high-frequency detail prediction and accurate illuminance estimation. To account for both, we propose generating a set of mirrored and diffuse spheres at different exposures, based on their 3D positions in the input. Making use of diffusion priors, we fine-tune powerful existing diffusion models on a large-scale customized dataset of indoor and outdoor scenes, paired with spatiotemporal light probes. For accurate spatial conditioning, we demonstrate that depth alone is insufficient and we introduce a new geometric condition to provide the relative position of the scene to the target 3D position. Finally, we combine diffuse and mirror predictions at different exposures into a single HDRI map leveraging differentiable rendering. We thoroughly evaluate our method and design choices to establish LiMo as state-of-the-art for both spatial control and prediction accuracy.", "date": "2025-12-15", "categories": []}
{"id": "2605.24195", "title": "Single View Seafloor Recovery from Imaging Sonar via Differentiable Rendering", "abstract": "Sonar is often the only modality suitable for high-resolution imaging underwater due to light attenuation and turbidity. Forward-looking imaging sonar provides measurements over range and horizontal angle but collapses vertical structure into a flat image, creating ambiguities that make 3D recovery challenging. A common use case for imaging sonar is underwater terrain mapping (bathymetry), yet current methods require many views, expensive multi-sensor setups, or significant training data, which limits use and adaptability to new environments.\n  We present a training-free method that recovers bathymetry from a single sonar image in under 30 seconds via differentiable rendering, conditioned on a known seafloor tilt. To our knowledge, this is the first differentiable rendering approach for single-view height recovery in sonar. Our method implements differentiable sonar ray tracing and optimizes an explicit height field to reproduce the target image. On synthetic datasets, our approach outperforms a supervised CNN under distribution shift and remains close on rough terrain, while the CNN wins in-distribution. By modeling physically grounded priors of the sonar process, our method adapts across sensor configurations and environments without training data.", "date": "2026-05-22", "categories": []}
{"id": "2605.24959", "title": "Three-Step Conditional Diffusion 3D Reconstruction for Light-Field Microscopy", "abstract": "Light-field microscopy (LFM) enables single-shot capture of multi-angular information from biological samples, supporting real-time volumetric imaging. However, traditional physics-based algorithms often suffer from limited spatial resolution, severe artifacts, and high computational costs. Existing learning-based methods improve inference efficiency but still face limitations in reconstruction accuracy and generalization capability. To address these challenges, this paper proposes a high-fidelity Three-Step Conditional Diffusion (TCD) 3D reconstruction method for LFM. Although conventional diffusion models have achieved remarkable success in generative modeling, their slow sampling process and the inherent trade-off between quality and efficiency hinder their application in real-time 3D imaging. We redesign the diffusion process through a deterministic three-step sampling strategy coupled with a lightweight conditional U-Net, establishing a new paradigm for fast and accurate volumetric reconstruction. Furthermore, an Inter-Class Detection (ICD) module is incorporated to identify out-of-distribution or anomalous inputs during inference, thereby enhancing model stability and reliability. Extensive experiments and cross-dataset evaluations demonstrate that TCD significantly outperforms state-of-the-art methods in both reconstruction fidelity and generalization, providing an efficient and practical 3D reconstruction solution for light-field microscopy.", "date": "2026-05-24", "categories": []}
{"id": "2605.25163", "title": "K-U-KAN: Koopman-Enhanced U-KAN for 3D Dental Reconstruction from a Single Panoramic X-ray Radiograph", "abstract": "A panoramic X-ray compresses a 3D jaw into a 2D strip; we aim to recover the missing depth cleanly and fast. Existing implicit neural representations render realistic volumes but are slow to train, sensitive to sampling and positional encodings, and costly in practice. Pure CNN baselines are efficient yet struggle with the dental arch's long-range geometry, blur fine enamel-dentin boundaries, and offer little interpretability. We present K-U-KAN, a three-stage pipeline that (i) lifts 2D features into depth-aware observables with Kolmogorov-Arnold Networks, (ii) advances these observables by a stable, phase-aware linear evolution via a Koopman token block, and (iii) places the predicted depth bins onto focal-trough rays before a lightweight 3D attention U-KAN refines the volume. This marriage of physics (Beer-Lambert image formation), geometry (horseshoe focal trough), and learned linear dynamics yields sharp anatomy, fewer artifacts, and robust behavior on native radiographic intensities with batch size one. On held-out data, K-U-KAN matches transformer/implicit baselines on signal and structure metrics, clearly improves perceptual quality, and trains in roughly half the time-making single-view PX $\\to$ CBCT reconstruction more practical for clinical pipelines.", "date": "2026-05-24", "categories": []}
{"id": "2605.25589", "title": "Artifact Correction for Echo-Planar Imaging at Low-Field and Ultra-Low-Field MRI", "abstract": "Purpose: Echo-planar imaging (EPI) in low-field (LF) and ultra-low-field MRI (ULF) suffers from severe Nyquist ghost artifacts due to odd-even k-space misalignment. This study develops a reference-free artifact correction pipeline that reduces reliance on conventional reference scans while achieving improved ghost suppression. Methods: Starting from the traditional reference-scan-based ghost artifact correction method, we first introduce a peak-alignment-based ghost artifact correction method to correct odd-even line displacement without reference data. To further reduce residual artifacts, an interpolation-and-resampling strategy is applied. The combined method was evaluated using EPI and diffusion-weighted EPI data in LF and ULF. Results: The proposed pipeline effectively mitigated Nyquist ghosts, improved structural continuity, and enhanced signal uniformity. Peak-alignment-based ghost artifact correction method alone provided comparable artifact suppression to reference-scan-based ghost artifact correction method, while interpolation and resampling further suppressed residual artifacts, enabling reliable visualization of brain structures under ULF conditions. Conclusion: A practical, reference-free correction pipeline is presented for LF and ULF EPI, combining peak-alignment-based ghost artifact correction method and interpolation-resampling to achieve efficient ghost suppression and expand the clinical applicability of low-field MRI systems, providing both theoretical guidance and practical experience for ULF EPI-based DWI imaging.", "date": "2026-05-25", "categories": []}
{"id": "2605.25892", "title": "SP-MoMamba: Superpixel-driven Mixture of State Space Experts for Efficient Image Super-Resolution", "abstract": "State space models (SSMs) have emerged as a powerful paradigm for efficient single-image super-resolution (SR) due to their linear complexity and long-range modeling capabilities. However, existing Mamba-based methods typically rely on data-agnostic rigid scanning, which reshapes 2D images into 1D sequences over a fixed grid, inevitably disrupting spatial-semantic topology and introducing artifacts. Inspired by the \\textbf{Gestalt perceptual grouping theory}, we propose \\textbf{SP-MoMamba}, a superpixel-driven mixture of state space experts designed for content-aware SR. Our core idea is to transform the traditional rigid scanning into a \\textbf{semantic-level interaction} by treating superpixels as fundamental units. Specifically, we introduce the \\textbf{Superpixel-driven State Space Model (SP-SSM)}, which compresses semantically homogeneous regions into high-order tokens to preserve global topological consistency. To address the conflict between fixed scanning scales and diverse semantic granularities, we develop the \\textbf{Multi-Scale Superpixel Mixture of State Space Experts (MSS-MoE)}. This module utilizes a dynamic routing mechanism to adaptively assign scale-specific experts, effectively capturing multi-scale textures while reducing computational redundancy. Furthermore, to prevent the loss of high-frequency details during global abstraction, we introduce a \\textbf{Local Spatial Modulation Expert (LSME)} to complement the global modeling, ensuring a precise reconstruction of sharp edges and fine structures. Extensive experiments on standard benchmarks demonstrate that SP-MoMamba achieves superior reconstruction fidelity and a more favorable efficiency-performance trade-off compared to state-of-the-art efficient SR methods.", "date": "2026-05-25", "categories": []}
{"id": "2605.22423", "title": "Moment-Reenacting: Inverse Motion Degradation with Cross-shutter Guidance", "abstract": "Motion degradation, manifested as blur in global shutter (GS) images or rolling shutter (RS) distortion in RS counterparts, remains a fundamental challenge in computational imaging, especially under fast motion or low-light conditions. While prior works have treated blur decomposition and RS temporal super-resolution as separate tasks, this separation fails to exploit their intrinsic complementarity. In this paper, we propose a unified framework to invert motion degradation and reenact imaging moment by jointly leveraging the complementary characteristics of GS blur and RS distortion. To this end, we introduce a novel dual-shutter setup that captures synchronized blur-RS image pairs and demonstrate that this combination effectively resolves temporal and spatial ambiguities inherent in both modalities. For allowing flexible performance-cost trade-offs, we further extend this dual-shutter setup to a stereo Blur-RS configuration with a narrow baseline. In addition, we construct a triaxial imaging system to collect a real-world dataset with aligned GS-RS pairs and ground-truth high-speed frames, enabling robust training and evaluation beyond synthetic data. Our proposed network explicitly disentangles motion into context-aware and temporally-sensitive representations via a dual-stream motion interpretation module, followed by a self-prompted frame reconstruction stage. Extensive experiments validate the superiority and generalizability of our approach, establishing a new paradigm for realistic high-speed video reconstruction under complex motion degradations. Codes and more resources are available at https://jixiang2016.github.io/dualBR_site/.", "date": "2026-05-21", "categories": []}
{"id": "2605.23264", "title": "Coloring the Noise: Adversarial Sobolev Alignment for Faithful Image Super Resolution", "abstract": "Generative priors in Image Super-Resolution (SR) often compromise faithful restoration, we attribute this limitation to a fundamental spectral misalignment between isotropic objectives and the intrinsic natural image manifold. While Direct Preference Optimization offers a path to alignment, its reliance on spectrally flat Gaussian noise fails to distinguish authentic high-frequency details from hallucinations. To bridge this geometric gap, we propose ASASR, a theoretically grounded framework that recasts the generative flow into a Sobolev-induced Riemannian geometry by explicitly coloring the noise transition kernel to mirror natural spectral decay. Driving this geometric alignment, we integrate a parametric adversary grounded in the Riesz Representation Theorem, which synthesizes targeted negative samples equivalent to worst-case Sobolev gradients to direct optimization along the tangent space of plausible structural failures. Extensive evaluations demonstrate that ASASR outperforms leading generative baselines, particularly in preserving spectral consistency and structural fidelity, offering a robust solution that effectively mitigates artifacts.", "date": "2026-05-22", "categories": []}
{"id": "2605.23282", "title": "Discontinuous Galerkin Neural Operator for Pathology Defocus Deblurring", "abstract": "Defocus deblurring in pathological microscopy remains challenging due to the spatially varying and locally discontinuous nature of optical blur induced by a position-dependent integral imaging process.\n  Existing deep learning methods, constrained by shift-invariance assumptions and limited interpretability, are not well suited to such heterogeneous blur patterns.\n  Neural operators provide a principled alternative by modeling defocus formation directly as an integral operator, offering a new perspective on defocus deblurring.\n  However, most existing neural operator architectures for low-level vision rely on globally parameterized kernels that assume smoothness and stationarity, limiting their ability to model heterogeneous and locally discontinuous blur patterns.\n  To address this limitation, we propose the Discontinuous Galerkin Neural Operator (DGNO), which parameterizes the integral kernel using a discontinuous Galerkin formulation with element-local volume operators and interface numerical fluxes.\n  DGNO provides a principled combination of locality,\n  heterogeneity modeling, and global coherence while preserving the underlying physics\n  of optical image formation.\n  Extensive and insightful experiments demonstrate that DGNO surpasses state-of-the-arts, delivering sharper reconstructions, robust handling of spatially varying blur, and scalable high-resolution performance. The code will be released at https://github.com/DeepMed-Lab-ECNU/Single-Image-Deblur.", "date": "2026-05-22", "categories": []}
{"id": "2605.23451", "title": "Efficient One-Step Diffusion Restoration Model with Compact Token Compression and Linear Attention", "abstract": "Real-world image super-resolution aims to recover high-quality images from complex and unknown real-world degradations. However, existing generative Real-ISR methods largely inherit the dense latent representations and quadratic-cost global modeling paradigm developed for high-resolution image synthesis, causing computation, memory usage, and inference latency to scale unfavorably with resolution and thus limiting practical deployment. We argue that the key bottleneck lies not in insufficient restoration priors, but in excessive token redundancy and costly token interactions during high-resolution restoration. Motivated by this observation, we revisit Real-ISR from the perspectives of compact latent representation and linear-complexity modeling, and propose SANA-SR, an efficient one-step restoration framework. Specifically, SANA-SR employs a deep compression autoencoder with a 32x compression ratio to drastically reduce latent tokens while preserving restoration-relevant structures and textures. On top of this compact latent space, we introduce a linear-attention DiT with LoRA fine-tuning, enabling efficient high-resolution restoration with linear-complexity token mixing. Extensive experiments on all benchmark datasets demonstrate that SANA-SR achieves highly competitive and often superior quantitative performance against existing methods, while restoring clearer and more realistic textures. Moreover, after pruning, the deployed model runs in 0.019s with 407.95G MACs and 344M parameters, highlighting its strong potential for practical mobile deployment.", "date": "2026-05-22", "categories": []}
{"id": "2507.13339", "title": "SpectraLift: Physics-Guided Spectral-Inversion Network for Self-Supervised Hyperspectral Image Super-Resolution", "abstract": "High-spatial-resolution hyperspectral images (HSI) are essential for applications such as remote sensing and medical imaging, yet HSI sensors inherently trade spatial detail for spectral richness. Fusing high-spatial-resolution multispectral images (HR-MSI) with low-spatial-resolution hyperspectral images (LR-HSI) is a promising route to recover fine spatial structures without sacrificing spectral fidelity. Most state-of-the-art methods for HSI-MSI fusion demand point spread function (PSF) calibration or ground truth high resolution HSI (HR-HSI), both of which are impractical to obtain in real world settings. We present SpectraLift, a fully self-supervised framework that fuses LR-HSI and HR-MSI inputs using only the MSI's Spectral Response Function (SRF). SpectraLift trains a lightweight per-pixel multi-layer perceptron (MLP) network using ($i$)~a synthetic low-spatial-resolution multispectral image (LR-MSI) obtained by applying the SRF to the LR-HSI as input, ($ii$)~the LR-HSI as the output, and ($iii$)~an $\\ell_1$ spectral reconstruction loss between the estimated and true LR-HSI as the optimization objective. At inference, SpectraLift uses the trained network to map the HR-MSI pixel-wise into a HR-HSI estimate. SpectraLift converges in minutes, is agnostic to spatial blur and resolution, and outperforms state-of-the-art methods on PSNR, SAM, SSIM, and RMSE benchmarks.", "date": "2025-07-17", "categories": []}
{"id": "2510.20814", "title": "SpectraMorph: Structured Latent Learning for Self-Supervised Hyperspectral Super-Resolution", "abstract": "Hyperspectral sensors capture dense spectra per pixel but suffer from low spatial resolution, causing blurred boundaries and mixed-pixel effects. Co-registered companion sensors such as multispectral, RGB, or panchromatic cameras provide high-resolution spatial detail, motivating hyperspectral super-resolution through the fusion of hyperspectral and multispectral images (HSI-MSI). Existing deep learning based methods achieve strong performance but rely on opaque regressors that lack interpretability and often fail when the MSI has very few bands. We propose SpectraMorph, a physics-guided self-supervised fusion framework with a structured latent space. Instead of direct regression, SpectraMorph enforces an unmixing bottleneck: endmember signatures are extracted from the low-resolution HSI, and a compact multilayer perceptron predicts abundance-like maps from the MSI. Spectra are reconstructed by linear mixing, with training performed in a self-supervised manner via the MSI sensor's spectral response function. SpectraMorph produces interpretable intermediates, trains in under a minute, and remains robust even with a single-band (pan-chromatic) MSI. Experiments on synthetic and real-world datasets show SpectraMorph consistently outperforming state-of-the-art unsupervised/self-supervised baselines while remaining very competitive against supervised baselines.", "date": "2025-10-23", "categories": []}
{"id": "2603.21717", "title": "Uncertainty-Aware Distribution-to-Distribution Flow Matching for Scientific Imaging", "abstract": "Distribution-to-distribution generative models support scientific imaging tasks ranging from modeling cellular perturbation responses to translating medical images across conditions. Trustworthy generation requires reliability, or generalization across labs, devices, and experimental conditions, and accountability, or detecting out-of-distribution cases where predictions may be unreliable. We leverage Stochastic Flow Matching (SFM), a marginal-preserving stochastic extension of flow matching for improved generalization under distribution shift. SFM augments deterministic flows with a diffusion term together with a learned score-based drift correction, retaining the learned transport marginals while modeling conditional variability. Building on this SFM framework, we introduce Bayesian Stochastic Flow Matching (BSFM) as a companion uncertainty quantification mechanism and develop AVUQ (Antithetic Variance-reduction Uncertainty Quantification) to approximately estimate epistemic and aleatoric uncertainty via sample-efficient antithetic sampling with approximate posterior inference. We further use AVUQ to yield anomaly scores for unreliable generation detection. Experiments on cellular imaging (BBBC021, JUMP) and brain fMRI (Theory of Mind) across diverse unseen scenarios show that SFM improves generalization while AVUQ provides effective uncertainty-based anomaly scores under practical sampling budgets.", "date": "2026-03-23", "categories": []}
{"id": "2605.21671", "title": "HyperBench: Standardizing and Scaling Synthetic Evaluation for Hyperspectral Super-Resolution", "abstract": "Hyperspectral super-resolution (HSR) reconstructs a high-spatial-resolution hyperspectral image by fusing a low-resolution hyperspectral image (LR-HSI) with a high-resolution multispectral image (HR-MSI). In the absence of real-world paired data, HSR methods are evaluated almost exclusively on synthetic experiments derived from hyperspectral datasets through Wald's protocol. Despite the protocol's widespread adoption, its practical implementation varies markedly across research works, typically relying on a single (usually Gaussian) or very few point spread functions (PSFs), one or two spectral response functions (SRFs), and a couple of spatial downsampling factors. As a result, reported performance figures are difficult to compare across the literature, in addition to being often difficult to reproduce; furthermore, they may not generalize across realistic sensing conditions. We introduce HyperBench, a unified and extensible framework that standardizes synthetic experimentation for HSR. HyperBench supports diverse degradation configurations spanning ten PSFs, four SRFs derived from operational multispectral sensors, configurable spatial downsampling factors, and matched additive white Gaussian noise; its goal is to automate large-scale evaluation and structured logging. By decoupling model development from experimental design, the framework enables reproducible, apples-to-apples cross-method comparison with minimal friction. We use HyperBench to evaluate six recently proposed HSR methods across a 70-configuration sweep on four widely used hyperspectral scenes and observe that the inter-method PSNR spread widens from approximately 5 dB on the easiest PSF to over 13 dB on the hardest - a fragility that is structurally invisible to the prevailing single-configuration evaluation protocol. HyperBench code is available at https://github.com/ritikgshah/HyperBench .", "date": "2026-05-20", "categories": []}
{"id": "2605.21913", "title": "Multi-scale interaction network for stereo image super-resolution", "abstract": "Stereo image super-resolution aims to generate high-resolution images by leveraging complementary information from binocular systems. Although previous studies have achieved impressive results, the potential of intra-view and cross-view information has not been fully exploited. To address this issue, we propose a novel multi-scale interaction network for stereo image super-resolution. Specifically, we design a Multi-scale Spatial-Channel Attention Module that utilizes multi-scale large separable kernel attention and simple channel attention to improve intra-view feature extraction. Additionally, we propose a Dual-View Epipolar Attention Module, utilizing an optimal transport algorithm to achieve more accurate matching along the epipolar line. Extensive experimental and ablation studies show that our method achieves competitive results that outperform most SOTA methods.", "date": "2026-05-21", "categories": []}
{"id": "2605.22104", "title": "OPERA: An Agent for Image Restoration with End-to-End Joint Planning-Execution Optimization", "abstract": "Real-world image restoration is challenging due to complex and interacting mixed degradations. Recent agent-based approaches address this problem by composing multiple task-specific restoration tools. However, empirical analysis reveals that their performance is fundamentally limited by implicitly constrained planning spaces and the lack of coordination among independently pretrained tools. To address these issues, we propose OPERA (Optimized Planning-Execution Restoration Agent), a framework that jointly optimizes restoration planning and tool execution in an end-to-end manner. On the planning side, OPERA uses reinforcement learning to directly optimize tool composition over a combinatorial plan space, with the final restoration quality as the reward. On the execution side, OPERA introduces agent-guided co-training of restoration tools, enabling them to learn cooperative behaviors under sequential composition. Extensive experiments on multi-degradation benchmarks and real-world datasets demonstrate that OPERA consistently outperforms both all-in-one restoration models and existing agent-based methods across diverse and complex degradation scenarios.", "date": "2026-05-21", "categories": []}
{"id": "2605.22121", "title": "MotionDPS: Motion-Compensated 3D Brain MRI Reconstruction", "abstract": "Magnetic resonance imaging (MRI) is highly susceptible to patient motion due to its relatively long acquisition times and the fact that data are acquired sequentially in k-space. Even small patient movements introduce phase inconsistencies across measurements, leading to severe artifacts such as blurring, ghosting, and geometric distortions that can compromise diagnostic quality. Retrospective motion compensation remains challenging, particularly in accelerated acquisitions, due to the ill-posed nature of the joint reconstruction and motion estimation problem. In this work, we propose a unified Bayesian framework for motion-compensated 3D MRI that jointly estimates the anatomical image, rigid-body motion parameters, and coil sensitivity maps directly from motion-corrupted k-space data. Our approach integrates pretrained 3D complex-valued score-based diffusion models as expressive anatomical image priors within a physics-based forward model. Inference is performed by alternating diffusion posterior image updates with efficient proximal optimization steps for motion and coil sensitivity estimation, enabling fully unsupervised reconstruction without the need for paired motion-free training data. Experiments on simulated and real-motion brain MRI datasets demonstrate that the proposed method achieves improved image quality and motion robustness compared to state-of-the-art classical and learning-based motion correction techniques, particularly in the presence of severe motion and high acceleration.", "date": "2026-05-21", "categories": []}
{"id": "2605.22186", "title": "Event-Illumination Collaborative Low-light Image Enhancement with a High-resolution Real-world Dataset", "abstract": "Event-based low-light image enhancement (LIE) methods mainly focus on incorporating high dynamic range (HDR) information from events while overlooking the essential global illumination in images and the inherent noise sensitivity of event signals in real-world scenarios. To address these issues, we propose EIC-LIE, an event-illumination collaborative LIE framework. Concretely, we first design an Event-Illumination Collaborative Interaction (EICI) module, which contains two key processes: forward gathering, which gathers HDR features across varying lighting conditions, and backward injection, which provides complementary content for illumination and event representations. Next, we introduce an Illumination-aware Event Filter (IAEF) that dynamically reduces event noise based on brightness statistics derived from images. Additionally, we build a beam-splitter-based hybrid imaging system to collect high-quality event-image pairs with temporal synchronization from dynamic scenes, providing the first high-resolution, real-world event-based LIE dataset. Extensive experiments show that our EIC-LIE outperforms state-of-the-art methods on five real-world and synthetic datasets, significantly surpassing previous methods with improvements of up to 1.24dB in PSNR and 0.069 in SSIM. The code and dataset are released at https://github.com/QUEAHREN/EIC-LIE.", "date": "2026-05-21", "categories": []}
{"id": "2605.20624", "title": "Accelerating Video Inverse Problem Solvers with Autoregressive Diffusion Models", "abstract": "Diffusion models provide powerful priors for zero-shot video inverse problems, but their real-time deployment is hindered by two inefficiencies: high initial latency caused by holistic video restoration, and low throughput resulting from multiple VAE passes to enforce measurement consistency in pixel space. To overcome these limitations, we propose Autoregressive Video Inverse problem Solver (AVIS). The AVIS framework leverages autoregressive video diffusion models to restore videos in a streaming manner, naturally eliminating latency bottlenecks. Specifically, AVIS initializes reverse diffusion with a measurement-consistent estimate, reducing the required sampling steps. Compared to leading non-autoregressive solvers, AVIS drastically reduces initial latency from 114s to 4s and increases throughput from 0.71 to 1.18 FPS while achieving superior restoration quality. We further introduce a highly accelerated variant, dubbed AVIS Flash, that enforces measurement consistency solely on the first chunk. AVIS Flash substantially boosts throughput to 5.91 FPS on a single RTX 4090 GPU while maintaining competitive performance and achieving a favorable efficiency-performance trade-off, paving the way toward real-time deployment.", "date": "2026-05-20", "categories": []}
{"id": "2605.20687", "title": "Motion-Robust Deep Reconstruction for Free-Breathing Cardiac Cine MRI", "abstract": "Conventional cardiac cine MRI relies on breath-hold Cartesian acquisitions, which are vulnerable to motion artifacts and can be uncomfortable or infeasible, particularly for pediatric and other noncompliant patients who cannot reliably hold their breath. Free-breathing radial acquisitions can alleviate these limitations, but robust reconstruction at high acceleration remains challenging due to prominent streak artifacts. To address these limitations, we propose Cine-DL, a clinically oriented framework that couples targeted k-space preprocessing with fast, model-based deep reconstruction. In this pipeline, raw free-breathing radial data undergo retrospective cardiac binning and respiratory gating to resolve cardiac phases and discard motion-corrupted spokes. We then introduce Streak Optimized Coil Compression (SOC), which explicitly preserves cardiac signals while suppressing peripheral interference that typically drives the streak artifacts. The resulting 2D+t cine series is reconstructed with an unrolled network that alternates a ResNet proximal operator with physics-based data consistency updates solved via conjugate gradient. We further employ a memory-efficient training strategy that reduces peak memory usage. We evaluate Cine-DL on free-breathing volunteer data against established baselines (k-t SENSE and iGRASP) and demonstrate clinical translation via hospital deployment on newly acquired patient data. Our experiments show that Cine-DL consistently improves quantitative metrics and visual fidelity, supporting a practical route toward routine, time-sensitive clinical adoption of free-breathing cine MRI.", "date": "2026-05-20", "categories": []}
{"id": "2605.20820", "title": "AIR: Amortized Image Reconstruction Framework for Self-Supervised Feed-Forward 2D Gaussian Splatting", "abstract": "2D Gaussian splatting provides an efficient explicit representation for image reconstruction, but existing methods still require costly per-image iterative optimization or rely on handcrafted priors for primitive allocation. We present AIR, a self-supervised feed-forward framework that amortizes iterative Gaussian fitting into a single network pass, eliminating per-image test-time optimization. AIR adopts a stage-wise residual architecture that progressively predicts additional Gaussian primitives from reconstruction residuals, together with an explicit Stage Control mechanism that activates new primitives only in under-reconstructed regions. A Predict--Optimize--Distill training strategy stabilizes multi-stage prediction by distilling short-horizon optimized Gaussian increments back into the predictor. The stabilized predictor is then jointly finetuned across stages and equipped with an image-adaptive quantizer for compact Gaussian storage. Experiments on Kodak and DIV2K show that AIR achieves better reconstruction quality than representative Gaussian-based baselines while reducing encoding time to 160--300\\,ms. Code: https://github.com/whoiszzj/AIR.git", "date": "2026-05-20", "categories": []}
{"id": "2605.21094", "title": "UOTIP: Unbalanced Optimal Transport Map for Unpaired Inverse Problems", "abstract": "We investigate unpaired image inverse problems, a challenging setting where only independent, non-paired sets of noisy measurements and clean target signals are available for training. We propose a novel inverse problem solver based on Unbalanced Optimal Transport, called Unbalanced Optimal Transport Map for Inverse Problems (UOTIP). Our method formulates the reconstruction task, predicting clean target signals from noisy measurements, as learning a UOT Map from noisy measurement distribution to clean signal distribution by incorporating a likelihood-based cost function. By relaxing the exact marginal constraint, the UOT framework provides key advantages to our model: robustness to multi-level observation noise, adaptability to class imbalance between noisy and clean datasets, and generalizability to diverse noise-type scenarios. Furthermore, we theoretically demonstrate that incorporating a quadratic cost term ensures the existence and uniqueness of the transport map by satisfying the twist condition, even for ill-posed inverse problems. Our experiments demonstrate that UOTIP achieves state-of-the-art performance on unpaired image inverse problem benchmarks, across linear and nonlinear inverse problems.", "date": "2026-05-20", "categories": []}
{"id": "2605.17470", "title": "EchoSR: Efficient Context Harnessing for Lightweight Image Super-Resolution", "abstract": "Image super-resolution (SR) aims to reconstruct high-quality, high-resolution (HR) images from low-resolution (LR) inputs and plays a critical role in various downstream applications. Despite recent advancements, balancing reconstruction fidelity and computational efficiency remains a fundamental challenge, particularly in resource-constrained scenarios. While existing lightweight methods attempt to expand receptive fields, many of them either incur substantial computational overhead, naively scale up kernel sizes, or lack mechanisms for coherent multi-scale integration, limiting their overall effectiveness and scalability. To address these limitations, we propose EchoSR, an efficient context-harnessing framework for lightweight image super-resolution, which unifies multi-scale receptive field modeling and hierarchical context fusion. EchoSR decouples feature learning into disentangled local, multi-scale, and global modeling stages through an efficient context-harnessing strategy, and further promotes seamless cross-scale integration via a cross-scale overlapping fusion mechanism. Extensive experiments have shown that EchoSR consistently outperforms state-of-the-art lightweight super-resolution methods across multiple benchmarks, while also achieving a faster speed $(\\sim 2\\times)$. The source code is available at https://github.com/funnyWang-Echoes/EchoSR.", "date": "2026-05-17", "categories": []}
{"id": "2510.04382", "title": "Adaptive double-phase Rudin--Osher--Fatemi denoising model", "abstract": "Even though more than 30 years have passed since the seminal Rudin--Osher--Fatemi (ROF) paper on total variation (TV) denoising, it remains relevant, in particular in scientific applications such as astronomical imaging. However, it is known to suffer from artifacts such as the staircasing effect. Many variants of the model have been proposed with the aim of countering this. Recently, against the backdrop of immense research output on double-phase problems in the mathematical analysis community, a double-phase type integral functional, comprising of TV and a weighted term of quadratic growth, was suggested as a regularizer for image restoration.\n  Here, we propose an adaptive variant of the ROF denoising model based on that regularizer. It is designed to reduce staircasing with respect to the classical ROF model, while preserving the edges of the image in a similar fashion. We implement the model and test its performance on synthetic and natural images over a range of noise levels. Compared to {established} models {with similar interpretability to ROF}, we observe an improved or similar performance in terms of similarity metrics SSIM, PSNR, {and LPIPS}, while the staircasing effect is visibly reduced.", "date": "2025-10-05", "categories": []}
{"id": "2605.10185", "title": "DynGhost: Temporally-Modelled Transformer for Dynamic Ghost Imaging with Quantum Detectors", "abstract": "Ghost imaging reconstructs spatial information from a single-pixel bucket detector by correlating structured illumination patterns with scalar intensity measurements. While deep learning approaches have achieved promising results on static scenes, two critical limitations remain unaddressed: existing architectures fail to exploit temporal coherence across frames, leaving dynamic ghost imaging largely unsolved, and they assume additive Gaussian noise models that do not reflect the true Poissonian statistics of real single-photon hardware. We present DynGhost (Dynamic Ghost Imaging Transformer), a transformer architecture that addresses both limitations through alternating spatial and temporal attention blocks. Our quantum-aware training framework, based on physically accurate detector simulations (SNSPDs, SPADs, SiPMs) and Anscombe variance-stabilizing normalization, resolves the distribution shift that causes classical models to fail under realistic hardware constraints. Experiments across multiple benchmarks demonstrate that DynGhost outperforms both traditional reconstruction methods and existing deep learning architectures, with particular gains in dynamic and photon-starved settings.", "date": "2026-05-11", "categories": []}
{"id": "2605.16396", "title": "Beyond MMSE: Enhancing PnP Restoration with ProxiMAP", "abstract": "Plug-and-Play (PnP) methods have become standard tools for solving imaging inverse problems by replacing the intractable maximum a posteriori (MAP) denoiser with the MMSE one. While this mismatch has been widely treated as unavoidable, recent works have sought to close this gap by targeting the MAP with diffusion-model scores. We show this is problematic in practice: learned scores do not match the true ones, so MAP-targeting iterations converge to cartoon-like images rather than realistic ones, and better results are obtained by stopping short of convergence. We turn this observation into a design principle and introduce ProxiMAP, an iterative MAP approximation whose noise schedule keeps the iterate's residual noise matched to the denoiser's training noise. This keeps the denoiser in-distribution where its score is reliable, and yields implicit early stopping that avoids the failure mode above. ProxiMAP is a modular drop-in replacement for MMSE denoisers in standard PnP algorithms and consistently sharpens reconstructions across deblurring, inpainting, super-resolution, and phase retrieval. Building on the same principle, we propose a hybrid variant that applies ProxiMAP only in the late iterations of PnP, where the denoiser is most reliable -- matching or exceeding the full-replacement variant at a fraction of the cost.", "date": "2026-05-12", "categories": []}
{"id": "2605.16925", "title": "P2GS: Physical Prior-guided Gaussian Splatting for Photometrically Consistent Urban Reconstruction", "abstract": "3D Gaussian Splatting (3DGS) has recently emerged as a powerful explicit representation enabling fast, high-fidelity rendering, making it a promising foundation for closed-loop simulators and perception models in autonomous driving. However, conventional 3DGS implicitly assumes consistent exposure and tone mapping across views. Real driving data violates this assumption due to heterogeneous camera pipelines and dynamic outdoor illumination, baking exposure discrepancies and sensor noise into the radiance field and producing artifacts and inconsistent illumination especially in static backgrounds crucial for realistic simulation. These issues are amplified in autonomous driving, where sparse viewpoints, varying exposures, and outdoor lighting interact, while prior work mainly targets dynamic-object reconstruction and overlooks cross-view photometric consistency. To address this limitation, we introduce P2GS, a physically consistent Gaussian Splatting framework that jointly decomposes a view-invariant linear HDR radiance field, per-view exposure scales, and tone-mapping functions from only LDR images without HDR supervision. P2GS employs a unified optimization strategy grounded in the physical image-formation process, enforcing relative-exposure consistency and HDR-domain radiance regularization. This yields a radiance field robust to inter-camera illumination differences while preserving the real-time efficiency of standard 3DGS. Experiments across real and simulated driving environments show that P2GS matches or surpasses prior methods in LDR reconstruction while providing substantially improved photometric consistency, reliable exposure normalization, and physically coherent illumination across diverse scenes.", "date": "2026-05-16", "categories": []}
{"id": "2605.16967", "title": "Expandable, Compressible, Mineable: Open-World Thermal Image Restoration", "abstract": "In open-world settings, thermal infrared (TIR) image degradations continuously emerge and evolve, while most existing all-in-one restoration methods are built on a closed-set assumption and struggle to continually adapt to novel degradations. To address this, we propose ECMRNet, an Expandable, Compressible, and Mineable Restoration Network for open-world TIR restoration from a continual learning perspective. Conceptually, ECMRNet unifies continual degradation learning as an \"expand-compress-mine\" closed-loop process, enabling sustained adaptation to new degradations with controllable evolution. Structurally, ECMRNet decomposes intermediate representations into group-isolated subspaces, and achieves strict parameter isolation and fast adaptation to new degradations by freezing historical groups and isomorphically expanding new ones. To curb model growth as tasks accumulate, we present Structural Entropy Pruning, which identifies and removes redundant channel groups via two-dimensional structural entropy minimization, achieving information contribution-driven adaptive compression. Moreover, we design a Sub-degradation Knowledge Mining Module that dynamically retrieves and recombines transferable components from historical representations to improve restoration under compound degradations. Experimental results demonstrate that ECMRNet achieves superior overall performance across diverse single and compound degradations while using fewer parameters and lower computational cost. The source code is available at https://github.com/Kust-lp/ECMRNet.", "date": "2026-05-16", "categories": []}
{"id": "2605.17506", "title": "Degradation Frequency Curve: An Explicit Frequency-Quantified Representation for All-in-One Image Restoration", "abstract": "A fundamental difficulty in all-in-one blind image restoration is that degradation is usually treated as an implicit factor hidden in degraded-to-clean mapping, rather than as an explicit object that can be measured and manipulated. This limitation becomes more pronounced under mixed, compound, or unseen degradation conditions, where degradation effects are hard to assign to predefined labels or task-specific parameters. We propose the Degradation Frequency Curve (DFC), a structured spectral representation that quantifies degradation responses by measuring band-wise residual-to-degraded energy ratios in the frequency domain. DFC converts visually entangled and hard-to-describe degradation effects into a measurable degradation coordinate space. Moreover, DFC can be adaptively decomposed into band-wise spectral tokens, allowing local degradation responses to be represented as reusable restoration priors. Based on this representation, we develop the DFC-guided Image Restorer (DFC-IR), a token-conditioned multi-scale framework that progressively estimates DFCs from intermediate restorations and uses the resulting spectral tokens to guide degradation-aware restoration in a coarse-to-fine manner. Extensive experiments on standard, composite, unseen, and real-world degradation benchmarks show that DFC provides an effective representation basis for all-in-one restoration, leading to state-of-the-art performance and improved generalization under complex degradation profiles.", "date": "2026-05-17", "categories": []}
{"id": "2605.17865", "title": "Imaging Hidden Objects with Consumer LiDAR via Motion Induced Sampling", "abstract": "LiDARs are being increasingly deployed for consumer imaging in handheld, wearable, and robotic applications. These sensors can capture the time-of-flight of light at picosecond resolution, which in principle, enables them to capture information about objects hidden from their field of view. While such non-line-of-sight (NLOS) imaging capabilities have been shown on research-grade LiDARs, they are challenging to achieve on consumer devices due to poor signal quality resulting from low laser power, low spatial resolution, and object and camera motion. Inspired by burst photography and synthetic aperture radar, we propose a multi-frame fusion strategy to overcome these challenges and demonstrate NLOS imaging on consumer LiDAR. We first introduce the motion-induced aperture sampling model to unify the effects of object shape, object motion, and camera motion under a single measurement model. Using this model, we demonstrate several NLOS capabilities on a smartphone-grade LiDAR: (1) 3D reconstruction, (2) single and multi-object tracking, and (3) camera localization using hidden objects. Previously, NLOS imaging capabilities were largely restricted to bulky and expensive research-grade hardware that requires extensive setup and calibration. Our results represent a shift towards plug-and-play NLOS imaging, where anyone can image hidden objects with off-the-shelf hardware ($<100) and no additional setup. We believe that democratization of such capabilities will advance consumer applications of NLOS imaging.", "date": "2026-05-18", "categories": []}
{"id": "2605.15456", "title": "DIPA: Distilled Preconditioned Algorithms for Solving Imaging Inverse Problems", "abstract": "Solving imaging inverse problems has usually been addressed by designing proper prior models of the underlying signal. However, minimizing the data fidelity term poses significant challenges due to the ill-conditioned sensing matrix caused by physical constraints in the acquisition system. Thus, preconditioning techniques have been adopted in classical optimization theory to address ill-conditioned data-fidelity minimization by transforming the algorithm gradient step to achieve faster convergence and better numerical stability. We extend the preconditioning concept beyond convergence acceleration and use it to improve reconstruction quality. We introduce DIPA: Distilled Preconditioned Algorithms, where a preconditioning operator (PO) is optimized using teacher-guided distillation criteria. Unlike standard model-compression KD, the teacher and student differ by the sensing operators available during reconstruction: the teacher uses a simulated, better-conditioned, and more informative sensing matrix, whereas the student uses the physically feasible sensing matrix. We design different distillation loss functions to transfer different properties of the teacher algorithm to the preconditioned student. The PO can be linear (L-DIPA), allowing interpretability, or non-linear (N-DIPA), parametrized by a neural network, offering better scalability. We validate the proposed PO design across several imaging modalities, including magnetic resonance imaging, compressed sensing, and super-resolution imaging.", "date": "2026-05-14", "categories": []}
{"id": "2601.16981", "title": "SyncLight: Single-Edit Multi-View Relighting", "abstract": "We present SyncLight, a method to enable consistent, parametric control over light sources across multiple uncalibrated views of a static scene conditioned on a single view. While single-view relighting has advanced significantly, existing generative approaches struggle to maintain the rigorous lighting consistency essential for multi-camera broadcasts, stereoscopic cinema, and virtual production. SyncLight addresses this by enabling precise control over light intensity and color across a multi-view capture of a scene, conditioned on a single reference edit. Our method leverages a multi-view diffusion transformer trained using a latent bridge matching formulation, achieving high-fidelity relighting of the entire image set in a single inference step. To facilitate training, we introduce a large-scale hybrid dataset comprising diverse synthetic environments -- curated from existing sources and newly designed scenes -- alongside high-fidelity, real-world multi-view captures under calibrated illumination. Though trained only on image pairs, SyncLight generalizes zero-shot to an arbitrary number of viewpoints, effectively propagating lighting changes across all views, without requiring camera pose information. SyncLight enables practical relighting workflows for multi-view capture systems.", "date": "2026-01-23", "categories": []}
{"id": "2602.07029", "title": "Guidestar-Free Adaptive Optics with Asymmetric Apertures", "abstract": "This work introduces the first closed-loop adaptive optics (AO) system capable of optically correcting aberrations in real-time without a guidestar or a wavefront sensor. Nearly 40 years ago, Cederquist et al. demonstrated that asymmetric apertures enable phase retrieval (PR) algorithms to perform fully computational wavefront sensing, albeit at a high computational cost. More recently, Chimitt et al. extended this approach with machine learning and demonstrated real-time wavefront sensing using only a single (guidestar-based) point-spread-function (PSF) measurement. Inspired by these works, we introduce a guidestar-free AO framework built around asymmetric apertures and machine learning. Our approach combines three key elements: (1) an asymmetric aperture placed at the system's pupil plane that enables PR-based wavefront sensing, (2) a pair of machine learning algorithms that estimate the PSF from natural scene measurements and reconstruct phase aberrations, and (3) a spatial light modulator that performs optical correction. We experimentally validate this framework on dense natural scenes imaged through unknown obscurants. Our method outperforms state-of-the-art guidestar-free wavefront shaping methods, using an order of magnitude fewer measurements and three orders of magnitude less computation.", "date": "2026-02-02", "categories": []}
{"id": "2605.09020", "title": "The Direct Integration Theorem: A Rigorous Framework for Consistent Discrete Solutions of the Inverse Radon Problem", "abstract": "This paper presents a novel Direct Integration Theorem (DIT), derived as a non-trivial corollary of the classical Central Slice Theorem (CST). The DIT provides a mathematically consistent transition from the continuous to the discrete domain - a fundamental challenge in computed tomography - thereby eliminating the need for frequency-domain interpolation without resorting to conventional ramp-filtering. The proposed approach circumvents two principal limitations inherent in traditional methods: (i) the zero-frequency singularity and spectral distortions introduced by the mandatory ramp-filtering step, and (ii) discretization inaccuracies associated with frequency-domain interpolation. Based on the DIT, we develop a rigorous framework for consistent discrete solutions of the inverse Radon problem. Mathematical modeling demonstrates that this approach achieves quasi-exact reconstruction, with errors constrained solely by sampling parameters and grid geometry. Furthermore, while Filtered Back Projection (FBP) inherently distorts the variance of the reconstructed image, the DIT-based algorithm preserves it. Comparative simulations confirm that the proposed method eliminates common artifacts, such as intensity cupping, and consistently outperforms FBP in terms of PSNR, SSIM, and reprojection fidelity, faithfully restoring the original image's statistical characteristics.", "date": "2026-05-09", "categories": []}
{"id": "2605.13027", "title": "PRISM: Prior Rectification and Uncertainty-Aware Structure Modeling for Diffusion-Based Text Image Super-Resolution", "abstract": "Text image super-resolution (Text-SR) requires more than visually plausible detail synthesis: slight errors in stroke topology may alter character identity and break readability. Existing methods improve text fidelity with stronger recognition-based or generative priors, yet they still face two unresolved challenges under severe degradation: the text condition extracted from low-quality inputs can itself be unreliable, and a plausible global prior does not fully determine fine-grained stroke boundaries. We present PRISM, a single-step diffusion-based Text-SR framework that addresses these two challenges through Flow-Matching Prior Rectification (FMPR) and a Structure-guided Uncertainty-aware Residual Encoder (SURE). FMPR constructs a privileged training-time prior from paired low-quality/high-quality latents and learns a flow matching that transports degraded embeddings toward this restoration-oriented prior space, yielding more accurate and reliable global text guidance. SURE further predicts uncertainty-aware structural residuals to selectively absorb reliable local boundary evidence while suppressing ambiguous stroke cues. Together, these components enable explicit global prior rectification and local structure refinement within a single diffusion restoration pass. Experiments on both synthetic and real-world benchmarks show that PRISM achieves state-of-the-art performance with millisecond-level inference. Our dataset and code will be available at https://github.com/faithxuz/PRISM.", "date": "2026-05-13", "categories": []}
{"id": "2605.13146", "title": "On Hallucinations in Inverse Problems: Fundamental Limits and Provable Assessment Methods", "abstract": "Artificial intelligence (AI) has transformed imaging inverse problems, from medical diagnostics to Earth observation. Yet deep neural networks can produce hallucinations, realistic-looking but incorrect details, undermining their reliability, especially when ground truth data is unavailable. We develop a theoretical framework showing that such hallucinations are not merely artifacts of particular models, but can arise from the ill-posed nature of the inverse problem itself. We derive necessary and sufficient conditions for hallucinations, together with computable bounds on their magnitude that depend only on the forward model. Building on this theory, we introduce algorithms to: (1) estimate the minimum hallucination magnitude achievable by any reconstruction model for a given input; (2) assess the faithfulness of reconstructed details by a given reconstruction model. Experiments across three imaging tasks demonstrate that our approach applies broadly, including to modern generative models, and provides a principled way to quantify and evaluate AI hallucinations.", "date": "2026-05-13", "categories": []}
{"id": "2605.13158", "title": "Unifying Physically-Informed Weather Priors in A Single Model for Image Restoration Across Multiple Adverse Weather Conditions", "abstract": "Image restoration under multiple adverse weather conditions aims to develop a single model to recover the underlying scene with high visibility. Weather-related artifacts vary with the particle's distance to the camera according to the established scene visibility analysis, where close and faraway regions are more affected by falling drops and fog effects, respectively. Existing methods fail to consider this weather-specific physical visual process; thus, the restoration performance is limited. In this work, we analyze the common visual factors in adverse weather conditions and present a unified imaging model that considers the individually visible particles and fog-like aggregate scattering effects. Further, we design a novel weather-prior-based network, which leverages the weather-related prior information to help recover the scene by enhancing the features using the estimated occlusion and transmission. Experimental results in multiple adverse scenarios show the superiority of our method against state-of-the-art methods.", "date": "2026-05-13", "categories": []}
{"id": "2605.13457", "title": "OP4KSR: One-Step Patch-Free 4K Super-Resolution with Periodic Artifact Suppression", "abstract": "Diffusion-based real-world image super-resolution (Real-ISR) has achieved remarkable perceptual quality; however, directly super-resolving images to 4K remains limited by extreme memory consumption. Consequently, prior methods adopt patch-based inference, sacrificing global context and introducing semantic confusion, spatial inconsistency, and severe latency. We propose OP4KSR, a one-step patch-free 4K SR approach built upon the powerful Flux backbone. By leveraging the extreme-compression F16 VAE, OP4KSR makes 4K SR inference tractable under practical GPU budgets, preserving global spatial-semantic coherence while enabling highly efficient inference. However, adapting this one-step architecture intrinsically triggers severe periodic artifacts. We trace this to a RoPE base frequency allocation mismatch and intra-token spatial ambiguity, both exacerbated by the lack of iterative refinement. To suppress these artifacts, we couple RoPE base frequency rescaling (RFR) with an autocorrelation-based periodicity loss ($\\mathcal{L}_\\text{AP}$). Furthermore, we curate a dedicated training dataset alongside three benchmarks (one synthetic and two real-world) to advance 4K SR research. Extensive experiments demonstrate that OP4KSR achieves competitive perceptual quality with efficient inference, generating a $4096\\times4096$ output in only 5.75 seconds on a single NVIDIA H20 GPU.", "date": "2026-05-13", "categories": []}
{"id": "2605.13583", "title": "Phy-CoSF: Physics-Guided Continuous Spectral Fields Reconstruction and Super-Resolution for Snapshot Compressive Imaging", "abstract": "Recent advances have demonstrated that coded aperture snapshot spectral imaging (CASSI) systems show great potential for capturing 3D hyperspectral images (HSIs) from a single 2D measurement. Despite the inherent spectral continuity of scenes captured by CASSI, most existing reconstruction methods are restricted to fixed, discrete spectral outputs, thereby precluding continuous spectral reconstruction or spectral super-resolution. To address this challenge, we propose Phy-CoSF, which synergizes deep unfolding networks with implicit neural representations, establishing a new paradigm for continuous spectral reconstruction and super-resolution in CASSI. Specifically, we propose a two-phase architecture that bridges discrete-wavelength training with continuous spectral rendering, enabling the synthesis of high-fidelity HSIs at arbitrary target wavelengths. At the core of our framework lies the continuous spectral fields (CoSF) module, embedded within each unfolding stage as a dynamic prior, which comprises a triple-branch cross-domain feature mixer for comprehensive spatial-frequency-channel feature fusion, alongside a spectral synthesis head that generates spectral intensities by querying continuous wavelength coordinates. Extensive experimental results demonstrate that Phy-CoSF not only achieves continuous modeling at arbitrary spectral resolutions but also outperforms many state-of-the-art methods in both reconstruction fidelity and spectral detail preservation. Our code and more results are available at: https://github.com/PaiDii/Phy-CoSF.git.", "date": "2026-05-13", "categories": []}
{"id": "2605.13619", "title": "DeepFilters: Scattering-Aware Pupil Engineering with Learned Digital Filter Reconstruction for Extended Depth of Field Microscopy", "abstract": "Extended depth of field microscopy encodes axial information into a single acquisition through engineered point spread functions, but conventional and deep optics approaches are subject to degradation in scattering tissue. We introduce DeepFilters, a scattering-aware deep optics framework that jointly optimizes a parameterized pupil filter and a digital-filter-based reconstruction network through a calibrated differentiable forward model to achieve broad generalization without retraining. Incorporating empirical scattering kernels, physics-guided regularization, and a hybrid genetic-gradient initialization strategy, DeepFilters extends the PSF from 16 micron to >400 micron in clear media and enables signal recovery beyond 120 micron deep in biological tissues, validated across fixed brain slices and sea urchin embryos.", "date": "2026-05-13", "categories": []}
{"id": "2605.13664", "title": "HADAR-Based Thermal Infrared Hyperspectral Image Restoration", "abstract": "Thermal-infrared (TIR) hyperspectral imagery (HSI) provides critical scene information for various applications. However, its practical utility is severely limited by unique sensor degradations beyond the capabilities of existing restoration methods, which are ignorant of underlying thermal physics. Here, we propose HAIR (HADAR-based Image Restoration) as a physics-driven framework for ground-based TIR-HSI restoration. HAIR utilizes the HADAR rendering equation (HRE) and combines it with the atmospheric downwelling radiative transfer equation (RTE) to model TIR-HSI using temperature, emissivity, and texture (TeX) physical triplets. This physical model leads to a TeX decompose-synthesize strategy that guarantees physical consistency and spatio-spectral noise resilience, in stark contrast to existing approaches. Moreover, our framework uses a forward-modeled atmospheric downwelling reference, along with spectral smoothness of emissivity and blackbody radiation, to enable spectral calibration and generation that would otherwise be elusive. Our extensive experiments on the outdoor DARPA Invisible Headlights dataset and in-lab FTIR measurements show that HAIR consistently outperforms state-of-the-art methods across denoising, inpainting, spectral calibration, and spectral super-resolution, establishing a benchmark in objective accuracy and visual quality.", "date": "2026-05-13", "categories": []}
{"id": "2605.13744", "title": "Aligning Network Equivariance with Data Symmetry: A Theoretical Framework and Adaptive Approach for Image Restoration", "abstract": "Image restoration is an inherently ill posed inverse problem. Equivariant networks that embed geometric symmetry priors can mitigate this ill posedness and improve performance. However, current understanding of the relationship between network equivariance and data symmetry remains largely heuristic. Particularly for real world data with imperfect symmetry, existing research lacks a systematic theoretical framework to quantify symmetry, select transformation groups, or evaluate model data alignment. To bridge this gap, we conduct an analysis from an optimization perspective and formalize the intrinsic relationship among data symmetry priors, model equivariance, and generalization capability. Specifically, we propose for the first time a quantifiable definition of non strict symmetry at the dataset level (rather than sample level) and use it as a constraint to formulate the restoration inverse problem. We then show that the equivariance for restoration models can be naturally derived from this inverse problems incorporated the proposed symmetry constraints, and that the equivariance error of the optimal restoration operator is strictly bounded by the data symmetry error and the discretization mesh size. Furthermore, by analyzing the network's empirical risk, we demonstrate that aligning equivariance with data symmetry optimizes the bias variance trade off, minimizing the total expected risk. Guided by these insights, we propose a Sample Adaptive Equivariant Network that uses a hypernetwork and transformation learnable equivariant convolutions to dynamically align with each sample's inherent symmetry. Extensive experiments on super resolution, denoising, and deraining validate our theoretical findings and show significant superiority over standard baselines and traditional equivariant models. Our code and supplementary material are available at https://github.com/tanfy929/SA-Conv.", "date": "2026-05-13", "categories": []}
{"id": "2511.17038", "title": "DAPS++: Rethinking Diffusion Inverse Problems with Decoupled Posterior Annealing", "abstract": "From a Bayesian perspective, score-based diffusion solves inverse problems through joint inference, embedding the likelihood with the prior to guide the sampling process. However, this formulation fails to explain its practical behavior: the prior offers limited guidance, while reconstruction is largely driven by the measurement-consistency term, leading to an inference process that is effectively decoupled from the diffusion dynamics. We show that the diffusion prior in these solvers functions primarily as a warm initializer that places estimates near the data manifold, while reconstruction is driven almost entirely by measurement consistency. Based on this observation, we introduce \\textbf{DAPS++}, which fully decouples diffusion-based initialization from likelihood-driven refinement, allowing the likelihood term to guide inference more directly while maintaining numerical stability and providing insight into why unified diffusion trajectories remain effective in practice. By requiring fewer function evaluations (NFEs) and measurement-optimization steps, \\textbf{DAPS++} achieves high computational efficiency and robust reconstruction performance across diverse image restoration tasks.", "date": "2025-11-21", "categories": []}
{"id": "2512.05683", "title": "Physics-Informed Graph Neural Networks for Frequency-Aware Optical Aberration Correction", "abstract": "Optical aberrations significantly degrade image quality in microscopy, particularly when imaging deeper into samples. These aberrations arise from distortions in the optical wavefront and can be mathematically represented using Zernike polynomials. Existing methods often address only mild aberrations on limited sample types and modalities, typically treating the problem as a black-box mapping without leveraging the underlying optical physics of wavefront distortions. We propose ZRNet, a physics-informed framework that jointly performs Zernike coefficient prediction and optical image Restoration. We contribute a Zernike Graph module that explicitly models physical relationships between Zernike polynomials based on their azimuthal degrees-ensuring that learned corrections align with fundamental optical principles. To further enforce physical consistency between image restoration and Zernike prediction, we introduce a Frequency-Aware Alignment (FAA) loss, which better aligns Zernike coefficient prediction and image features in the Fourier domain. Extensive experiments on CytoImageNet demonstrates that our approach achieves state-of-the-art performance in both image restoration and Zernike coefficient prediction across diverse microscopy modalities and biological samples with complex, large-amplitude aberrations. We further validate on experimental PSF data from a physical microscope and demonstrate robustness to realistic sensor noise, confirming generalisation beyond simulated conditions. Code is available at https://github.com/janetkok/ZRNet.", "date": "2025-12-05", "categories": []}
{"id": "2512.07150", "title": "FlowLPS: Langevin-Proximal Sampling for Flow-based Inverse Problem Solvers", "abstract": "Deep generative models are powerful priors for imaging inverse problems, but training-free solvers for latent flow models face a practical finite-step trade-off. Optimization-heavy methods quickly improve measurement consistency, but in highly nonlinear latent spaces, their results can depend strongly on where local refinement is initialized, often degrading perceptual realism. In contrast, stochastic sampling methods better preserve posterior exploration, but often require many iterations to obtain sharp, measurement-consistent reconstructions. To address this trade-off, we propose FlowLPS, a training-free latent flow inverse solver based on Langevin-Proximal Sampling. At each reverse step, FlowLPS uses a few Langevin updates to perturb the model-predicted clean estimate in posterior-oriented directions, providing stochastic initializations for local refinement. It then applies local MAP-style proximal refinement to rapidly improve measurement consistency from the Langevin-updated estimate. We additionally use controlled pCN-style re-noising to stabilize the reverse trajectory while retaining trajectory coherence. Experiments on FFHQ and DIV2K across five linear inverse problems show that FlowLPS achieves a strong balance between measurement fidelity and perceptual quality, with additional experiments on pixel-space inverse problems and phase retrieval.", "date": "2025-12-08", "categories": []}
{"id": "2605.11475", "title": "Deep Probabilistic Unfolding for Quantized Compressive Sensing", "abstract": "We propose a deep probabilistic unfolding model to address the classical quantized compressive sensing problem that leverages an unfolding framework to enhance the reconstruction accuracy and efficiency. Unlike previous unfolding methods that apply L2 projection to measurements, we derive a closed-form, numerically stable likelihood gradient projection, which allows the model to respect the true quantization physics, turning the hard quantization constraint into a soft probabilistic guidance. Furthermore, an efficient, dual-domain Mamba module is specifically designed to dynamically capture and fuse the multi-scale local and global features, ensuring the interactions between the distant but correlated regions. Extensive experiments demonstrate the state-of-the-art performance of the proposed method over previous works, which is capable of promoting the application of quantized compressive sensing in real life.", "date": "2026-05-12", "categories": []}
{"id": "2605.11508", "title": "LiBrA-Net: Lie-Algebraic Bilateral Affine Fields for Real-Time 4K Video Dehazing", "abstract": "Currently, there is a gap in the field of ultra-high-definition (UHD) video dehazing due to the lack of a benchmark for evaluation. Furthermore, existing video dehazing methods cannot run on consumer-grade GPUs when processing continuous UHD sequences of 3--5 frames at a time. In this paper, we address both issues with a new benchmark and an efficient method. Our key observation is that atmospheric dehazing reduces to a per-pixel affine transform governed by the low-frequency depth field, which can be compactly encoded in bilateral grids whose prediction cost is decoupled from the output resolution. Building on this, we propose LiBrA-Net, which factorizes the spatiotemporal affine field into a spatial--color and a temporal bilateral sub-grid predicted at a fixed low resolution, fuses their coefficients in the $\\mathfrak{gl}(3)$ Lie algebra under group-theoretic regularization, maps the result to invertible GL(3) transforms via a Cayley parameterization, and restores high-frequency detail through a lightweight input-guided branch. We further release UHV-4K, the first paired 4K video dehazing benchmark with depth, transmission, and optical-flow annotations on every frame. Across UHV-4K, REVIDE, and HazeWorld, LiBrA-Net sets a new state of the art among compared video dehazing methods while running native 4K at 25 FPS on a single GPU with only 6.12 M parameters. Code and data are available at https://anonymous.4open.science/r/LiBrA-Net-42B8.", "date": "2026-05-12", "categories": []}
{"id": "2605.11583", "title": "NexOP: Joint Optimization of NEX-Aware k-space Sampling and Image Reconstruction for Low-Field MRI", "abstract": "Modern low-field magnetic resonance imaging (MRI) technology offers a compelling alternative to standard high-field MRI, with portable, low-cost systems. However, its clinical utility is limited by a low Signal-to-Noise Ratio (SNR), which hampers diagnostic image quality. A common approach to increase SNR is through repetitive signal acquisitions, known as NEX, but this results in excessively long scan durations. Although recent work has introduced methods to accelerate MRI scans through k-space sampling optimization, the NEX dimension remains unexploited; typically, a single sampling mask is used across all repetitions. Here we introduce NexOP, a deep-learning framework for joint optimization of the sampling and reconstruction in multi-NEX acquisitions, tailored for low-SNR settings. NexOP enables optimizing the sampling density probabilities across the extended k-space-NEX domain, under a fixed sampling-budget constraint, and introduces a new deep-learning architecture for reconstructing a single high-SNR image from multiple low-SNR measurements. Experiments with raw low-field (0.3T) brain data demonstrate that NexOP consistently outperforms competing methods, both quantitatively and qualitatively, across diverse acceleration factors and tissue contrasts. The results also demonstrate that NexOP yields non-uniform sampling strategies, with progressively decreasing sampling across repetitions, hence exploiting the NEX dimension efficiently. Moreover, we present a theoretical analysis supporting these numerical observations. Overall, this work proposes a sampling-reconstruction optimization framework highly suitable for low-field MRI, which can enable faster, higher-quality imaging with low-cost systems and contribute to advancing affordable and accessible healthcare.", "date": "2026-05-12", "categories": []}
{"id": "2605.11585", "title": "A Mixture Autoregressive Image Generative Model on Quadtree Regions for Gaussian Noise Removal via Variational Bayes and Gradient Methods", "abstract": "This paper addresses the problem of image denoising for grayscale images. We propose a probabilistic image generative model that combines a quadtree region-partitioning model with a mixture autoregressive model, and propose a framework that reduces MAP (maximum a posteriori)-estimation-based denoising to the maximization of a variational lower bound. To maximize this lower bound, we develop an algorithm that alternately applies variational Bayes and gradient methods. We particularly demonstrate that the gradient-based update rule can be computed analytically without numerical computation or approximation. We carried out some experiments to verify that the proposed algorithm actually removes image noise and to identify directions for future improvement.", "date": "2026-05-12", "categories": []}
{"id": "2605.11829", "title": "Bin Latent Transformer (BiLT): A shift-invariant autoencoder for calibration-free spectral unmixing of turbid media", "abstract": "The accurate recovery of constituent-level optical properties from integrating sphere measurements is a central analytical challenge in pharmaceutical analysis, food science, and biomedical diagnostics. Neural network autoencoders can extract spectrally resolved absorption and scattering coefficients for each constituent without prior knowledge, but their fully connected encoders bind learned features to absolute wavelength indices, causing accuracy loss under spectrometer calibration drift or hardware exchange. This work introduces the Bin Latent Transformer (BiLT)-Autoencoder, in which the dense encoder is replaced by a cross-attention scanner: 16 learnable probe vectors query a convolutional feature map, aggregating morphological spectral information independently of absolute wavelength position. A physics-constrained linear decoder with enforced absorption/scattering separation and a three-phase curriculum augmentation strategy complete the architecture. On a liquid phantom benchmark (intralipid and two ink absorbers; 496 samples), the model achieves $R^2 = 0.979$ and $0.975$ for $\\mu_a(\\lambda)$ and $\\mu_s'(\\lambda)$, respectively, on held-out test spectra, maintaining $R^2 > 0.90$ for $\\mu_a$ and $R^2 \\approx 0.99$ for $\\mu_s'$ across the full tested shift range of $\\pm 10$ spectral bands. The model generalises to a simulated spectrometer with a broader instrument line shape (${\\approx}24$nm FWHM) without retraining, retaining $R^2 \\approx 0.96$ and $0.974$ for the two channels. Attention map analysis reveals a physically interpretable two-component probe strategy: sparse anchor probes at absorption-edge wavelengths combined with a diffuse, SNR-driven ensemble at the high-transmittance long-wavelength region, which recruits additional probes dynamically under noise to provide implicit spectral averaging.", "date": "2026-05-12", "categories": []}
{"id": "2602.04189", "title": "Beyond Accuracy: Evaluating Posterior Fidelity of Diffusion Inverse Solvers", "abstract": "Uncertainty evaluation is critical in scientific and engineering inverse problems. However, existing benchmarks on Diffusion Inverse Solvers (DIS) primarily focus on reconstruction accuracy but overlook uncertainty and distributional behavior. Since stochastic inverse solvers represent uncertainty through diffusion-based posterior samples, evaluating how well their generated samples capture the target posterior distribution becomes an important aspect of uncertainty quantification. To address this limitation and better understand the distributional behavior of diffusion samplers, we conduct a systematic study to investigate the posterior fidelity of a broad range of existing DIS methods in controlled simulation settings with a known analytical true posterior. Furthermore, to enable posterior-aware evaluation on real-world inverse problems where ground-truth posterior is unavailable, we propose score-based Kernel Stein Discrepancy (score-KSD), a theoretically-grounded and ground-truth-free metric that measures the consistency of the distribution of generated samples from a DIS method with the target posterior score field, induced by the forward model and learned diffusion prior. Through both simulation experiments and real-world inverse problem solving, we validate the effectiveness of the proposed score-KSD and demonstrate that it provides meaningful posterior fidelity diagnostics beyond reconstruction accuracy, revealing that higher reconstruction accuracy does not necessarily imply better posterior consistency.", "date": "2026-02-04", "categories": []}
{"id": "2605.07429", "title": "Towards Photorealistic and Efficient Bokeh Rendering via Diffusion Framework", "abstract": "Existing mobile devices are constrained by compact optical designs, such as small apertures, which make it difficult to produce natural, optically realistic bokeh effects. Although recent learning-based methods have shown promising results, they still struggle with photos captured under high digital zoom levels, which often suffer from reduced resolution and loss of fine details. A naive solution is to enhance image quality before applying bokeh rendering, yet this two-stage pipeline reduces efficiency and introduces unnecessary error accumulation. To overcome these limitations, we propose MagicBokeh, a unified diffusion-based framework designed for high-quality and efficient bokeh rendering. Through an alternative training strategy and a focus-aware masked attention mechanism, our method jointly optimizes bokeh rendering and super-resolution, substantially improving both controllability and visual fidelity. Furthermore, we introduce degradation-aware depth module to enable more accurate depth estimation from low-quality inputs. Experimental results demonstrate that MagicBokeh efficiently produces photorealistic bokeh effects, particularly on real-world low-resolution images, paving the way for future advancements in bokeh rendering. Our code and models are available at https://github.com/vivoCameraResearch/MagicBokeh.", "date": "2026-05-08", "categories": []}
{"id": "2605.08173", "title": "CASISR: Circular Arbitrary-Scale Image Super-Resolution", "abstract": "The generalization performance (GP) of deep learning-based arbitrary-scale image super-resolution (ASISR) methods is subject to limited training datasets and unlimited testing datasets. It is vitally significant to enhance the GP of the pretrained ASISR models by making full use of the testing samples. The ASISR models usually employ an open-loop architecture from low-resolution (LR) images to super-resolution (SR) images. The degradation model from SR samples to LR samples is known bicubic down-sampling for the classical ASISR, is supposed down-sampling with additive random noise for the blind ASISR, and is learnable for the real-world ASISR. Combining the ASISR and degradation models, it is potentially possible to adopt a closed-loop architecture based on the automatic control theory for strengthening the GP of the ASISR methods. Therefore, this paper proposes a closed-loop architecture, circular ASISR (CASISR), to lift the capability of image reconstruction. A mathematical nonlinear loop equation is established to describe the CASISR, the reasonability of the CASISR is proven by conditional probability theory, and the stability of the CASISR is proven by Taylor series approximation. The first-order and second-order absolute difference images are defined to compare the image reconstruction performance of the ASISR and the CASISR methods. Comprehensive simulation experiments show that the proposed CASISR approach outperforms the eight state-of-the-art ASISR approaches in the quality of image reconstruction. Especially, the proposed CASISR is extraordinarily suitable for fractional SR scale factors and is extremely effective for text and stripe images with drastically changed edges.", "date": "2026-05-05", "categories": []}
{"id": "2605.08275", "title": "Model-based Dynamic 3D MRI Reconstructions using Neural Fields and Tensor Product Expansions", "abstract": "Conventional MRI reconstruction methods treat images and coil sensitivities as discrete objects, leading to high memory demands and limited structural awareness that hamper effective regularization. These limitations hinder accurate reconstruction in highly undersampled scenarios, such as dynamic 3D cardiac magnetic resonance (CMR). We introduce a discretization-free, memory-efficient, model-based framework for dynamic 2D and 3D MRI reconstruction from highly undersampled data. We represent magnetization and coil sensitivities as continuous objects -- differentiable functions -- using tensor products of univariate neural fields. This tensor product structure enables scalable optimization in high-dimensional spatiotemporal settings. Our method outperforms state-of-the-art model-based reconstructions in dynamic 2D and 3D MR settings, preserving structure and motion even under aggressive undersampling (e.g., acceleration factor 16).", "date": "2026-05-08", "categories": []}
{"id": "2605.08566", "title": "MicroDiffuse3D: A Foundation Model for 3D Microscopy Imaging Restoration", "abstract": "Chemical imaging enables label-free visualization of cells, tissues and living systems while providing direct biochemical information that is difficult to obtain with conventional fluorescence microscopy. Despite its promise in applications ranging from intraoperative diagnosis to drug-response analysis, its broader use remains limited by slow data acquisition, particularly for three-dimensional imaging. Here we present MicroDiffuse3D, a pretrained foundation model for 3D microscopy image restoration that recovers high-quality volumetric structure from degraded low-resolution measurements acquired at substantially higher throughput. We evaluated MicroDiffuse3D across three challenging restoration settings, including 3D super-resolution under 16-fold volumetric sparsity, joint degradation in resolution and noise, and 3D denoising in the low signal-to-noise ratio (SNR) regime, where the model delivered clear gains over strong baselines. Under the sparse 3D super-resolution setting, MicroDiffuse3D produced clearer continuity across depth with fewer artifacts and improved segmentation quality by 10.58% and line-profile concordance by 15.59%. Together, our results establish pretrained 3D restoration as a broadly applicable strategy for overcoming the throughput and SNR limitations in volumetric chemical imaging, enabling high-resolution analysis at scales and speeds that were previously difficult to achieve.", "date": "2026-05-08", "categories": []}
{"id": "2605.08854", "title": "Restoration-Aligned Generative Flow Models for Blind Motion Deblurring", "abstract": "Generative flow models offer powerful priors learned from large-scale natural images, but directly adapting them to restoration tasks such as motion deblurring causes severe fidelity degradation, as their training objective is inherently misaligned with restoration. We present DeblurFlow, a framework that resolves this misalignment by reformulating the flow trajectory itself: we replace the noise endpoint with the blur observation, which makes the underlying vector field coincide with the residual error between blur and clean images. Under this formulation, the standard flow matching loss naturally takes the form of a residual loss, allowing pretrained flow models to be optimized under restoration-aligned objectives via LoRA adaptation. This formulation further enables a dual-expert sampling strategy: a fidelity expert provides a high-fidelity initialization, e.g., PSNR 33.69 dB, and DeblurFlow enhances perceptual quality with only a marginal fidelity reduction to 33.05 dB, whereas directly applying a generative model on top of a fidelity expert decreases PSNR to 27.60 dB. To make this practical, we further introduce r-space, a latent space tailored for residual decoding rather than image reconstruction, which reduces encoder-decoder cost by up to 9$\\times$over standard VAE latents. Extensive experiments on GoPro, HIDE, RealBlur, and RWBI demonstrate that DeblurFlow achieves strong restoration fidelity and perceptual realism, while remaining computationally practical.", "date": "2026-05-09", "categories": []}
{"id": "2605.09477", "title": "Outlier-Robust Diffusion Solvers for Inverse Problems", "abstract": "Methods based on diffusion models (DMs) for solving inverse problems (IPs) have recently achieved remarkable performance. However, DM-based methods typically struggle against outliers, which are common in real-world measurements. In this work, to tackle IPs with outliers, we first refine the measurement via explicit noise estimation to mitigate the effect of noise. Subsequently, we formulate an iteratively reweighted least squares objective based on the Huber loss to address the outliers. We propose a method utilizing gradient descent to approximately solve the corresponding optimization problem for the robust objective. To avoid delicate tuning of the learning rate required by the gradient descent method, we further employ the conjugate gradient method with an efficient strategy for updating. Extensive experiments on multiple image datasets for linear and nonlinear tasks under various conditions demonstrate that our proposed methods exhibit robustness to outliers and outperform recent DM-based methods in most cases.", "date": "2026-05-10", "categories": []}
{"id": "2605.09628", "title": "DegBins: Degradation-Driven Binning for Depth Super-Resolution", "abstract": "Depth super-resolution (DSR) aims to recover a high-resolution (HR) depth map from its low-resolution (LR) counterpart. With color image guidance, this task is typically formulated as learning the residual between HR and LR in a low-dimensional feature space. However, this additive formulation is insufficient to accurately capture the complex relationship between HR and LR, especially under spatially varying degradations. In this paper, we introduce DegBins, a novel DSR framework that leverages degradation-driven binning to adaptively enhance residual modeling. Specifically, DegBins reformulates the regression-based DSR as a hybrid classification-regression problem, where the residual depth is represented as a linear combination of discrete depth bins weighted by their learned probability distribution, yielding more flexible and expressive representations. Furthermore, DegBins models the degradation relationship between HR and LR in a high-dimensional feature space, enabling adaptive bin range adjustment and probability optimization conditioned on local degradation characteristics. To progressively improve reconstruction quality, DegBins adopts a multi-stage refinement scheme, where each stage performs finer-grained bin partitioning and probability updating based on the former estimation. This coarse-to-fine design facilitates more accurate depth recovery, particularly in regions with severe degradations or complex structural variations. Extensive experiments across five benchmarks demonstrate that DegBins consistently outperforms existing state-of-the-art methods in terms of accuracy, robustness, and generalization.", "date": "2026-05-10", "categories": []}
{"id": "2605.10008", "title": "Measurement-Adapted Eigentask Representations for Photon-Limited Optical Readout", "abstract": "Optical readout in low-light imaging is fundamentally limited by measurement noise, including photon shot noise, detector noise, and quantization error. In this regime, downstream inference depends not only on the optical front end, but also on how noisy high-dimensional sensor measurements are represented before classification or decision-making. Here we show that eigentasks provide a measurement-adapted representation for optical sensor outputs by ordering readout features according to their resolvability under noise. Using experimental data from a lens-based optical imaging system and a reanalysis of published data from a single-photon-detection neural network, we find that eigentask representations frequently outperform standard baselines including principal component analysis and filtering-based compression. The advantage is most pronounced in photon-limited, few-shot, and higher-difficulty classification regimes. In few-shot MPEG-7 classification, for example, the advantage over other methods reaches about 10 percentage points as the number of classes increases. In these settings, eigentasks yield more informative low-dimensional features and improve sample-efficient downstream learning. These results identify measurement-adapted representation as a promising strategy for optical inference when photon budget, acquisition time, and task complexity are constrained.", "date": "2026-05-11", "categories": []}
{"id": "2605.10275", "title": "PolarVSR: A Unified Framework and Benchmark for Continuous Space-Time Polarization Video Reconstruction", "abstract": "Polarimetric imaging captures surface polarization characteristics, such as the Degree of Linear Polarization (DoLP) and the Angle of Polarization (AoP). In mainstream Division of-Focal-Plane (DoFP) color polarization imaging, recovering polarization parameters from captured mosaic arrays remains a challenging inverse problem. Existing DoFP cameras also face hardware bottlenecks and often cannot support high-frame-rate acquisition, limiting polarimetric imaging in dynamic video tasks. These limitations motivate joint spatial and temporal enhancement. To this end, we propose the first space-time polarization video reconstruction architecture. The method jointly models polarization directions in space and time and uses a polarization-aware implicit neural representation for continuous, high-fidelity upsampling. By analyzing temporal variations in polarization parameters, we further introduce a flow-guided polarization variation loss to supervise polarization dynamics. We also establish the first large-scale color DoFP polarization video benchmark to support this research direction. Extensive experiments on this benchmark demonstrate the effectiveness of the method.", "date": "2026-05-11", "categories": []}
{"id": "2605.10374", "title": "Halo Separation-guided Underwater Multi-scale Image Restoration", "abstract": "Underwater images captured by Autonomous Underwater Vehicles (AUVs) are inevitably affected by artificial light sources, which often produce halos in the foreground of the camera and seriously interfere with the quality of the image. The existing underwater image enhancement methods fail to fully consider this key problem, and the robustness of processing images under artificial light scenes is poor. In practical applications, since underwater image enhancement itself is a very challenging task, the influence of artificial light sources will lead to serious degradation of image performance and affect subsequent vision tasks. In order to effectively deal with this problem, this paper designs a single halo image correction method based on an iterative structure. The network is mainly divided into two sub-networks, one is the halo layer separation sub-network which aims to separate the halo by gradient minimization, and the other is the multi-scale recovery sub-network which aims to recover the image information masked by halo. The UIEB and EUVP synthetic datasets are used for training to ensure that the network can fully learn the characteristics and laws of underwater halo images. Then a large number of halo images taken in an underwater environment with real artificial light are collected for testing. In addition, the brightness distribution characteristics of underwater halo images are analyzed and the radial gradient is introduced to constraint eliminate halo to improve the effect of underwater image restoration.", "date": "2026-05-11", "categories": []}
{"id": "2605.10629", "title": "Product-of-Gaussian-Mixture Diffusion Models for Joint Nonlinear MRI Reconstruction", "abstract": "Recently, diffusion models have attracted considerable attention for magnetic resonance image reconstruction due to their high sample quality. However, most existing methods rely on large networks with opaque time-conditioning mechanisms, and require offline coil sensitivity estimation. This results in limited interpretability of the reconstruction process and reduced flexibility in the acquisition setup. To address these limitations, we jointly reconstruct the image and the coil sensitivities by combining the parameter-efficient product-of-Gaussian-mixture diffusion model as an image prior with a classical smoothness prior on the coil sensitivities. The proposed method is fast and robust to both contrast and anatomical distribution shifts as well as changing k-space trajectories. Finally, we propose a more expressive parameterization of the image prior which improves results in denoising and magnetic resonance image reconstruction.", "date": "2026-05-11", "categories": []}
{"id": "2604.24136", "title": "Bridging Restoration and Generation Manifolds in One-Step Diffusion for Real-World Super-Resolution", "abstract": "Pretrained diffusion models have revolutionized real-world image super-resolution (Real-ISR) but suffer from computational bottlenecks due to iterative sampling. Recent single-step distillation accelerates inference but faces a stark perception-distortion trade-off due to rigid timestep initialization, distributional trajectory mismatches, and fragile stochastic modulation. To address this, we present Adaptive Inversion and Degradation-aware Sampling for Real-ISR (IDaS-SR), a one-step framework bridging the deterministic restoration and stochastic generation manifolds. At its core, the Manifold Inversion Noise Estimator (MINE) resolves these initialization and trajectory mismatches by predicting a severity-aware timestep and inversion noise, precisely anchoring low-quality latents onto the diffusion trajectory. Furthermore, to mitigate fragile stochastic modulation, we propose CHARIOT, a continuous generative steering mechanism. By rescheduling trajectories and interpolating noise, it enables explicit navigation of the perception-distortion boundary without compromising structural priors. Extensive experiments demonstrate that IDaS-SR outperforms state-of-the-art methods, seamlessly transitioning from a rigorous structural restorer to a sophisticated texture hallucinator in a single inference step.", "date": "2026-04-27", "categories": []}
{"id": "2605.07359", "title": "UniISP: A Unified ISP Framework for Both Human and Machine Vision", "abstract": "Compared to RGB images, raw sensor data provides a richer representation of information, which is crucial for accurate recognition, particularly under challenging conditions such as low-light environments. The traditional Image Signal Processing (ISP) pipeline generates visually pleasing RGB images for human perception through a series of steps, but some of these operations may adversely impact the information integrity by introducing compression and loss. Furthermore, in computer vision tasks that directly utilize raw camera data, most existing methods integrate minimal ISP processing with downstream networks, yet the resulting images are often difficult to visualize or do not align with human aesthetic preferences. This paper proposes UniISP, a novel ISP framework designed to simultaneously meet the requirements of both human visual perception and computer vision applications. By incorporating a carefully designed Hybrid Attention Module (HAM) and employing supervised learning, the proposed method ensures that the generated images are visually appealing. Additionally, a Feature Adapter module is introduced to effectively propagate informative features from the ISP stage to subsequent downstream networks. Extensive experiments demonstrate that our approach achieves state-of-the-art performance across various scenarios and multiple datasets, proving its generalizability and effectiveness.", "date": "2026-05-08", "categories": []}
{"id": "2605.07495", "title": "Lightweight Unpaired Smartphone ISP Transfer with Semantic Pseudo-Pairing", "abstract": "Unpaired smartphone ISP is a challenging problem due to the lack of scene and color alignment between RAW and target RGB images. Many existing methods either require paired data or rely heavily on adversarial training, which can become unstable in the unpaired setting. In this work, we present a simple and effective approach developed for the NTIRE 2026 Learned Smartphone ISP Challenge with Unpaired Data. Our method first reconstructs larger images from training patches to recover global context. Then, we extract semantic embeddings with DINOv2, and use fused Gromov-Wasserstein (FGW) optimal transport to build pseudo pairs between RAW and RGB images at both image and patch levels. This semantic matching allows us to partially alleviate the unpairedness of the data and build these pseudo input-target pairs. Based on these pseudo pairs, we train a lightweight CNN with only 7K parameters for color rendering. The network is designed to be compact and focus on color transformation rather than structural change, which helps reduce artifacts and improve training stability. Our challenge submission achieves 22.569 PSNR, 0.675 SSIM, and 8.067 $\\Delta E$ on the final hidden test set, significantly improving over the baseline and achieving the 3rd best SSIM and $\\Delta E$ among all challenge entries. Our code is available at github.com/nuniniyujin/Unpaired-ISP .", "date": "2026-05-08", "categories": []}
{"id": "2603.04673", "title": "sFRC for assessing hallucinations in medical image restoration", "abstract": "Deep learning (DL) methods are currently being explored to restore images from sparse-view-, limited-data-, and undersampled-based acquisitions in medical applications. Although outputs from DL may appear visually appealing based on likability/subjective criteria (such as less noise, smooth features), they may also suffer from hallucinations. This issue is further exacerbated by a lack of easy-to-use techniques and robust metrics for the identification of hallucinations in DL outputs. In this work, we propose performing Fourier Ring Correlation (FRC) analysis over small patches and concomitantly (s)canning across DL outputs and their reference counterparts to detect hallucinations (termed as sFRC). We describe the rationale behind sFRC and provide its mathematical formulation. The parameters essential to sFRC may be set using predefined hallucinated features annotated by subject matter experts or using imaging theory-based hallucination maps. We use sFRC to detect hallucinations for three undersampled medical imaging problems: CT super-resolution, CT sparse view, and MRI subsampled restoration. In the testing phase, we demonstrate sFRC's effectiveness in detecting hallucinated features for the CT problem and sFRC's agreement with imaging theory-based outputs on hallucinated feature maps for the MR problem. Finally, we quantify the hallucination rates of DL methods on in-distribution versus out-of-distribution data and under increasing subsampling rates to characterize the robustness of DL methods. Beyond DL-based methods, sFRC's effectiveness in detecting hallucinations for a conventional regularization-based restoration method and a state-of-the-art unrolled method is also shown.", "date": "2026-03-04", "categories": []}
{"id": "2605.03749", "title": "FluxFlow: Conservative Flow-Matching for Astronomical Image Super-Resolution", "abstract": "Ground-to-space astronomical super-resolution requires recovering space-quality images from ground-based observations that are simultaneously limited by pixel sampling resolution and atmospheric seeing, which imposes a stochastic, spatially varying PSF that cannot be resolved through upsampling alone. Existing methods rely on synthetic training pairs that fail to capture real atmospheric statistics and are prone to either over-smoothed reconstructions or hallucination sources with no physical counterpart in the observed sky. We propose FluxFlow, a conservative pixel-space flow-matching framework that incorporates observation uncertainty and source-region importance weights during training, and a training-free Wiener-regularized test-time correction to suppress hallucination sources while preserving recovered detail. We further construct the DESI--HST Dataset, the large-scale real-world benchmark comprising 19,500 real co-registered ground-to-space image pairs with real atmospheric PSF variation. Experiments demonstrate that FluxFlow consistently outperforms existing baseline methods in both photometric and scientific accuracy.", "date": "2026-05-05", "categories": []}
{"id": "2605.05990", "title": "iPhoneBlur: A Difficulty-Stratified Benchmark for Consumer Device Motion Deblurring", "abstract": "Motion blur restoration on consumer mobile devices is typically evaluated using aggregate metrics that obscure performance variation across blur difficulty, masking model behavior under real deployment conditions. This work introduces iPhoneBlur, a difficulty-stratified benchmark of 7,400 image pairs synthesized from high-framerate iPhone 17 Pro videos captured in diverse real-world scenarios. Samples are partitioned into Easy, Medium, and Hard categories through PSNR-guided adaptive temporal windowing, with stratification validated by monotonic 2.2x increase in optical flow magnitude across tiers. Each sample includes comprehensive metadata enabling investigation of ISP-aware and difficulty-adaptive restoration strategies. Spectral analysis confirms synthesized blur exhibits high-frequency suppression patterns consistent with authentic motion degradation. Evaluation of six architectures reveals consistent 7-9 dB performance degradation from Easy to Hard subsets, a substantial gap entirely hidden by aggregate reporting. The benchmark further exposes a domain gap between professional and consumer cameras which targeted fine-tuning substantially recovers. By coupling difficulty stratification with deployment-critical metadata, iPhoneBlur enables systematic assessment of model reliability and failure modes for resource-constrained edge systems.", "date": "2026-05-07", "categories": []}
{"id": "2605.06127", "title": "Continuous Expert Assembly: Instance-Conditioned Low-Rank Residuals for All-in-One Image Restoration", "abstract": "Real-world image degradation is often unknown, spatially non-uniform, and compositional, requiring all-in-one restoration models to adapt a single set of weights to diverse local corruption patterns without test-time degradation labels. Existing methods typically modulate a shared backbone with global prompts or degradation descriptors, or route features through predefined expert pools. However, compact global conditioning can bottleneck localized degradation evidence, while static expert routing may produce homogeneous updates or rely on unstable sparse assignments. We propose \\textbf{Continuous Expert Assembly} (CEA), a token-wise dynamic parameterization framework for all-in-one image restoration. CEA employs a lightweight \\textbf{Cross-Attention Hyper-Adapter} to probe intermediate spatial features and synthesize instance-conditioned low-rank routing bases and residual directions. Each spatial token then assembles its own residual update via dense signed dot-product affinities over the generated rank-wise components, avoiding external prompts, static expert banks, and discrete Top-\n  selection. The resulting assembly rule also admits a linear-attention perspective, making its dense token-wise routing behavior transparent. Experiments on AIO-3, AIO-5, and CDD-11 show that CEA improves average restoration quality over strong prompt-, descriptor-, and expert-based baselines, with the clearest gains on spatially varying and compositional degradations, while maintaining favorable parameter, FLOP, and runtime efficiency.", "date": "2026-05-07", "categories": []}
{"id": "2605.06214", "title": "Differentiable Adaptive 4D Structured Illumination for Joint Capture of Shape and Reflectance", "abstract": "We present a differentiable framework to adaptively compute 4D illumination conditions with respect to an object, for efficient, high-quality simultaneous acquisition of its shape and reflectance, with a unified spatial-angular structured light and a single camera. Using a simple histogram-based pixel-level probability model for depth and reflectance, we differentiably link the next illumination condition(s) with a loss that encourages the reduction in depth uncertainty. As new structured illumination is cast, corresponding image measurements are used to update the uncertainty at each pixel. Finally, a fine-tuning-based approach reconstructs the depth map and reflectance parameter maps, by minimizing the differences between all physical measurements and their simulated counterparts. The effectiveness of our framework is demonstrated on physical objects with wide variations in shape and appearance. Our depth results compare favorably with state-of-the-art techniques, while our reflectance results are comparable when validated against photographs.", "date": "2026-05-07", "categories": []}
{"id": "2605.06359", "title": "The frame-level leakage trap: rethinking evaluation protocols for intrinsic image decomposition, with source-separable uncertainty as a case study", "abstract": "Evaluation protocols for learned intrinsic image decomposition on MPI Sintel have been inconsistent. Several prior works split the dataset by frames, which allows spatially similar frames of the same scene to appear in both train and test partitions. We quantify this leakage effect for the first time, across three architectures: a frame-level split inflates test R_PSNR by 1.6 to 2.0 dB (p less than 0.01 for all three, paired t-test across 3 seeds) relative to a scene-level split, confirming an architecture-independent protocol effect. A three-point gradient (random/temporal/scene) shows the gap is continuous, and under extended training the frame-level inflation exceeds 10 dB. We advocate scene-level splits as the community standard and provide reference numbers for six representative models under this protocol. As a case study within the corrected protocol, we present a physics-informed decomposition I = R composed with S + N with a source-separable three-way heteroscedastic uncertainty head. We empirically verify channel specialization: the non-Lambertian uncertainty channel shows r = 0.67 cross-correlation with non-Lambertian residual error, more than 4 times the texture channel's correlation. We further demonstrate downstream utility: filtering out the 75% highest-uncertainty pixels reduces reconstruction MSE by 77% on retained pixels, whereas random filtering produces no improvement. The specialization also holds on out-of-distribution real photographs. We report negative results for a more elaborate variant combining frequency decomposition, cross-task supervision, evidential learning, contrastive loss, and test-time adaptation. Our method reaches 15.98 plus or minus 0.41 dB R_PSNR, within 0.8 dB of a 5-member Deep Ensemble at one-fifth the cost, with the unique capability of source-separated uncertainty.", "date": "2026-05-07", "categories": []}
{"id": "2605.04234", "title": "Disentangled Learning Improves Implicit Neural Representations for Medical Reconstruction", "abstract": "Implicit neural representations (INRs) have emerged as a powerful paradigm for medical imaging via physics-informed unsupervised learning. Classical INRs optimize an entire network from scratch for each subject, leading to inefficient training and suboptimal imaging quality. Recent initialization-based approaches attempt to inject population priors into pre-trained networks, yet they rely on high-quality images and often suffer from catastrophic forgetting during fine-tuning. We present DisINR, a novel INR framework that explicitly disentangles shared and subject-specific representations. DisINR introduces a shared encoder-decoder pair and subject-specific encoders, whose features are jointly decoded for image reconstruction. By integrating differentiable forward models, it pre-trains the shared modules directly from limited raw measurements, removing the need for pre-acquired high-quality images. During test-time adaptation, only the subject-specific encoder is optimized, while the shared pair remains frozen, effectively preserving learned priors. Extensive evaluations on three representative medical imaging tasks show that DisINR significantly outperforms state-of-the-art INRs in both reconstruction accuracy and efficiency.", "date": "2026-05-05", "categories": []}
{"id": "2605.04581", "title": "GTF: Omnidirectional EPI Transformer for Light Field Super-Resolution", "abstract": "Light field (LF) image super-resolution benefits from Epipolar Plane Images (EPIs), whose line slopes explicitly encode disparity. However, existing Transformer-based LF SR methods mainly attend to horizontal and vertical EPIs, leaving diagonal epipolar geometry underexplored. We present GTF, an omnidirectional EPI Transformer that explicitly models horizontal, vertical, 45-degree, and 135-degree EPIs within a unified reconstruction framework. GTF combines directional EPI processing, MacPI-based prior injection, adaptive directional fusion, and a topology-preserving feed-forward network to better exploit LF geometry. For the NTIRE 2026 fidelity tracks, we use GTF as the main model, while a lightweight GTF-Tiny variant targets the efficiency track. On five standard LF SR benchmarks covering both real-captured and synthetic scenes, GTF reaches 32.78 dB without inference-time enhancement, and stronger inference settings with EPSW and test-time augmentation further improve performance. Under the NTIRE 2026 efficiency constraint, GTF-Tiny attains 32.57 dB with only 0.915M parameters and 19.81 GFLOPs. In the NTIRE 2026 Light Field Image Super-Resolution Challenge, our submissions rank 3rd on Track 1 and Track 3 and 4th on Track 2. Architecture-evolution, channel-width, and inference analyses further support the effectiveness of diagonal EPI modeling, directional fusion, and the lightweight design.", "date": "2026-05-06", "categories": []}
{"id": "2605.00527", "title": "Multi-frame Restoration for High-rate Lissajous Confocal Laser Endomicroscopy", "abstract": "Lissajous confocal laser endomicroscopy (CLE) is a promising solution for high speed in vivo optical biopsy for handheld scenarios. However, Lissajous scanning traces a resonant trajectory and samples only the visited pixels per frame; at high frame rates, many pixels remain unvisited, creating structured holes. In this work, we introduce the first benchmark for high-rate Lissajous CLE, consisting of low-quality video clips paired with high-quality reference images. The reference images are wide-FOV mosaics obtained by stitching stabilized, slow-scan frames of the same tissue, enabling temporally aligned supervision. Using this dataset, we propose MIRA, a lightweight recurrent framework for Lissajous CLE restoration that iteratively aggregates temporal context through feature reuse and displacement alignment. Our experiments demonstrate that MIRA outperforms both lightweight and high-complexity baselines in restoration quality while maintaining a favorable computational efficiency suitable for clinical deployment.", "date": "2026-05-01", "categories": []}
{"id": "2605.03626", "title": "RPBA-Net: An Interpretable Residual Pyramid Bilateral Affine Network for RAW-Domain ISP Enhancement", "abstract": "To address module fragmentation, uninterpretable mappings, and deployment constraints in RAW-domain demosaicing, color correction, and detail enhancement, this paper proposes RPBA-Net, an interpretable residual pyramid bilateral affine network for RAW-domain ISP enhancement. Given packed RAW as input, the method performs residual affine base reconstruction by estimating a base RGB representation and learning identity-guided residual affine corrections, thereby unifying demosaicing and enhancement. It further builds pyramid bilateral affine grids and combines guide-driven autoregressive adaptive slicing with adaptive cross-layer fusion to hierarchically model global tone restoration and local texture enhancement. In addition, smoothness, cross-scale consistency, and magnitude regularization terms are introduced to improve model stability, controllability, and structural interpretability. Extensive experiments demonstrate that RPBA-Net surpasses representative RAW-to-sRGB methods and achieves state-of-the-art performance in reconstruction fidelity and perceptual quality, while maintaining low model complexity and strong deployment potential for mobile and embedded platforms.", "date": "2026-05-05", "categories": []}
{"id": "2605.03837", "title": "Conditions for well-posed color recovery in scattering media", "abstract": "Recovering scene color from images captured in scattering media is a fundamental inverse problem in optical imaging. Yet the problem is intrinsically ill-posed as multiple solutions can explain the same observation, and prediction error cannot be controlled without understanding the space of candidate solutions. Here, we present sufficient conditions under which color recovery in a scattering medium becomes well-posed. Observing that ill-posedness stems from (i) projection of spectral signals onto pixel intensities, and (ii) unknown medium parameters, we demonstrate that sensor improvements alone cannot resolve medium-induced distortions without additional constraints. We identify recovery patterns, cross-pixel relationships that naturally occur in images, and prove, for an ideal hyperspectral camera, that they restrict the solution to a unique candidate. This opens the door to a new class of vision algorithms grounded in first principles, enabling quantitative analysis of images in scattering environments.", "date": "2026-05-05", "categories": []}
{"id": "2407.06150", "title": "PanDORA: Casual HDR Radiance Acquisition of Indoor Scenes for Image-based Lighting", "abstract": "Most novel view synthesis methods -- including Neural Radiance Fields (NeRF) -- struggle to capture the high dynamic range (HDR) radiance required for realistic image-based lighting (IBL). This limitation stems from a reliance on low dynamic range (LDR) imagery, which fails to capture the intensity of light sources found in indoor environments. While exposure bracketing can recover this range, it is often too slow for practical, large-scale acquisition. In this work, we introduce PanDORA: PANoramic Dual-Observer Radiance Acquisition, a system specifically designed for the fast and affordable capture of high-quality HDR radiance maps for IBL. Our approach utilizes two 360{\\deg} cameras mounted on a portable monopod to simultaneously record videos at different exposures. These videos are processed by our proposed two-stage NeRF-based algorithm featuring a novel self-calibrating pipeline to estimate camera parameters. This pipeline produces non-saturated HDR radiance fields that accurately capture the radiance of a scene. When evaluated on a new dataset of real indoor environments featuring HDR ground truth lighting, PanDORA demonstrates superior fidelity in reconstructing the peak intensities necessary for downstream rendering tasks, providing a scalable and efficient solution for capturing real-world IBLs.", "date": "2024-07-08", "categories": []}
{"id": "2503.09559", "title": "Interlaced R2D2 DNN Series for Scalable Non-Cartesian MRI with Sensitivity Self-calibration", "abstract": "We introduce interlaced R2D2 (iR2D2), a DNN series paradigm for scalable image reconstruction from accelerated non-Cartesian k-space acquisitions in MRI with sensitivity map self-calibration. While unrolled DNN architectures provide robust image formation, embedding non-uniform fast Fourier transform operators within the backpropagation graph becomes impractical to train at large scale, e.g., in 2D MRI with a large number of coils, or for higher-dimensional imaging. To address this scalability challenge, we leverage the R2D2 paradigm as a learned version of the Matching Pursuit algorithm that was recently introduced in radio astronomy for fast large-scale Fourier imaging. R2D2's reconstruction is formed as a series of residual images iteratively estimated as outputs of DNN modules taking the previous iteration's data residual as input. Specific to MRI, precomputed sensitivity maps derived from undersampled data can yield an inaccurate measurement operator, which may adversely affect the performance of iterative algorithms such as R2D2. Thus, we extend the R2D2 framework to iR2D2 by introducing a bespoke interlaced architecture that alternates between two R2D2 DNN series to jointly self-calibrate sensitivity maps and form the MR image. We further enhance iR2D2 to operate as an adaptive solver governed by an error-controlled update condition that enforces a sufficient residual energy descent, a dynamic capability fundamentally incompatible with the predefined forward passes of unrolled architectures. Extensive experiments in simulation and on real data, targeting undersampled radial k-space sampling, demonstrate that iR2D2 significantly improves upon R2D2 and outperforms state-of-the-art benchmarks, delivering scalable, high-fidelity imaging with corrected sensitivity profiles.", "date": "2025-03-12", "categories": []}
{"id": "2508.09179", "title": "HiFi-Mamba: Dual-Stream W-Laplacian Enhanced Mamba for High-Fidelity MRI Reconstruction", "abstract": "Reconstructing high-fidelity MR images from undersampled k-space data remains a challenging problem in MRI. While Mamba variants for vision tasks offer promising long-range modeling capabilities with linear-time complexity, their direct application to MRI reconstruction inherits two key limitations: (1) insensitivity to high-frequency anatomical details; and (2) reliance on redundant multi-directional scanning. To address these limitations, we introduce High-Fidelity Mamba (HiFi-Mamba), a novel dual-stream Mamba-based architecture comprising stacked W-Laplacian (WL) and HiFi-Mamba blocks. Specifically, the WL block performs fidelity-preserving spectral decoupling, producing complementary low- and high-frequency streams. This separation enables the HiFi-Mamba block to focus on low-frequency structures, enhancing global feature modeling. Concurrently, the HiFi-Mamba block selectively integrates high-frequency features through adaptive state-space modulation, preserving comprehensive spectral details. To eliminate the scanning redundancy, the HiFi-Mamba block adopts a streamlined unidirectional traversal strategy that preserves long-range modeling capability with improved computational efficiency. Extensive experiments on standard MRI reconstruction benchmarks demonstrate that HiFi-Mamba consistently outperforms state-of-the-art CNN-based, Transformer-based, and other Mamba-based models in reconstruction accuracy while maintaining a compact and efficient model design.", "date": "2025-08-07", "categories": []}
{"id": "2510.06516", "title": "Limited-Angle Tomography Reconstruction via Projector Guided 3D Diffusion", "abstract": "Limited-angle electron tomography aims to reconstruct 3D shapes from 2D projections of Transmission Electron Microscopy (TEM) within a restricted range and number of tilting angles, but it suffers from the missing-wedge problem that causes severe reconstruction artifacts. Deep learning approaches have shown promising results in alleviating these artifacts, yet they typically require large high-quality training datasets with known 3D ground truth which are difficult to obtain in electron microscopy. To address these challenges, we propose TEMDiff, a novel 3D diffusion-based iterative reconstruction framework. Our method is trained on readily available volumetric FIB-SEM data using a simulator that maps them to TEM tilt series, enabling the model to learn realistic structural priors without requiring clean TEM ground truth. By operating directly on 3D volumes, TEMDiff implicitly enforces consistency across slices without the need for additional regularization. On simulated electron tomography datasets with limited angular coverage, TEMDiff outperforms state-of-the-art methods in reconstruction quality. We further demonstrate that a trained TEMDiff model generalizes well to real-world TEM tilts obtained under different conditions and can recover accurate structures from tilt ranges as narrow as 8 degrees, with 2-degree increments, without any retraining or fine-tuning.", "date": "2025-10-07", "categories": []}
{"id": "2605.00878", "title": "Single Image Defogging Using a Fourth-Order Telegraph PDE Guided by Physical Haze Modeling", "abstract": "In real-world scenarios, image defogging is an inverse problem due to unknown scene depth, atmospheric scattering, and the common absence of ground truth . To resolve the issue, we propose a hybrid defogging model that integrates a fourth-order nonlinear PDE with a physical haze formation model. We used Dark Channel Prior to estimate atmospheric parameters and to generate a guidance image, while the final restoration is performed via a fourth-order PDE-based evolution. A fourth-order PDE of the type telegraph is then evolved, incorporating an edge-adaptive diffusion coefficient and a fidelity term weighted by the transmission map. Fourth-order diffusion effectively suppresses haze while preserving structural details, and the hyperbolic formulation improves numerical stability and convergence behavior. We use relative error norm criteria for the convergence of our PDE. The proposed method is compared with Dark Channel prior, modified Dark Channel prior, and variational-based single-image defogging techniques. When we have ground truth available, we use MSE and SSIM for quantitative evaluation, whereas no-reference metrics, including FADE, Contrast Restoration Index, Average Gradient, and Entropy, are applied to real-world foggy images. Experimental results demonstrate that the proposed hybrid PDE-based method provides comparable visual quality and maintains structural details.", "date": "2026-04-26", "categories": []}
{"id": "2605.00885", "title": "Multi-Branch Non-Homogeneous Image Dehazing via Concentration Partitioning and Image Fusion", "abstract": "Existing single image dehazing methods have demonstrated satisfactory performance on homogeneous thin-haze images; however, they often struggle with non-homogeneous hazy images that exhibit spatially varying haze concentrations and abrupt density transitions across different regions. To address this fundamental limitation, we propose a novel multi-branch deep neural network framework, termed Concentration Partitioning and Image Fusion Network (CPIFNet), which decomposes the challenging non-homogeneous dehazing problem into a set of tractable homogeneous sub-problems. Our key insight is that a single non-homogeneous hazy image can be viewed as a composite of multiple local regions, each exhibiting approximately homogeneous haze characteristics. CPIFNet employs a two-stage architecture consisting of an Image Enhancement Network (IENet) stage and an Image Fusion Network (IFNet) stage. In the first stage, multiple IENet branches are independently trained on homogeneous haze datasets of different concentration levels, producing enhancement models that excel at restoring regions matching their respective haze densities. In the second stage, the IFNet intelligently aggregates the advantageous regions from all enhancement outputs through deep feature stacking and merging, yielding a unified high-quality dehazed result. Furthermore, we introduce a comprehensive loss function incorporating reconstruction, perceptual, structural, and color losses to jointly supervise both stages.", "date": "2026-04-27", "categories": []}
{"id": "2605.00901", "title": "RA-CMF: Region-Adaptive Conditional MeanFlow for CT Image Reconstruction", "abstract": "The use of CT imaging is important for screening, diagnosis, therapy planning, and prognosis of lung cancers. Unfortunately, due to differences in imaging protocols and scanner models, CT images acquired by different means may show large differences in noise statistics, contrast, and texture. In this study, we develop a novel conditional MeanFlow pipeline for CT image reconstruction. We introduce a conditional MeanFlow network that models the enhancement trajectory by predicting image-conditioned flow fields given intermediate image states. The image enhancement network is trained with a MeanFlow consistency loss along with the image reconstruction loss. In order to provide an adaptive refinement process in terms of spatial location of enhancements, we integrate a regional reinforcement learning-driven policy network into our approach. The policy network receives information about the MeanFlow rollouts and provides predictions in terms of tile-wise refinement budgets, stopping criteria, and total budget allocation of enhancement processes. Our policy network is trained through reinforcement learning in a policy gradient framework, where the goal of the training reward is to maximize improvement of enhancements while minimizing unnecessary computations and avoiding instabilities. In this way, our approach combines conditional flow-based enhancement with reinforcement learning-based spatial enhancement control. This allows our approach to focus more attention on enhancing difficult areas while stabilizing areas already showing sufficient quality. Our results show high accuracy in the tumor ROI, with the average radiomic feature CCC being 0.96, an average PSNR of 31.30 $\\pm$ 4.16, and average SSIM of 0.94 $\\pm$ 0.07. Moreover, there is an improvement in the overall quality of images, with an average PSNR of 34.23 $\\pm$ 1.71 and average SSIM of 0.95 $\\pm$ 0.01.", "date": "2026-04-28", "categories": []}
{"id": "2605.01122", "title": "Machine Learning-Augmented Acceleration of Iterative Ptychographic Reconstruction", "abstract": "Iterative ptychographic reconstruction algorithms are widely used for coherent diffractive imaging but can exhibit slow convergence under realistic experimental conditions. We propose a machine learning-augmented approach that accelerates iterative ptychographic reconstruction by introducing a learned fast-forward operator applied during reconstruction. Following an initial warm-up using standard iterations, the fast-forward operator advances the reconstruction toward a more converged state, after which conventional iterative updates are resumed. This strategy preserves the physical consistency and flexibility of established ptychographic solvers while reducing the number of iterations required for convergence. The model is trained on diverse ptychographic datasets and evaluated on experimental data acquired in a different year, demonstrating robustness and temporal generalization. Compared with conventional iterative solvers, the machine learning-augmented method achieves comparable reconstruction quality while converging faster in terms of Poisson negative log-likelihood, yielding over a two-fold reduction in wall-clock time. The approach has been integrated into an existing reconstruction pipeline and deployed in production at a synchrotron beamline, demonstrating practicality for real-time experimental operation.", "date": "2026-05-01", "categories": []}
{"id": "2605.01185", "title": "Phase-map synthesis from magnitude-only MR images using conditional score-based diffusion models with application in training of accelerated MRI reconstruction models", "abstract": "Accelerated magnetic resonance imaging (MRI) enabled by the training of deep learning (DL)-based image recon. models requires large and diverse raw k-space datasets. In most clinical MRI applications, due to storage and patient privacy concerns, raw k-space data is discarded and magnitude-only images are the only component saved. Consequently, a large portion of the DL-based MRI recon. literature has either relied on small training datasets or has used one of the few available open-source k-space datasets. At the same time, the growing number of anonymized magnitude-only image registries/databases motivates the development of techniques that can use them as training datasets for generalizable DL-based recon. models. Here we propose to address this challenge by employing a generative approach based on conditional score-based diffusion models (SBDMs): given a magnitude-only MR image, it synthesizes a phase map (in the image domain) that realistically corresponds to the magnitude-only image. We evaluate its generative capabilities in a downstream DL-based recon. task whereby a large k-space dataset is generated by combining the SBDM-synthesized phase-maps and the corresponding magnitude-only images, and this k-space dataset is then used to train a DL model for accelerated MRI recon. We compare the performance of the resulting DL model versus those trained according to (a) a naive approach that uses smooth phase, (b) a k-space training dataset generated using synthesized phase maps derived from a generative adversarial network, and (c) the ground truth k-space data. Our results suggest that the DL model trained from SBDM-synthesized k-space data outperforms the other approaches in terms of quantitative metrics as well as qualitatively observed recon. fidelity, i.e., whether the reconstructed images include erroneous or hallucinated features that could adversely impact diagnostic accuracy.", "date": "2026-05-02", "categories": []}
{"id": "2605.02464", "title": "ExpoCM: Exposure-Aware One-Step Generative Single-Image HDR Reconstruction", "abstract": "Single-image HDR reconstruction aims to recover high dynamic range radiance from a single low dynamic range (LDR) input, but remains highly ill-posed due to detail saturation in over-exposed regions and noise amplification in under-exposed areas. While recent diffusion-based approaches offer powerful generative priors, they often overlook the exposure-dependent nature of the degradation and incur substantial computational costs from iterative sampling. To address these challenges, we propose ExpoCM, a novel one-step generative HDR reconstruction framework that reformulates HDR reconstruction as a Probability Flow ODE (PF-ODE) and constructs exposure-aware consistency trajectories via exposure-dependent perturbations. Specifically, a soft exposure mask is first constructed to separate the LDR image into over-, under-, and well-exposed regions. Based on this partition, region-conditioned consistency trajectories are designed to hallucinate saturated details, suppress noise in dark regions, and preserve reliable structures within a single, distillation-free inference step. To further enhance perceptual quality, we introduce an Exposure-guided Luminance-Chromaticity Loss in the CIE~$\\text{L}^*\\text{a}^*\\text{b}^*$ space, which assigns exposure-aware weights to luminance and chromaticity components, effectively mitigating brightness bias and color drift. Extensive experiments on the HDR-REAL, HDR-EYE, and AIM2025 benchmarks demonstrate that ExpoCM achieves state-of-the-art fidelity and perceptual accuracy, while enabling over 400$\\times$ and 20$\\times$ faster inference compared to DDPM (1000 steps) and DDIM (50 steps), respectively.", "date": "2026-05-04", "categories": []}
{"id": "2605.02627", "title": "Rethinking Low-Light Image Enhancement: A Log-Domain Intensity--Chromaticity Decoupling Perspective", "abstract": "Explicit reconstruction constraints derived from the decoupled representation are further imposed to suppress abnormal channel amplification and chromatic noise. Experiments on LOLv2-Real, MIT-Adobe FiveK, and LSRW show that the proposed method achieves competitive or superior quantitative and visual performance, reaching 29.71 dB PSNR and 0.89 SSIM on LOLv2-Real. DarkFace experiments further indicate improved downstream face detection under low-light conditions. Code and pretrained models are available at: https://github.com/mubaisam/ICD.", "date": "2026-05-04", "categories": []}
{"id": "2605.02767", "title": "TOC-SR: Task-Optimal Compact diffusion for Image Super Resolution", "abstract": "Diffusion models have recently demonstrated strong performance for image restoration tasks, including super-resolution. However, their large model size and iterative sampling procedures make them computationally expensive for practical deployment. In this work, we present TOC-SR, a framework for building efficient one-step super-resolution models by first discovering a compact diffusion backbone. Starting from a sixteen-channel latent diffusion model, we construct parameter-efficient surrogate blocks using feature-wise generative distillation and perform architecture discovery using epsilon-constrained Bayesian Optimization to minimize model complexity while preserving generative fidelity. The resulting compact diffusion backbone achieves a 6.6x reduction in parameters and a 2.8x reduction in GMACs compared to the expanded diffusion model. We then adapt this backbone for super-resolution and distill the diffusion process into a single-step generator. Experiments demonstrate that the proposed approach enables efficient super-resolution while maintaining strong reconstruction quality.", "date": "2026-05-04", "categories": []}
{"id": "2602.17205", "title": "Deeper detection limits in astronomical imaging using self-supervised spatiotemporal denoising", "abstract": "The detection limit of astronomical imaging observations is limited by several noise sources. Some of that noise is correlated between neighbouring image pixels and exposures, so in principle could be learned and corrected. We present an astronomical self-supervised transformer-based denoising algorithm (ASTERIS), that integrates spatiotemporal information across multiple exposures. Benchmarking on mock data indicates that ASTERIS improves detection limits by 1.0 magnitude at 90% completeness and purity, while preserving the point spread function and photometric accuracy. Observational validation using data from the James Webb Space Telescope (JWST) and Subaru telescope identifies previously undetectable features, including low-surface-brightness galaxy structures and gravitationally-lensed arcs. Applied to deep JWST images, ASTERIS identifies three times more redshift > 9 galaxy candidates, with rest-frame ultraviolet luminosity 1.0 magnitude fainter, than previous methods.", "date": "2026-02-19", "categories": []}
{"id": "2604.21960", "title": "Conditional Diffusion Posterior Alignment for Sparse-View CT Reconstruction", "abstract": "Computed Tomography (CT) is a widely used imaging modality in medical and industrial applications. To limit radiation exposure and measurement time, there is a growing interest in sparse-view CT, where the number of projection views is significantly reduced. Deep neural networks have shown great promise in improving reconstruction quality in sparse-view CT, especially generative diffusion models. However, these methods struggle to scale to large 3D volumes due to several reasons: (i) the high memory and computational requirements of 3D models, (ii) the lack of large 3D training datasets, and (iii) the inconsistencies across slices when using 2D models independently on each slice. We overcome these limitations and scale diffusion-based sparse-view CT reconstruction to large 3D volumes by combining conditional diffusion with explicit data consistency. We propose Conditional Diffusion Posterior Alignment (CDPA) to enable scalable 3D sparse-view CT reconstruction. A 2D U-Net diffusion model is conditioned on an initial 3D reconstruction to improve inter-slice consistency, combined with data-consistency alignment to match measured projections. Experiments on synthetic and real Cone Beam CT (CBCT) data show state-of-the-art performance, with ablations that confirm the synergistic effects of the proposed pipeline. Finally, we show that the same principles also strengthen fast denoising U-Nets, yielding near-diffusion quality at a fraction of the computational cost.", "date": "2026-04-23", "categories": []}
{"id": "2605.00029", "title": "Broadband Wide Field of View Imaging with Computational Mirrors", "abstract": "Traditional glass-based optics are typically optimized for narrow spectral bands, such as the visible (400-700nm) or shortwave infrared (1000-1800nm). While the emergence of VIS-SWIR sensors (400-1700nm) offers transformative potential, refractive optics struggle to focus this entire range simultaneously. Mirrors represent a promising achromatic alternative; however, they are often sidelined by field curvature, and off-axis aberrations. This paper introduces Computational Mirrors, a framework that enables high-resolution, wide-field-of-view imaging across the complete VIS-SWIR spectrum using a single sensor. Our method is built on the observation that distinct regions of the field of view reach focus at varying distances from the mirror. By capturing a minimal focal stack (2-4 images), we utilize a computational backend to recover a sharp, all-in-focus image. A key contribution of this work is SeidelConv, a novel, physics-inspired, spatially-varying point spread function (PSF) model designed to accurately characterize and correct the off-axis aberrations inherent in simple concave mirrors. We demonstrate the efficacy of our approach using a first-of-its-kind 50mm F/1 optical system equipped with a VIS-SWIR sensor. Our system produces sharp images across RGB, NIR, and SWIR wavelengths without requiring refocusing, revealing material details invisible within individual spectral bands. We further validate the scalability of our approach with a 100mm F/2 system optimized for long-range imaging.", "date": "2026-04-22", "categories": []}
{"id": "2602.21361", "title": "Towards single-shot coherent imaging via overlap-free ptychography", "abstract": "Ptychographic imaging at synchrotron and XFEL sources requires dense overlapping scans, limiting throughput and increasing dose. Extending coherent diffractive imaging to overlap-free operation on extended samples remains an open problem. Here, we extend PtychoPINN (O. Hoidn \\emph{et al.}, \\emph{Scientific Reports} \\textbf{13}, 22789, 2023) to deliver \\emph{overlap-free, single-shot} reconstructions in a Fresnel coherent diffraction imaging (CDI) geometry while also accelerating conventional multi-shot ptychography. The framework couples a differentiable forward model of coherent scattering with a Poisson photon-counting likelihood; real-space overlap enters as a tunable parameter via coordinate-based grouping rather than a hard requirement. On synthetic benchmarks, reconstructions remain accurate at low counts ($\\sim\\!10^4$ photons/frame), and overlap-free single-shot reconstruction with an experimental probe reaches amplitude structural similarity (SSIM) 0.904, compared with 0.968 for overlap-constrained reconstruction. Against a data-saturated supervised model with the same backbone (16,384 training images), PtychoPINN achieves higher SSIM with only 1,024 images and generalizes to unseen illumination profiles. Per-graphics processing unit (GPU) throughput is approximately $40\\times$ that of least-squares maximum-likelihood (LSQ-ML) reconstruction at matched $128\\times128$ resolution. These results, validated on experimental data from the Advanced Photon Source and the Linac Coherent Light Source, unify single-exposure Fresnel CDI and overlapped ptychography within one framework, supporting dose-efficient, high-throughput imaging at modern light sources.", "date": "2026-02-24", "categories": []}
{"id": "2604.27653", "title": "FUN: A Focal U-Net Combining Reconstruction and Object Detection for Snapshot Spectral Imaging", "abstract": "Conventional push-broom hyperspectral imaging suffers from slow acquisition speeds, precluding real-time object detection; in contrast, snapshot spectral imaging enables instantaneous hyperspectral images (HSIs) capture, making real-time object detection feasible, yet its potential is often compromised by time-consuming post-capture reconstruction. To address this issue, we propose the Focal U-shaped Network (FUN), a novel end-to-end framework that jointly performs HSI reconstruction and object detection via multi-task learning. FUN employs a shared U-shaped backbone, where reconstruction provides underlying spectral information while detection guides semantic-aware priors learning, facilitating mutually beneficial task interaction. Crucially, we introduce focal modulation, an efficient alternative to self-attention that modulates spatial and spectral features while reducing quadratic computational complexity, enabling a self-attention-free architecture for joint reconstruction and detection. Furthermore, we contribute a new HSI object detection dataset with 8712 annotated objects across 363 HSIs to facilitate evaluation of the proposed method. Experiments demonstrate that FUN achieves state-of-the-art performance on both tasks, using 40% fewer parameters and 30% less computation than recent alternatives, making it promising for future real-time edge deployment. The code and datasets are available: https://github.com/ShawnDong98/FUN.", "date": "2026-04-30", "categories": []}
{"id": "2604.27702", "title": "RayFormer: Modeling Inter- and Intra-Ray Similarity for NeRF-Based Video Snapshot Compressive Imaging", "abstract": "Video snapshot compressive imaging (SCI) enables the reconstruction of dynamic scenes from a single snapshot measurement. Recently, NeRF-based methods have shown promising reconstruction performance. However, such methods typically adopt random ray sampling strategies and fail to capture content structural similarities, resulting in limited reconstruction quality. To address these issues, we first propose a patch-level ray sampling strategy to enable the modeling of content structure. Then, we propose an Inter- and Intra-Ray Transformer (RayFormer) to capture the structural similarities, modeling both inter-ray similarities among spatially neighboring points at the same depth and intra-ray correlations between adjacent points along the viewing ray. Finally, benefiting from the patch-level sampling strategy, the total variation prior is incorporated into the objective function to enhance spatial smoothness and suppress artifacts. Experiments in both simulated and real-world scenes demonstrate that the proposed method achieves state-of-the-art (SOTA) reconstruction performance.", "date": "2026-04-30", "categories": []}
{"id": "2512.23726", "title": "q3-MuPa: Quick, Quiet, Quantitative Multi-Parametric MRI using Physics-Informed Diffusion Models", "abstract": "The 3D fast silent multi-parametric mapping sequence with zero echo time (MuPa-ZTE) is a novel quantitative MRI (qMRI) acquisition that enables nearly silent scanning by using a 3D phyllotaxis sampling scheme. MuPa-ZTE improves patient comfort and motion robustness, and generates quantitative maps of T1, T2, and proton density using the acquired weighted image series. In this work, we propose a diffusion model-based qMRI mapping method that leverages both a deep generative model and physics-based data consistency to further improve the mapping performance. Furthermore, our method enables additional acquisition acceleration, allowing high-quality qMRI mapping from a fourfold-accelerated MuPa-ZTE scan (approximately 1 minute). Specifically, we trained a denoising diffusion probabilistic model (DDPM) to map MuPa-ZTE image series to qMRI maps, and we incorporated the MuPa-ZTE forward signal model as an explicit data consistency (DC) constraint during inference. We compared our mapping method against a baseline dictionary matching approach and a purely data-driven diffusion model. The diffusion models were trained entirely on synthetic data generated from digital brain phantoms, eliminating the need for large real-scan datasets. We evaluated on synthetic data, a NISM/ISMRM phantom, healthy volunteers, and a patient with brain metastases. The results demonstrated that our method produces 3D qMRI maps with high accuracy, reduced noise and better preservation of structural details. Notably, it generalised well to real scans despite training on synthetic data alone. The combination of the MuPa-ZTE acquisition and our physics-informed diffusion model is termed q3-MuPa, a quick, quiet, and quantitative multi-parametric mapping framework, and our findings highlight its strong clinical potential.", "date": "2025-12-19", "categories": []}
{"id": "2604.26664", "title": "Circular Phase Representation and Geometry-Aware Optimization for Ptychographic Image Reconstruction", "abstract": "Traditional iterative reconstruction methods are accurate but computationally expensive, limiting their use in high-throughput and real-time ptychography. Recent deep learning approaches improve speed, but often predict phase as a Euclidean scalar despite its $2\\pi$ periodicity, which can introduce wrapping artifacts, discontinuities at $\\pm\\pi$, and a mismatch between the loss and the underlying signal geometry. We present a deep learning framework for ptychographic reconstruction that models phase on the unit circle using cosine and sine components. Phase error is optimized with a differentiable geodesic loss, which avoids branch-cut discontinuities and provides bounded gradients. The network further incorporates saturation-aware dual-gain input scaling, parallel encoder branches, and three decoders for amplitude, cosine, and sine prediction, together with a composite loss that promotes circular consistency and structural fidelity. Experiments on synthetic and experimental datasets show consistent improvements in both amplitude and phase reconstruction over existing deep learning methods. Frequency-domain analysis further shows better preservation of mid- and high-frequency phase content. The proposed method also provides substantial speedup over iterative solvers while maintaining physically consistent reconstructions.", "date": "2026-04-29", "categories": []}
{"id": "2604.26920", "title": "Color-Encoded Illumination for High-Speed Volumetric Scene Reconstruction", "abstract": "The task of capturing and rendering 3D dynamic scenes from 2D images has become increasingly popular in recent years. However, most conventional cameras are bandwidth-limited to 30-60 FPS, restricting these methods to static or slowly evolving scenes. While overcoming bandwidth limitations is difficult for general scenes, recent years have seen a flurry of computational imaging methods that yield high-speed videos using conventional cameras for specific applications (e.g., motion capture and particle image velocimetry). However, most of these methods require modifications to a camera's optics or the addition of mechanically moving components, limiting them to a single-view high-speed capture. Consequently, these methods cannot be readily used to capture a 3D representation of rapid scene motion. In this paper, we propose a novel method to capture and reconstruct a volumetric representation of a high-speed scene using only unaugmented low-speed cameras. Instead of modifying the hardware or optics of each individual camera, we encode high-speed scene dynamics by illuminating the scene with a rapid, sequential color-coded sequence. This results in simultaneous multi-view capture of the scene, where high-speed temporal information is encoded in the spatial intensity and color variations of the captured images. To construct a high-speed volumetric representation of the dynamic scene, we develop a novel dynamic Gaussian Splatting-based approach that decodes the temporal information from the images. We evaluate our approach on simulated scenes and real-world experiments using a multi-camera imaging setup, showing first-of-a-kind high-speed volumetric scene reconstructions.", "date": "2026-04-29", "categories": []}
{"id": "2511.07743", "title": "UltraGS: Real-Time Physically-Decoupled Gaussian Splatting for Ultrasound Novel View Synthesis", "abstract": "Ultrasound imaging is a cornerstone of non-invasive clinical diagnostics, yet its limited field of view poses challenges for novel view synthesis. We present UltraGS, a real-time framework that adapts Gaussian Splatting to sensorless ultrasound imaging by integrating explicit radiance fields with lightweight, physics-inspired acoustic modeling. UltraGS employs depth-aware Gaussian primitives with learnable fields of view to improve geometric consistency under unconstrained probe motion, and introduces PD Rendering, a differentiable acoustic operator that combines low-order spherical harmonics with first-order wave effects for efficient intensity synthesis. We further present a clinical ultrasound dataset acquired under real-world scanning protocols. Extensive evaluations across three datasets demonstrate that UltraGS establishes a new performance-efficiency frontier, achieving state-of-the-art results in PSNR (up to 29.55) and SSIM (up to 0.89) while achieving real-time synthesis at 64.69 fps on a single GPU. The code and dataset are open-sourced at: https://github.com/Bean-Young/UltraGS.", "date": "2025-11-11", "categories": []}
{"id": "2604.10359", "title": "Multinex: Lightweight Low-light Image Enhancement via Multi-prior Retinex", "abstract": "Low-light image enhancement (LLIE) aims to restore natural visibility, color fidelity, and structural detail under severe illumination degradation. State-of-the-art (SOTA) LLIE techniques often rely on large models and multi-stage training, limiting practicality for edge deployment. Moreover, their dependence on a single color space introduces instability and visible exposure or color artifacts. To address these, we propose Multinex, an ultra-lightweight structured framework that integrates multiple fine-grained representations within a principled Retinex residual formulation. It decomposes an image into illumination and color prior stacks derived from distinct analytic representations, and learns to fuse these representations into luminance and reflectance adjustments required to correct exposure. By prioritizing enhancement over reconstruction and exploiting lightweight neural operations, Multinex significantly reduces computational cost, exemplified by its lightweight (45K parameters) and nano (0.7K parameters) versions. Extensive benchmarks show that all lightweight variants significantly outperform their corresponding lightweight SOTA models, and reach comparable performance to heavy models. Paper page available at https://albrateanu.github.io/multinex.", "date": "2026-04-11", "categories": []}
{"id": "2604.25310", "title": "Rapid tracking through strongly scattering media with physics-informed neuromorphic speckle analysis", "abstract": "This work addresses the critical problem of tracking fast-moving objects through strongly scattering media in a low-light environment. Different from existing approaches that use frame-based cameras with fixed exposure times, which trade off signal-to-noise ratio for temporal resolution, we introduce computational neuromorphic tracking (CNT), a physics-informed framework that combines asynchronous event sensing with task-driven speckle analysis for robust motion estimation. We formulate the neuromorphic speckle aggregation as a spatiotemporal speckle representation, jointly optimizing the temporal and spatial parameters to maximize tracking stability under extreme conditions. Extensive experiments demonstrate that our method enables robust motion tracking of 10x faster motion and under 10x dimmer illumination compared to conventional systems. These improvements significantly broaden the operational regime for tracking through scattering media, providing an efficient and scalable solution for demanding scenarios involving rapid motion and low-light conditions.", "date": "2026-04-28", "categories": []}
{"id": "2604.25489", "title": "Adaptable phase retrieval for coherent transition radiation spectroscopy based on differentiable physics information", "abstract": "Coherent transition radiation (CTR) spectroscopy is a critical diagnostic for characterizing the longitudinal structure of relativistic electron bunches in laser-plasma and conventional accelerators. In practice, recovering the bunch profile from a measured CTR spectrum is an ill-posed phase-retrieval problem. Traditionally, this is addressed using Gerchberg-Saxton (GS)-type iterative algorithms. However, these implementations often rely on explicit inverse propagators, making them difficult to adapt to sophisticated experimental forward models. In this work, we introduce a flexible gradient-based framework for CTR phase retrieval. By leveraging a differentiable forward model, we propose a phase-only gradient descent (GD-Phase) approach that enforces the measured spectral amplitude as a hard constraint while optimizing the Fourier phase under physical real-space priors. Using synthetic CTR spectra spanning multi-peaked and strongly modulated profiles, we benchmark GD-Phase against traditional GS and a real-space amplitude-parametrized gradient descent (GD-Amp) algorithm. Unlike traditional methods, this formulation allows for the seamless inclusion of arbitrary differentiable experimental effects into the reconstruction loop. We demonstrate that this physics-informed approach not only reproduces the fidelity of GS methods but also establishes a robust baseline for incorporating multi-diagnostic constraints and uncertainty quantification. This enables the systematic extension to higher-dimensional, multimodal, and uncertainty-aware diagnostics, facilitating fast and scalable phase retrieval in realistic experimental settings.", "date": "2026-04-28", "categories": []}
{"id": "2508.19068", "title": "Learning Binary Sampling Patterns for Single-Pixel Imaging using Bilevel Optimisation", "abstract": "Single-Pixel Imaging (SPI) enables the reconstruction of objects using a single detector through sequential illuminations with structured light patterns. The choice of illumination patterns is critical, particularly in highly undersampled regimes, where it directly determines reconstruction quality and acquisition speed. Instead of relying on handcrafted or fixed patterns, we propose to learn task-specific patterns directly from data. Practical SPI hardware only supports binary patterns, making binary pattern design a necessary consideration. We propose a bilevel optimisation method for learning task-specific binary illumination patterns optimised for applications such as single-pixel fluorescence microscopy. We address the non-differentiable nature of binary optimisation using the Straight-Through Estimator. In addition, we incorporate learned variational regularisation, improving reconstruction quality and robustness. We demonstrate our method on the CytoImageNet microscopy dataset. We show that our learned patterns achieve superior reconstruction performance compared to baseline methods and end-to-end deep learning, particularly in highly undersampled regimes and in scarce-data settings.", "date": "2025-08-26", "categories": []}
{"id": "2601.06166", "title": "B-FIRE: Binning-Free Diffusion Implicit Neural Representation for Hyper-Accelerated Motion-Resolved MRI", "abstract": "Accelerated dynamic volumetric magnetic resonance imaging (4DMRI) is essential for applications relying on motion resolution. Existing 4DMRI produces acceptable artifacts of averaged breathing phases, which can blur and misrepresent instantaneous dynamic information. Recovery of such information requires a new paradigm to reconstruct extremely undersampled non-Cartesian k-space data. We propose B-FIRE, a binning-free diffusion implicit neural representation framework for hyper-accelerated MR reconstruction capable of reflecting instantaneous 3D abdominal anatomy. B-FIRE employs a CNN-INR encoder-decoder backbone optimized using diffusion with a comprehensive loss that enforces image-domain fidelity and frequency-aware constraints. Motion binned image pairs were used as training references, while inference was performed on binning-free undersampled data. Experiments were conducted on a T1-weighted StarVIBE liver MRI cohort, with accelerations ranging from 8 spokes per frame (RV8) to RV1. B-FIRE was compared against direct NuFFT, GRASP-CS, and an unrolled CNN method. Reconstruction fidelity, motion trajectory consistency, and inference latency were evaluated.", "date": "2026-01-07", "categories": []}
{"id": "2604.05500", "title": "CLIP-Guided Data Augmentation for Night-Time Image Dehazing", "abstract": "Nighttime image dehazing faces a more complex degradation pattern than its daytime counterpart, as haze scattering couples with low illumination, non-uniform lighting, and strong light interference. Under limited supervision, this complexity aggravates domain drift and training instability, since target-domain samples are scarce while naively introducing external data may weaken adaptation due to distribution mismatch. This paper presents our solution to the NTIRE 2026 Night Time Image Dehazing Challenge, built as a unified framework that integrates domain-aligned data construction, stage-wise training, and inference-time enhancement. Specifically, a pre-trained CLIP visual encoder screens candidate external samples by similarity to construct training data closer to the target domain. NAFNet is then trained in two stages, first adapting to the target domain and then expanding to broader degradation patterns. At inference time, TLC, x8 self-ensemble, and weighted snapshot fusion are combined to improve output stability. Rather than relying on complex network redesign, the proposed framework offers a practical and effective pipeline for nighttime image dehazing.", "date": "2026-04-07", "categories": []}
{"id": "2604.11468", "title": "Beyond Model Design: Data-Centric Training and Self-Ensemble for Gaussian Color Image Denoising", "abstract": "This paper presents our solution to the NTIRE 2026 Image Denoising Challenge (Gaussian color image denoising at fixed noise level $\\sigma = 50$). Rather than proposing a new restoration backbone, we revisit the performance boundary of the mature Restormer architecture from two complementary directions: stronger data-centric training and more complete Test-Time capability release. Starting from the public Restormer $\\sigma\\!=\\!50$ baseline, we expand the standard multi-dataset training recipe with larger and more diverse public image corpora and organize optimization into two stages. At inference, we apply $\\times 8$ geometric self-ensemble to further release model capacity. A TLC-style local inference wrapper is retained for implementation consistency; however, systematic ablation reveals its quantitative contribution to be negligible in this setting. On the challenge validation set of 100 images, our final submission achieves 30.762 dB PSNR and 0.861 SSIM, improving over the public Restormer $\\sigma\\!=\\!50$ pretrained baseline by up to 3.366 dB PSNR. Ablation studies show that the dominant gain originates from the expanded training corpus and the two-stage optimization schedule, and self-ensemble provides marginal but consistent improvement.", "date": "2026-04-13", "categories": []}
{"id": "2604.23675", "title": "GS-DOT: Gaussian splatting-based image reconstruction for diffuse optical tomography", "abstract": "This work presents GS-DOT, a novel image reconstruction framework based on Gaussian Splatting (GS) for diffuse optical tomography (DOT). Inspired by GS for rendering applications, absorption coefficients are represented as a sparse sum of anisotropic Gaussian primitives optimized to fit measured time-resolved point-spread functions through analytic gradients and Adam optimization. This is the first adaptation of GS algorithms in the photon diffusion regime, where the ray transport function is replaced by the diffusion functions to enable accurate modeling of light transport in highly scattering media. Validation on synthetic tissue models demonstrate high accuracy in localization and quantification of reconstructed absorption maps for both clean and noisy signals. GS-DOT has demonstrated high robustness to noise and showed a huge reduction in memory demand.", "date": "2026-04-26", "categories": []}
{"id": "2604.23709", "title": "ZID-Net: Zero-Inference Diffusion Prior Decoupling Network for Single Image Dehazing", "abstract": "Single image dehazing is often constrained by a trade-off between restoration quality and computational efficiency. While efficient, CNN networks struggle to learn robust priors for dense and non-homogeneous haze. Conversely, diffusion models provide strong generative priors but suffer from severe inference latency and sampling instability. To address these limitations, we propose ZID-Net, a novel framework that explicitly decouples diffusion supervision from feed-forward inference. For efficient inference, we design a frequency-spatial decoupled feed-forward backbone. Within this backbone, a Channel-Spatial Laplacian Mask (CSLM) filters haze-amplified noise to extract purified structural details, while Lightweight Global Context Blocks (LGCBs) establish long-range spatial dependencies to capture the global variations of haze. A Dynamic Feature Arbitration Block (DFAB) then adaptively fuses these semantic and structural features for robust reconstruction. To provide this backbone with physical priors without the inference cost, we introduce a Zero-Inference Prior Propagation Head (ZI-PPH) during training. ZI-PPH leverages a conditional diffusion process to predict residual noise, providing degradation-aware structural supervision to the backbone. By discarding the diffusion branch at test time, ZID-Net integrates diffusion priors into a pure feed-forward architecture for accurate and efficient restoration. ZID-Net achieves 40.75 dB PSNR on the synthetic RESIDE dataset and outperforms existing methods with a 1.13 dB gain on real-world datasets. Additionally, it yields a 3.06 dB PSNR gain on the StateHaze1k remote sensing dataset with an inference time of just 19.35 ms. The project code is available at: https://github.com/XoomitLXH/ZID-Net.", "date": "2026-04-26", "categories": []}
{"id": "2604.24000", "title": "Shared-kernel Wavelet Neural Networks for Poisson Image Reconstruction", "abstract": "The Laplacian operator transforms the image into its Laplacian field, which usually is sparse and satisfies a stable distribution. On the other hand, an image can be uniquely reconstructed from its Laplacian field via solving a Poisson equation with a proper boundary condition. Such uniqueness is mathematically guaranteed. Thanks to these properties, we propose to use the sparse Laplacian field to present the image. We first show that the Laplacian field is sparse and satisfies a stable distribution on hundreds images. Then, we show that the image can be accurately reconstruct from its Laplacian field. For the reconstruction task, we propose a shared-kernel wavelet neural network, which solves the Poisson equation and has three advantages. First, it has less than {\\bf 0.0002M} parameters, which is compact enough for most of devices. Second, it has linear computation complexity, leading to a real-time reconstruction. Third, it achieves higher accuracy than previous methods. Several numerical experiments are conducted to show the effectiveness and efficiency of the sparse Laplacian field and the proposed Poisson solver. The proposed method can be applied in a large range of applications such as image compression, low light enhancement, object tracking, etc.", "date": "2026-04-27", "categories": []}
{"id": "2509.10897", "title": "TV Subgradient-Guided Multi-Source Fusion for Spectral Imaging in Dual-Camera CASSI Systems", "abstract": "Balancing spectral, spatial, and temporal resolutions is a key challenge in spectral imaging. The Dual-Camera Coded Aperture Snapshot Spectral Imaging (DC-CASSI) system alleviates this trade-off but suffers from severely ill-posed reconstruction problems due to its high compression ratio. Existing methods are constrained by scene-specific tuning or excessive reliance on paired training data. To address these issues, we propose a Total Variation (TV) subgradient-guided multi-source fusion framework for DC-CASSI reconstruction, comprising three core components: (1) An end-to-end Single-Disperser CASSI (SD-CASSI) observation model based on the tensor-form Kronecker $\\delta$, which establishes a rigorous mathematical foundation for physical constraints while enabling efficient adjoint operator implementation; (2) An adaptive spatial reference generator that integrates SD-CASSI's physical model and RGB subspace constraint, generating the reference image as reliable spatial prior; (3) A TV subgradient-guided regularization term that encodes local structural directions from the reference image into spectral reconstruction, achieving high-quality fused results. The framework is validated on simulated datasets and real-world datasets. Experimental results demonstrate that it achieves state-of-the-art reconstruction performance and robust noise resilience. This work not only establishes an interpretable theoretical foundation for subgradient-guided fusion but also provides a practical fusion-based paradigm for high-fidelity spectral image reconstruction in DC-CASSI systems. Source code: https://github.com/bestwishes43/ADMM-TVDS.", "date": "2025-09-13", "categories": []}
{"id": "2511.18513", "title": "LRDUN: A Low-Rank Deep Unfolding Network for Efficient Spectral Compressive Imaging", "abstract": "Deep unfolding networks (DUNs) have achieved remarkable success and become the mainstream paradigm for spectral compressive imaging (SCI) reconstruction. Existing DUNs are derived from full-HSI imaging models, where each stage operates directly on the high-dimensional HSI, refining the entire data cube based on the single 2D coded measurement. However, this paradigm leads to computational redundancy and suffers from the ill-posed nature of mapping 2D residuals back to 3D space of HSI. In this paper, we propose two novel imaging models corresponding to the spectral basis and subspace image by explicitly integrating low-rank (LR) decomposition with the sensing model. Compared to recovering the full HSI, estimating these compact low-dimensional components significantly mitigates the ill-posedness. Building upon these novel models, we develop the Low-Rank Deep Unfolding Network (LRDUN), which jointly solves the two subproblems within an unfolded proximal gradient descent (PGD) framework. Furthermore, we introduce a Generalized Feature Unfolding Mechanism (GFUM) that decouples the physical rank in the data-fidelity term from the feature dimensionality in the prior module, enhancing the representational capacity and flexibility of the network. Extensive experiments on simulated and real datasets demonstrate that the proposed LRDUN achieves state-of-the-art (SOTA) reconstruction quality with significantly reduced computational cost.", "date": "2025-11-23", "categories": []}
{"id": "2604.10275", "title": "FastSHADE: Fast Self-augmented Hierarchical Asymmetric Denoising for Efficient inference on mobile devices", "abstract": "Real-time image denoising is essential for modern mobile photography but remains challenging due to the strict latency and power constraints of edge devices. This paper presents FastSHADE (Fast Self-augmented Hierarchical Asymmetric Denoising), a lightweight U-Net-style network tailored for real-time, high-fidelity restoration on mobile GPUs. Our method features a multi-stage architecture incorporating a novel Asymmetric Frequency Denoising Block (AFDB) that decouples spatial structure extraction from high-frequency noise suppression to maximize efficiency, and a Spatially Gated Upsampler (SGU) that optimizes high-resolution skip connection fusion. To address generalization, we introduce an efficient Noise Shifting Self-Augmentation strategy that enhances data diversity without inducing domain shifts. Evaluations on the MAI2021 benchmark demonstrate that our scalable model family establishes a highly efficient speed-fidelity trade-off. Our base FastSHADE-M variant maintains real-time latency (<50 ms on an Adreno 840 GPU) while preserving structural integrity, and our scaled-up FastSHADE-XL establishes a new state-of-the-art for overall image quality, achieving 37.94 dB PSNR.", "date": "2026-04-11", "categories": []}
{"id": "2505.16487", "title": "Generative Prior-Guided Neural Interface Reconstruction for 3D Electrical Impedance Tomography", "abstract": "Reconstructing complex 3D interfaces from indirect measurements remains a grand challenge in scientific computing, particularly for ill-posed inverse problems like Electrical Impedance Tomography (EIT). Traditional shape optimization struggles with topological changes and regularization tuning, while emerging deep learning approaches often compromise physical fidelity or require prohibitive amounts of paired training data. We present a transformative ``solver-in-the-loop'' framework that bridges this divide by coupling a pre-trained 3D generative prior with a rigorous boundary integral equation (BIE) solver. Unlike Physics-Informed Neural Networks (PINNs) that treat physics as soft constraints, our architecture enforces the governing elliptic PDE as a hard constraint at every optimization step, ensuring strict physical consistency. Simultaneously, we navigate a compact latent manifold of plausible geometries learned by a differentiable neural shape representation, effectively regularizing the ill-posed problem through data-driven priors rather than heuristic smoothing. By propagating adjoint shape derivatives directly through the neural decoder, we achieve fast, stable convergence with dramatically reduced degrees of freedom. Extensive experiments on 3D high-contrast EIT demonstrate that this principled hybrid approach yields superior geometric accuracy and data efficiency which is difficult to achieve using traditional methods, establishing a robust new paradigm for physics-constrained geometric discovery.", "date": "2025-05-22", "categories": []}
{"id": "2602.19470", "title": "Physics-informed Active Polarimetric 3D Imaging for Specular Surfaces", "abstract": "3D imaging of specular surfaces remains challenging in real-world scenarios, such as in-line inspection or hand-held scanning, requiring fast and accurate measurement of complex geometries. Optical metrology techniques such as deflectometry achieve high accuracy but typically rely on multi-shot acquisition, making them unsuitable for dynamic environments. Fourier-based single-shot approaches alleviate this constraint, yet their performance deteriorates when measuring surfaces with high spatial frequency structure or large curvature. Alternatively, polarimetric 3D imaging in computer vision operates in a single-shot fashion and exhibits robustness to geometric complexity. However, its accuracy is fundamentally limited by the orthographic imaging assumption. In this paper, we propose a physics-informed deep learning framework for single-shot 3D imaging of complex specular surfaces. Polarization cues provide orientation priors that assist in interpreting geometric information encoded by structured illumination. These complementary cues are processed through a dual-encoder architecture with mutual feature modulation, allowing the network to resolve their nonlinear coupling and directly infer surface normals. The proposed method achieves accurate and robust normal estimation in single-shot with fast inference, enabling practical 3D imaging of complex specular surfaces.", "date": "2026-02-23", "categories": []}
{"id": "2604.19976", "title": "Lucky High Dynamic Range Smartphone Imaging", "abstract": "While the human eye can perceive an impressive twenty stops of dynamic range, smartphone camera sensors remain limited to about twelve stops despite decades of research. A variety of high dynamic range (HDR) image capture and processing techniques have been proposed, and, in practice, they can extend the dynamic range by 3-5 stops for handheld photography. This paper proposes an approach that robustly captures dynamic range using a handheld smartphone camera and lightweight networks suitable for running on mobile devices. Our method operates indirectly on linear raw pixels in bracketed exposures. Every pixel in the final HDR image is a convex combination of input pixels in the neighborhood, adjusted for exposure, and thus avoids hallucination artifacts typical of recent deep image synthesis networks. We validate our system on both synthetic imagery and unseen real bracketed images -- we confirm zero-shot generalization of the method to smartphone camera captures. Our iterative inference architecture is capable of processing an arbitrary number of bracketed input photos, and we show examples from capture stacks containing 3--9 images. Our training process relies only on synthetic captures yet generalizes to unseen real photos from several cameras. Moreover, we show that this training scheme improves other SOTA methods over their pretrained counterparts.", "date": "2026-04-21", "categories": []}
{"id": "2604.19989", "title": "Online CS-based SAR Edge-Mapping", "abstract": "With modern defense applications increasingly relying on inexpensive, small Unmanned Aerial Vehicles (UAVs), a major challenge lies in designing intelligent and computationally efficient onboard Automatic Target Recognition (ATR) algorithms to carry out operational objectives. This is especially critical in Synthetic Aperture Radar (SAR), where processing techniques such as ATR are often carried out post data collection, requiring onboard systems to bear the memory burden of storing the back-scattered signals. To alleviate this high cost, we propose an online, direct, edge-mapping technique which bypasses the image reconstruction step to classify scenes and targets. Furthermore, by reconstructing the scene as an edge-map we inherently promote sparsity, requiring fewer measurements and computational power than classic SAR reconstruction algorithms such as backprojection.", "date": "2026-04-21", "categories": []}
{"id": "2604.20128", "title": "Semi-Supervised Flow Matching for Mosaiced and Panchromatic Fusion Imaging", "abstract": "Fusing a low resolution (LR) mosaiced hyperspectral image (HSI) with a high resolution (HR) panchromatic (PAN) image offers a promising avenue for video-rate HR-HSI imaging via single-shot acquisition, yet its severely ill-posed nature remains a significant challenge. In this work, we propose a novel semi-supervised flow matching framework for mosaiced and PAN image fusion. Unlike previous diffusion-based approaches constrained by specific protocols or handcrafted assumptions, our method seamlessly integrates an unsupervised scheme with flow matching, resulting in a generalizable and efficient generative framework. Specifically, our method follows a two-stage training pipeline. First, we pretrain an unsupervised prior network to produce an initial pseudo HR-HSI. Building on this, we then train a conditional flow matching model to generate the target HR-HSI, introducing a random voting mechanism that iteratively refines the initial HR-HSI estimate, enabling robust and effective fusion. During inference, we employ a conflict-free gradient guidance strategy that ensures spectrally and spatially consistent HR-HSI reconstruction. Experiments on multiple benchmark datasets demonstrate that our method achieves superior quantitative and qualitative performance by a significant margin compared to representative baselines. Beyond mosaiced and PAN fusion, our approach provides a flexible generative framework that can be readily extended to other image fusion tasks and integrated with unsupervised or blind image restoration algorithms.", "date": "2026-04-22", "categories": []}
{"id": "2604.20154", "title": "Maximum Likelihood Reconstruction for Multi-Look Digital Holography with Markov-Modeled Speckle Correlation", "abstract": "Multi-look acquisition is a widely used strategy for reducing speckle noise in coherent imaging systems such as digital holography. By acquiring multiple measurements, speckle can be suppressed through averaging or joint reconstruction, typically under the assumption that speckle realizations across looks are statistically independent. In practice, however, hardware constraints limit measurement diversity, leading to inter-look correlation that degrades the performance of conventional methods. In this work, we study the reconstruction of speckle-free reflectivity from complex-valued multi-look measurements in the presence of correlated speckle. We model the inter-look dependence using a first-order Markov process and derive the corresponding likelihood under a first-order Markov approximation, resulting in a constrained maximum likelihood estimation problem. To solve this problem, we develop an efficient projected gradient descent framework that combines gradient-based updates with implicit regularization via deep image priors, and leverages Monte Carlo approximation and matrix-free operators for scalable computation. Simulation results demonstrate that the proposed approach remains robust under strong inter-look correlation, achieving performance close to the ideal independent-look scenario and consistently outperforming methods that ignore such dependencies. These results highlight the importance of explicitly modeling inter-look correlation and provide a practical framework for multi-look holographic reconstruction under realistic acquisition conditions. Our code is available at: https://github.com/Computational-Imaging-RU/MLE-Holography-Markov.", "date": "2026-04-22", "categories": []}
{"id": "2604.20291", "title": "Efficient INT8 Single-Image Super-Resolution via Deployment-Aware Quantization and Teacher-Guided Training", "abstract": "Efficient single-image super-resolution (SISR) requires balancing reconstruction fidelity, model compactness, and robustness under low-bit deployment, which is especially challenging for x3 SR. We present a deployment-oriented quantized SISR framework based on an extract-refine-upsample design. The student performs most computation in the low-resolution space and uses a lightweight re-parameterizable backbone with PixelShuffle reconstruction, yielding a compact inference graph. To improve quality without significantly increasing complexity, we adopt a three-stage training pipeline: Stage 1 learns a basic reconstruction mapping with spatial supervision; Stage 2 refines fidelity using Charbonnier loss, DCT-domain supervision, and confidence-weighted output-level distillation from a Mamba-based teacher; and Stage 3 applies quantization-aware training directly on the fused deploy graph. We further use weight clipping and BatchNorm recalibration to improve quantization stability. On the MAI 2026 Quantized 4K Image Super-Resolution Challenge test set, our final AIO MAI submission achieves 29.79 dB PSNR and 0.8634 SSIM, obtaining a final score of 1.8 under the target mobile INT8 deployment setting. Ablation on Stage 3 optimization shows that teacher-guided supervision improves the dynamic INT8 TFLite reconstruction from 29.91 dB/0.853 to 30.0003 dB/0.856, while the fixed-shape deployable INT8 TFLite artifact attains 30.006 dB/0.857.", "date": "2026-04-22", "categories": []}
{"id": "2604.20594", "title": "Physics-Informed Conditional Diffusion for Motion-Robust Retinal Temporal Laser Speckle Contrast Imaging", "abstract": "Retinal laser speckle contrast imaging (LSCI) is a noninvasive optical modality for monitoring retinal blood flow dynamics. However, conventional temporal LSCI (tLSCI) reconstruction relies on sufficiently long speckle sequences to obtain stable temporal statistics, which makes it vulnerable to acquisition disturbances and limits effective temporal resolution. A physically informed reconstruction framework, termed RetinaDiff (Retinal Diffusion Model), is proposed for retinal tLSCI that is robust to motion and requires only a few frames. In RetinaDiff, registration based on phase correlation is first applied to stabilize the raw speckle sequence before contrast computation, reducing interframe misalignment so that fluctuations at each pixel primarily reflect true flow dynamics. This step provides a physics prior corrected for motion and a high quality multiframe tLSCI reference. Next, guided by the physics prior, a conditional diffusion model performs inverse reconstruction by jointly conditioning on the registered speckle sequence and the corrected prior. Experiments on data acquired with a retinal LSCI system developed in house show improved structural continuity and statistical stability compared with direct reconstruction from few frames and representative baselines. The framework also remains effective in a small number of extremely challenging cases, where both the direct 5-frame input and the conventional multiframe reconstruction are severely degraded. Overall, this work provides a practical and physically grounded route for reliable retinal tLSCI reconstruction from extremely limited frames. The source code and model weights will be publicly available at https://github.com/QianChen113/RetinaDiff.", "date": "2026-04-22", "categories": []}
{"id": "2604.16794", "title": "Improving Radio Interferometry Imaging by Explicitly Modeling Cross-Domain Consistency in Reconstruction", "abstract": "Radio astronomy plays a crucial role in understanding the universe, particularly within the realm of non-thermal astrophysics. Images of celestial objects are derived from the signals (called visibility) measured by radio telescopes. Such imaging results, called dirty images, contain artifacts due to factors such as sparsity and therefore require reconstruction to improve imaging quality. Existing methods typically restrict reconstruction to a unimodal domain, either to the dirty image after imaging or to the sparse visibility prior to imaging. Focusing solely on each unimodal reconstruction results in the loss of complementary in-context information in either the visibility or image domain, leading to an incomplete modeling of mutual dependency and consistency. To address these challenges, we propose CDCRec, a multimodal radio interferometric data reconstruction method that explicitly models cross-domain consistency. We design a hierarchical multi-task and multi-stage framework to enhance the exploration of interplays between domains during reconstruction. Our experimental results demonstrate that CDCRec improves imaging performance through enhanced cross-domain correlation extraction. In particular, our self-supervised complementary modeling strategy is better than current methods at interferometric domain translations that rely heavily on recovering dense information from constrained source-domain data.", "date": "2026-04-18", "categories": []}
{"id": "2604.16919", "title": "Noise-Adaptive Diffusion Sampling for Inverse Problems Without Task-Specific Tuning", "abstract": "Diffusion models (DMs) have recently shown remarkable performance on inverse problems (IPs). Optimization-based methods can fast solve IPs using DMs as powerful regularizers, but they are susceptible to local minima and noise overfitting. Although DMs can provide strong priors for Bayesian approaches, enforcing measurement consistency during the denoising process leads to manifold infeasibility issues. We propose Noise-space Hamiltonian Monte Carlo (N-HMC), a posterior sampling method that treats reverse diffusion as a deterministic mapping from initial noise to clean images. N-HMC enables comprehensive exploration of the solution space, avoiding local optima. By moving inference entirely into the initial-noise space, N-HMC keeps proposals on the learned data manifold. We provide a comprehensive theoretical analysis of our approach and extend the framework to a noise-adaptive variant (NA-NHMC) that effectively handles IPs with unknown noise type and level. Extensive experiments across four linear and three nonlinear inverse problems demonstrate that NA-NHMC achieves superior reconstruction quality with robust performance across different hyperparameters and initializations, significantly outperforming recent state-of-the-art methods. The code is available at https://github.com/NA-HMC/NA-HMC.", "date": "2026-04-18", "categories": []}
{"id": "2604.16969", "title": "Hyperspectral Unmixing Hierarchies", "abstract": "Unmixing reveals the spatial distribution and spectral details of different constituents, called endmembers, in a hyperspectral image. Because unmixing has limited ground truth requirements, can accommodate mixed pixels, and is closely tied to light propagation, it is a uniquely powerful tool for analyzing hyperspectral images. However, spectral variability inhibits unmixing performance, the proper way to determine the number of endmembers is ambiguous, and the clarity of the endmembers degrades as more are included. Hierarchical structure is a possible solution to all three problems.\n  Here, hierarchical unmixing is defined by imposing a hierarchical abundance sum constraint on Deep Nonnegative Matrix Factorization. Binary Linear Unmixing Tactile Hierarchies (BLUTHs) solve the hierarchical unmixing problem with a simple network architecture. Sparsity modulation unmixing growth tailors the topology of a BLUTH to each scene. The structure imposed by BLUTHs allows endmembers with varying levels of spectral contrast to be revealed, mitigating the challenge of spectral variability.\n  The performance of BLUTHs exceeds state-of-the-art unmixing algorithms on laboratory scenes, particularly with regard to abundance estimation, while their performance remains competitive on remote sensing scenes. In addition, ocean color unmixing by BLUTHs is demonstrated on hyperspectral scenes from the HYPSO and PACE satellites.", "date": "2026-04-18", "categories": []}
{"id": "2604.17727", "title": "Voronoi-guided Bilateral 2D Gaussian Splatting for Arbitrary-Scale Hyperspectral Image Super-Resolution", "abstract": "Most existing hyperspectral image super-resolution methods require modifications for different scales, limiting their flexibility in arbitrary-scale reconstruction. 2D Gaussian splatting provides a continuous representation that is compatible with arbitrary-scale super-resolution. Existing methods often rely on rasterization strategies, which may limit flexible spatial modeling. Extending them to hyperspectral image super-resolution remains challenging, as the task requires adaptive spatial reconstruction while preserving spectral fidelity. This paper proposes GaussianHSI, a Gaussian-Splatting-based framework for arbitrary-scale hyperspectral image super-resolution. We develop a Voronoi-Guided Bilateral 2D Gaussian Splatting for spatial reconstruction. After predicting a set of Gaussian functions to represent the input, it associates each target pixel with relevant Gaussian functions through Voronoi-guided selection. The target pixel is then reconstructed by aggregating the selected Gaussian functions with reference-aware bilateral weighting, which considers both geometric relevance and consistency with low-resolution features. We further introduce a Spectral Detail Enhancement module to improve spectral reconstruction. Extensive experiments on benchmark datasets demonstrate the effectiveness of GaussianHSI over state-of-the-art methods for arbitrary-scale hyperspectral image super-resolution.", "date": "2026-04-20", "categories": []}
{"id": "2604.17734", "title": "Score-Based Matching with Target Guidance for Cryo-EM Denoising", "abstract": "Cryo-electron microscopy (cryo-EM) enables single-particle analysis of biological macromolecules under strict low-dose imaging conditions, but the resulting micrographs often exhibit extremely low signal-to-noise ratios and weak particle visibility. Image denoising is therefore an important preprocessing step for downstream cryo-EM analysis, including particle picking, 2D classification, and 3D reconstruction. Existing cryo-EM denoising methods are commonly trained with pixel-wise or Noise2Noise-style objectives, which can improve visual quality but do not explicitly account for structural consistency required by downstream analysis. In this work, we propose a score-based denoising framework for cryo-EM that learns the clean-data score to recover particle signals while better preserving structural information. Building on this formulation, we further introduce a target-guided variant that incorporates reference-density guidance to stabilize score learning under weak and ambiguous signal conditions. Rather than simply amplifying particle-like responses, our framework better suppresses structured low-frequency background, which improves particle--background separability for downstream analysis. Experiments on multiple cryo-EM datasets show that our score-based methods consistently improve downstream particle picking and produce more structure-consistent 3D reconstructions. Experiments on multiple cryo-EM datasets show that our methods improve downstream particle picking and produce more structure-consistent reconstructions.", "date": "2026-04-20", "categories": []}
{"id": "2604.18032", "title": "CFSR: Geometry-Conditioned Shadow Removal via Physical Disentanglement", "abstract": "Traditional shadow removal networks often treat image restoration as an unconstrained mapping, lacking the physical interpretability required to balance localized texture recovery with global illumination consistency. To address this, we propose CFSR, a multi-modal prior-driven framework that reframes shadow removal as a physics-constrained restoration process. By seamlessly integrating 3D geometric cues with large-scale foundation model semantics, CFSR effectively bridges the 2D-3D domain gap. Specifically, we first map observations into a custom HVI color space to suppress shadow-induced noise and robustly fuse RGB data with estimated depth priors. At its core, our Geometric & Semantic Dual Explicit Guided Attention mechanism utilizes DINO features and 3D surface normals to directly modulate the attention affinity matrix, structurally enforcing physical lighting constraints. To recover severely degraded regions, we inject holistic priors via a frozen CLIP encoder. Finally, our Frequency Collaborative Reconstruction Module (FCRM) achieves an optimal synthesis by decoupling the decoding process. Conditioned on geometric priors, FCRM seamlessly harmonizes the reconstruction of sharp high-frequency occlusion boundaries with the restoration of low-frequency global illumination. Extensive experiments demonstrate that CFSR achieves state-of-the-art performance across multiple challenging benchmarks.", "date": "2026-04-20", "categories": []}
{"id": "2403.18026", "title": "Deep Learning-Enabled Modality Transfer Between Independent Microscopes for High-Throughput Imaging", "abstract": "High-throughput biological imaging is often constrained by a trade-off between acquisition speed and image quality. Fast imaging modalities, such as wide-field fluorescence microscopy, enable large-scale data acquisition but suffer from reduced contrast and resolution, whereas high-resolution techniques, including confocal microscopy or single-molecule localization microscopy-based super-resolution techniques, provide superior image quality at the cost of throughput and instrument time. Here, we present a deep learning-based approach for modality transfer across independent microscopes, enabling the transformation of low-quality images acquired on fast systems into high-quality representations comparable to those obtained using advanced imaging platforms. To achieve this, we employ a generative adversarial network (GAN)-based model trained on paired datasets acquired on physically separate wide-field and confocal microscopes, demonstrating that image quality can be reliably transferred between independent instruments. Quantitative evaluation shows substantial improvement in structural similarity and signal fidelity, with median SSIM and PSNR of 0.94 and 31.87, respectively, compared to 0.83 and 21.48 for the original wide-field images. These results indicate that key structural features can be recovered with high accuracy. Importantly, this approach enables a workflow in which high-throughput imaging can be performed on fast, accessible microscopy systems while preserving the ability to computationally recover high-quality structural information. High-resolution microscopy can then be reserved for targeted validation, reducing acquisition time and improving overall experimental efficiency. Together, our results establish deep learning-enabled modality transfer as a practical strategy for bridging independent microscopy systems and supporting scalable, high-content imaging workflows.", "date": "2024-03-26", "categories": []}
{"id": "2506.13215", "title": "DVP-MVS++: Synergize Depth-Normal-Edge and Harmonized Visibility Prior for Multi-View Stereo", "abstract": "Recently, patch deformation-based methods have demonstrated significant effectiveness in multi-view stereo due to their incorporation of deformable and expandable perception for reconstructing textureless areas. However, these methods generally focus on identifying reliable pixel correlations to mitigate matching ambiguity of patch deformation, while neglecting the deformation instability caused by edge-skipping and visibility occlusions, which may cause potential estimation deviations. To address these issues, we propose DVP-MVS++, an innovative approach that synergizes both depth-normal-edge aligned and harmonized cross-view priors for robust and visibility-aware patch deformation. Specifically, to avoid edge-skipping, we first apply DepthPro, Metric3Dv2 and Roberts operator to generate coarse depth maps, normal maps and edge maps, respectively. These maps are then aligned via an erosion-dilation strategy to produce fine-grained homogeneous boundaries for facilitating robust patch deformation. Moreover, we reformulate view selection weights as visibility maps, and then implement both an enhanced cross-view depth reprojection and an area-maximization strategy to help reliably restore visible areas and effectively balance deformed patch, thus acquiring harmonized cross-view priors for visibility-aware patch deformation. Additionally, we obtain geometry consistency by adopting both aggregated normals via view selection and projection depth differences via epipolar lines, and then employ SHIQ for highlight correction to enable geometry consistency with highlight-aware perception, thus improving reconstruction quality during propagation and refinement stage. Evaluation results on ETH3D, Tanks & Temples and Strecha datasets exhibit the state-of-the-art performance and robust generalization capability of our proposed method.", "date": "2025-06-16", "categories": []}
{"id": "2604.15654", "title": "From Zero to Detail: A Progressive Spectral Decoupling Paradigm for UHD Image Restoration with New Benchmark", "abstract": "Ultra-high-definition (UHD) image restoration poses unique challenges due to the high spatial resolution, diverse content, and fine-grained structures present in UHD images. To address these issues, we introduce a progressive spectral decomposition for the restoration process, decomposing it into three stages: zero-frequency \\textbf{enhancement}, low-frequency \\textbf{restoration}, and high-frequency \\textbf{refinement}. Based on this formulation, we propose a novel framework, \\textbf{ERR}, which integrates three cooperative sub-networks: the zero-frequency enhancer (ZFE), the low-frequency restorer (LFR), and the high-frequency refiner (HFR). The ZFE incorporates global priors to learn holistic mappings, the LFR reconstructs the main content by focusing on coarse-scale information, and the HFR adopts our proposed frequency-windowed Kolmogorov-Arnold Network (FW-KAN) to recover fine textures and intricate details for high-fidelity restoration. To further advance research in UHD image restoration, we also construct a large-scale, high-quality benchmark dataset, \\textbf{LSUHDIR}, comprising 82{,}126 UHD images with diverse scenes and rich content. Our proposed methods demonstrate superior performance across a range of UHD image restoration tasks, and extensive ablation studies confirm the contribution and necessity of each module. Project page: https://github.com/NJU-PCALab/ERR.", "date": "2026-04-17", "categories": []}
{"id": "2604.16201", "title": "DENALI: A Dataset Enabling Non-Line-of-Sight Spatial Reasoning with Low-Cost LiDARs", "abstract": "Consumer LiDARs in mobile devices and robots typically output a single depth value per pixel. Yet internally, they record full time-resolved histograms containing direct and multi-bounce light returns; these multi-bounce returns encode rich non-line-of-sight (NLOS) cues that can enable perception of hidden objects in a scene. However, severe hardware limitations of consumer LiDARs make NLOS reconstruction with conventional methods difficult. In this work, we motivate a complementary direction: enabling NLOS perception with low-cost LiDARs through data-driven inference. We present DENALI, the first large-scale real-world dataset of space-time histograms from low-cost LiDARs capturing hidden objects. We capture time-resolved LiDAR histograms for 72,000 hidden-object scenes across diverse object shapes, positions, lighting conditions, and spatial resolutions. Using our dataset, we show that consumer LiDARs can enable accurate, data-driven NLOS perception. We further identify key scene and modeling factors that limit performance, as well as simulation-fidelity gaps that hinder current sim-to-real transfer, motivating future work toward scalable NLOS vision with consumer LiDARs.", "date": "2026-04-17", "categories": []}
{"id": "2410.08329", "title": "Survey of Deep Learning and Physics-Based Approaches in Computational Wave Imaging", "abstract": "Computational wave imaging (CWI) extracts hidden structure and physical properties of a volume of material by analyzing wave signals that traverse that volume. Applications include seismic exploration of the Earth's subsurface, acoustic imaging and non-destructive testing in material science, and ultrasound computed tomography in medicine. Current approaches for solving CWI problems can be divided into two categories: those rooted in traditional physics, and those based on deep learning. Physics-based methods stand out for their ability to provide high-resolution and quantitatively accurate estimates of acoustic properties within the medium. However, they can be computationally intensive and are susceptible to ill-posedness and nonconvexity typical of CWI problems. Machine learning-based computational methods have recently emerged, offering a different perspective to address these challenges. Diverse scientific communities have independently pursued the integration of deep learning in CWI. This review discusses how contemporary scientific machine-learning (ML) techniques, and deep neural networks in particular, have been developed to enhance and integrate with traditional physics-based methods for solving CWI problems.\n  We present a structured framework that consolidates existing research spanning multiple domains, including computational imaging, wave physics, and data science. This study concludes with important lessons learned from existing ML-based methods and identifies technical hurdles and emerging trends through a systematic analysis of the extensive literature on this topic.", "date": "2024-10-10", "categories": []}
{"id": "2602.07069", "title": "Bird-SR: Bidirectional Reward-Guided Diffusion for Real-World Image Super-Resolution", "abstract": "Powered by multimodal text-to-image priors, diffusion-based super-resolution excels at synthesizing intricate details; however, models trained on synthetic low-resolution (LR) and high-resolution (HR) image pairs often degrade when applied to real-world LR images due to significant distribution shifts. We propose Bird-SR, a bidirectional reward-guided diffusion framework that formulates super-resolution as trajectory-level preference optimization via reward feedback learning (ReFL), jointly leveraging synthetic LR-HR pairs and real-world LR images. For structural fidelity easily affected in ReFL, the model is directly optimized on synthetic pairs at early diffusion steps, which also facilitates structure preservation for real-world inputs under smaller distribution gap in structure levels. For perceptual enhancement, quality-guided rewards are applied to both synthetic and real LR images at the later trajectory phase. To mitigate reward hacking, the rewards for synthetic results are formulated in a relative advantage space bounded by their ground-truth counterparts, while real-world optimization is regularized via a semantic alignment constraint. Furthermore, to balance structural and perceptual learning, we introduce a dynamic fidelity-perception weighting strategy that emphasizes structure preservation at early stages and progressively shifts focus toward perceptual optimization at later diffusion steps. Extensive experiments on real-world SR benchmarks demonstrate that Bird-SR consistently outperforms state-of-the-art methods in perceptual quality while preserving structural consistency, validating its effectiveness for real-world super-resolution. Our code can be obtained at https://github.com/fanzh03/Bird-SR.", "date": "2026-02-05", "categories": []}
{"id": "2604.14558", "title": "The Fourth Challenge on Image Super-Resolution ($\\times$4) at NTIRE 2026: Benchmark Results and Method Overview", "abstract": "This paper presents the NTIRE 2026 image super-resolution ($\\times$4) challenge, one of the associated competitions of the NTIRE 2026 Workshop at CVPR 2026. The challenge aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs generated through bicubic downsampling with a $\\times$4 scaling factor. The objective is to develop effective super-resolution solutions and analyze recent advances in the field. To reflect the evolving objectives of image super-resolution, the challenge includes two tracks: (1) a restoration track, which emphasizes pixel-wise fidelity and ranks submissions based on PSNR; and (2) a perceptual track, which focuses on visual realism and evaluates results using a perceptual score. A total of 194 participants registered for the challenge, with 31 teams submitting valid entries. This report summarizes the challenge design, datasets, evaluation protocol, main results, and methods of participating teams. The challenge provides a unified benchmark and offers insights into current progress and future directions in image super-resolution.", "date": "2026-04-16", "categories": []}
{"id": "2604.14632", "title": "High-Speed Full-Color HDR Imaging via Unwrapping Modulo-Encoded Spike Streams", "abstract": "Conventional RGB-based high dynamic range (HDR) imaging faces a fundamental trade-off between motion artifacts in multi-exposure captures and irreversible information loss in single-shot techniques. Modulo sensors offer a promising alternative by encoding theoretically unbounded dynamic range into wrapped measurements. However, existing modulo solutions remain bottlenecked by iterative unwrapping overhead and hardware constraints limiting them to low-speed, grayscale capture. In this work, we present a complete modulo-based HDR imaging system that enables high-speed, full-color HDR acquisition by synergistically advancing both the sensing formulation and the unwrapping algorithm. At the core of our approach is an exposure-decoupled formulation of modulo imaging that allows multiple measurements to be interleaved in time, preserving a clean, observation-wise measurement model. Building upon this, we introduce an iteration-free unwrapping algorithm that integrates diffusion-based generative priors with the physical least absolute remainder property of modulo images, supporting highly efficient, physics-consistent HDR reconstruction. Finally, to validate the practical viability of our system, we demonstrate a proof-of-concept hardware implementation based on modulo-encoded spike streams. This setup preserves the native high temporal resolution of spike cameras, achieving 1000 FPS full-color imaging while reducing output data bandwidth from approximately 20 Gbps to 6 Gbps. Extensive evaluations indicate that our coordinated approach successfully overcomes key systemic bottlenecks, demonstrating the feasibility of deploying modulo imaging in dynamic scenarios.", "date": "2026-04-16", "categories": []}
{"id": "2604.14788", "title": "Sequence Search: Automated Sequence Design using Neural Architecture Search", "abstract": "Developing an MR sequence is challenging and remains largely constrained by human intuition. Recently, AI-driven approaches have been proposed; however, most require an initial sequence for parameter optimization or extensive training datasets, limiting their general applicability. In this study, we propose \"Sequence Search,\" an automated sequence design framework based on neural architecture search. The method takes tissue properties, imaging parameters, and design objectives as inputs and generates pulse sequences satisfying the design objectives, without requiring prior knowledge of conventional sequence structures. Sequence Search iteratively generates candidate sequences through neural architecture search and optimizes them via a differentiable Bloch simulator and objective-specific loss functions using gradient-based learning. The framework successfully replicated conventional spin-echo, T2-weighted spin-echo, and inversion recovery sequences. Less intuitive solutions were also discovered, such as three-RF spin-echo-like sequences with reduced RF energy and refocusing phases deviating from the conventional Hahn-echo. This work establishes a generalizable framework for automated MR sequence design, highlighting the potential to explore configurations beyond conventional designs based on human intuition.", "date": "2026-04-16", "categories": []}
{"id": "2510.01608", "title": "NPN: Non-Linear Projections of the Null-Space for Imaging Inverse Problems", "abstract": "Imaging inverse problems aim to recover high-dimensional signals from undersampled, noisy measurements, a fundamentally ill-posed task with infinite solutions in the null-space of the sensing operator. To resolve this ambiguity, prior information is typically incorporated through handcrafted regularizers or learned models that constrain the solution space. However, these priors typically ignore the task-specific structure of that null-space. In this work, we propose Non-Linear Projections of the Null-Space (NPN), a novel class of regularization that, instead of enforcing structural constraints in the image domain, promotes solutions that lie in a low-dimensional projection of the sensing matrix's null-space with a neural network. Our approach has two key advantages: (1) Interpretability: by focusing on the structure of the null-space, we design sensing-matrix-specific priors that capture information orthogonal to the signal components that are fundamentally blind to the sensing process. (2) Flexibility: NPN is adaptable to various inverse problems, compatible with existing reconstruction frameworks, and complementary to conventional image-domain priors. We provide theoretical guarantees on convergence and reconstruction accuracy when used within plug-and-play methods. Empirical results across diverse sensing matrices demonstrate that NPN priors consistently enhance reconstruction fidelity in various imaging inverse problems, such as compressive sensing, deblurring, super-resolution, computed tomography, and magnetic resonance imaging, with plug-and-play methods, unrolling networks, deep image prior, and diffusion models.", "date": "2025-10-02", "categories": []}
{"id": "2407.17182", "title": "A DeepONet for inverting the Neumann-to-Dirichlet Operator in Electrical Impedance Tomography: An approximation theoretic perspective and numerical results", "abstract": "In this work, we consider the non-invasive medical imaging modality of Electrical Impedance Tomography (EIT), where the goal is to recover the conductivity in a medium from boundary current-to-voltage measurements, i.e., the Neumann-to-Dirichlet (N--t--D) operator. We formulate this inverse problem as an operator-learning task, where the aim is to approximate the implicitly defined map from N--t--D operators to admissible conductivities. To this end, we employ a Deep Operator Network (DeepONet) architecture, thereby extending operator learning beyond the classical function-to-function setting to the more challenging operator-to-function regime. We establish a universal approximation theorem that guarantees that such operator-to-function maps can be approximated arbitrarily well by DeepONets. Furthermore, we provide a computational implementation of our approach and compare it against the iteratively regularized Gauss--Newton (IRGN) method. Our results show that the proposed framework yields accurate and robust reconstructions, outperforms the baseline, and demonstrates strong generalization. To our knowledge, this is the first work that combines rigorous approximation-theoretic guarantees with DeepONet-based inversion for EIT, thereby opening a principled and interpretable pathway for use of DeepONets in such inverse problems.", "date": "2024-07-24", "categories": []}
{"id": "2603.01591", "title": "FAST-DIPS: Adjoint-Free Analytic Steps and Hard-Constrained Likelihood Correction for Diffusion-Prior Inverse Problems", "abstract": "Training-free diffusion priors enable inverse-problem solvers without retraining, but for nonlinear forward operators data consistency often relies on repeated derivatives or inner optimization/MCMC loops with conservative step sizes, incurring many iterations and denoiser/score evaluations. We propose a training-free solver that replaces these inner loops with a hard measurement-space feasibility constraint (closed-form projection) and an analytic, model-optimal step size, enabling a small, fixed compute budget per noise level. Anchored at the denoiser prediction, the correction is approximated via an adjoint-free, ADMM-style splitting with projection and a few steepest-descent updates, using one VJP and either one JVP or a forward-difference probe, followed by backtracking and decoupled re-annealing. We prove local model optimality and descent under backtracking for the step-size rule, and derive an explicit KL bound for mode-substitution re-annealing under a local Gaussian conditional surrogate. We also develop a latent variant and a one-parameter pixel$\\rightarrow$latent hybrid schedule. Experiments achieve competitive PSNR/SSIM/LPIPS with up to 19.5$\\times$ speedup, without hand-coded adjoints or inner MCMC.", "date": "2026-03-02", "categories": []}
{"id": "2604.12257", "title": "Style-Decoupled Adaptive Routing Network for Underwater Image Enhancement", "abstract": "Underwater Image Enhancement (UIE) is essential for robust visual perception in marine applications. However, existing methods predominantly rely on uniform mapping tailored to average dataset distributions, leading to over-processing mildly degraded images or insufficient recovery for severe ones. To address this challenge, we propose a novel adaptive enhancement framework, SDAR-Net. Unlike existing uniform paradigms, it first decouples specific degradation styles from the input and subsequently modulates the enhancement process adaptively. Specifically, since underwater degradation primarily shifts the appearance while keeping the scene structure, SDAR-Net formulates image features into dynamic degradation style embeddings and static scene structural representations through a carefully designed training framework. Subsequently, we introduce an adaptive routing mechanism. By evaluating style features and adaptively predicting soft weights at different enhancement states, it guides the weighted fusion of the corresponding image representations, accurately satisfying the adaptive restoration demands of each image. Extensive experiments show that SDAR-Net achieves a new state-of-the-art (SOTA) performance with a PSNR of 25.72 dB on real-world benchmark, and demonstrates its utility in downstream vision tasks. Our code is available at https://github.com/WHU-USI3DV/SDAR-Net.", "date": "2026-04-14", "categories": []}
{"id": "2604.12917", "title": "M3D-Stereo: A Multiple-Medium and Multiple-Degradation Dataset for Stereo Image Restoration", "abstract": "Image restoration under adverse conditions, such as underwater, haze or fog, and low-light environments, remains a highly challenging problem due to complex physical degradations and severe information loss. Existing datasets are predominantly limited to a single degradation type or heavily rely on synthetic data without stereo consistency, inherently restricting their applicability in real-world scenarios. To address this, we introduce M3D-Stereo, a stereo dataset with 7904 high-resolution image pairs for image restoration research acquired in multiple media with multiple controlled degradation levels. It encompasses four degradation scenarios: underwater scatter, haze/fog, underwater low-light, and haze low-light. Each scenario forms a subset, and is divided into six levels of progressive degradation, allowing fine-grained evaluations of restoration methods with increasing severity of degradation. Collected via a laboratory setup, the dataset provides aligned stereo image pairs along with their pixel-wise consistent clear ground truths. Two restoration tasks, single-level and mixed-level degradation, were performed to verify its validity. M3D-Stereo establishes a better controlled and more realistic benchmark to evaluate image restoration and stereo matching methods in complex degradation environments. It is made public under LGPLv3 license.", "date": "2026-04-14", "categories": []}
{"id": "2404.13972", "title": "Non-Uniform Exposure Imaging via Neuromorphic Shutter Control", "abstract": "By leveraging the blur-noise trade-off, imaging with non-uniform exposures largely extends the image acquisition flexibility in harsh environments. However, the limitation of conventional cameras in perceiving intra-frame dynamic information prevents existing methods from being implemented in the real-world frame acquisition for real-time adaptive camera shutter control. To address this challenge, we propose a novel Neuromorphic Shutter Control (NSC) system to avoid motion blurs and alleviate instant noises, where the extremely low latency of events is leveraged to monitor the real-time motion and facilitate the scene-adaptive exposure. Furthermore, to stabilize the inconsistent Signal-to-Noise Ratio (SNR) caused by the non-uniform exposure times, we propose an event-based image denoising network within a self-supervised learning paradigm, i.e., SEID, exploring the statistics of image noises and inter-frame motion information of events to obtain artificial supervision signals for high-quality imaging in real-world scenes. To illustrate the effectiveness of the proposed NSC, we implement it in hardware by building a hybrid-camera imaging prototype system, with which we collect a real-world dataset containing well-synchronized frames and events in diverse scenarios with different target scenes and motion patterns. Experiments on the synthetic and real-world datasets demonstrate the superiority of our method over state-of-the-art approaches.", "date": "2024-04-22", "categories": []}
{"id": "2507.11931", "title": "Dark-EvGS: Event Camera as an Eye for Radiance Field in the Dark", "abstract": "In low-light environments, conventional cameras often struggle to capture clear multi-view images of objects due to dynamic range limitations and motion blur caused by long exposure. Event cameras, with their high-dynamic range and high-speed properties, have the potential to mitigate these issues. Additionally, 3D Gaussian Splatting (GS) enables radiance field reconstruction, facilitating bright frame synthesis from multiple viewpoints in low-light conditions. However, naively using an event-assisted 3D GS approach still faced challenges because, in low light, events are noisy, frames lack quality, and the color tone may be inconsistent. To address these issues, we propose Dark-EvGS, the first event-assisted 3D GS framework that enables the reconstruction of bright frames from arbitrary viewpoints along the camera trajectory. Triplet-level supervision is proposed to gain holistic knowledge, granular details, and sharp scene rendering. The color tone matching block is proposed to guarantee the color consistency of the rendered frames. Furthermore, we introduce the first real-captured dataset for the event-guided bright frame synthesis task via 3D GS-based radiance field reconstruction. Experiments demonstrate that our method achieves better results than existing methods, conquering radiance field reconstruction under challenging low-light conditions. The code and sample data are included in the supplementary material.", "date": "2025-07-16", "categories": []}
{"id": "2509.22736", "title": "PnP-CM: Consistency Models as Plug-and-Play Priors for Inverse Problems", "abstract": "Diffusion models have found extensive use in solving inverse problems, by sampling from an approximate posterior distribution of data given the measurements. Recently, consistency models (CMs) have been proposed to directly predict the final output from any point on the diffusion ODE trajectory, enabling high-quality sampling in just a few neural function evaluations (NFEs). CMs have also been utilized for inverse problems, but existing CM-based solvers either require additional task-specific training or utilize data fidelity operations with slow convergence, limiting their applicability to large-scale problems and making them difficult to extend to nonlinear settings. In this work, we reinterpret CMs as proximal operators of a prior, enabling their integration into plug-and-play (PnP) frameworks. Specifically, we propose PnP-CM, an ADMM-based PnP solver that provides a unified framework for solving a wide range of inverse problems, and incorporates noise perturbations and momentum-based updates to improve performance in the low-NFE regime. We evaluate our approach on a diverse set of linear and nonlinear inverse problems. We also train and apply CMs to MRI data for the first time. Our results show that PnP-CM achieves high-quality reconstructions in as few as 4 NFEs, and produces meaningful results in 2 steps, highlighting its effectiveness in real-world inverse problems while outperforming existing CM-based approaches.", "date": "2025-09-25", "categories": []}
{"id": "2512.09333", "title": "Improved Physics-Driven Neural Network to Solve Inverse Scattering Problems", "abstract": "This paper presents an improved physics-driven neural network (IPDNN) framework for solving electromagnetic inverse scattering problems (ISPs). A new Gaussian-localized oscillation-suppressing window (GLOW) activation function is introduced to stabilize convergence and enable a lightweight yet accurate network architecture. A dynamic scatter subregion identification strategy is further developed to adaptively refine the computational domain, preventing missed detections and reducing computational cost. Moreover, transfer learning is incorporated to extend the solver's applicability to practical scenarios, integrating the physical interpretability of iterative algorithms with the real-time inference capability of neural networks. Numerical simulations and experimental results demonstrate that the proposed solver achieves superior reconstruction accuracy, robustness, and efficiency compared with existing state-of-the-art methods.", "date": "2025-12-10", "categories": []}
{"id": "2604.03402", "title": "DRIFT: Deep Restoration, ISP Fusion, and Tone-mapping", "abstract": "Smartphone cameras have gained immense popularity with the adoption of high-resolution and high-dynamic range imaging. As a result, high-performance camera Image Signal Processors (ISPs) are crucial in generating high-quality images for the end user while keeping computational costs low. In this paper, we propose DRIFT (Deep Restoration, ISP Fusion, and Tone-mapping): an efficient AI mobile camera pipeline that generates high quality RGB images from hand-held raw captures. The first stage of DRIFT is a Multi-Frame Processing (MFP) network that is trained using a adversarial perceptual loss to perform multi-frame alignment, denoising, demosaicing, and super-resolution. Then, the output of DRIFT-MFP is processed by a novel deep-learning based tone-mapping (DRIFT-TM) solution that allows for tone tunability, ensures tone-consistency with a reference pipeline, and can be run efficiently for high-resolution images on a mobile device. We show qualitative and quantitative comparisons against state-of-the-art MFP and tone-mapping methods to demonstrate the effectiveness of our approach.", "date": "2026-04-03", "categories": []}
{"id": "2604.09643", "title": "PA-SFM: Tracker-free differentiable acoustic radiation for freehand 3D photoacoustic imaging", "abstract": "Three-dimensional (3D) handheld photoacoustic tomography typically relies on bulky and expensive external positioning sensors to correct motion artifacts, which severely limits its clinical flexibility and accessibility. To address this challenge, we present PA-SFM, a tracker-free framework that leverages exclusively single-modality photoacoustic data for both sensor pose recovery and high-fidelity 3D reconstruction via differentiable acoustic radiation modeling. Unlike traditional structure-from-motion (SFM) methods based on visual features, PA-SFM integrates the acoustic wave equation into a differentiable programming pipeline. By leveraging a high-performance, GPU-accelerated acoustic radiation kernel, the framework simultaneously optimizes the 3D photoacoustic source distribution and the sensor array pose via gradient descent. To ensure robust convergence in freehand scenarios, we introduce a coarse-to-fine optimization strategy that incorporates geometric consistency checks and rigid-body constraints to eliminate motion outliers. We validated the proposed method through both numerical simulations and in-vivo rat experiments. The results demonstrate that PA-SFM achieves sub-millimeter positioning accuracy and restores high-resolution 3D vascular structures comparable to ground-truth benchmarks, offering a low-cost, software-defined solution for clinical freehand photoacoustic imaging. The source code is publicly available at \\href{https://github.com/JaegerCQ/PA-SFM}{https://github.com/JaegerCQ/PA-SFM}.", "date": "2026-03-23", "categories": []}
{"id": "2604.10037", "title": "Compact single-shot ranging and near-far imaging using metasurfaces", "abstract": "We present a metasurface imaging system capable of simultaneously capturing two images at close range (1-2~cm) and an additional image at long range (about 40~cm) on a shared photosensor. The close-range image pair focuses at 1.4~cm and 2.0~cm, respectively, which forms a focal stack, enabling passive ranging with an accuracy of $\\pm$1~mm from 12~mm to 20~mm through a computationally efficient depth-from-defocus algorithm for a simplified scenario. The entire system is compact, with a total track length of 15~mm, making it suitable for seamless integration into edge platforms for defense and other resource-constrained applications.", "date": "2026-04-11", "categories": []}
{"id": "2604.10246", "title": "A Comparison of Multi-View Stereo Methods for Photogrammetric 3D Reconstruction: From Traditional to Learning-Based Approaches", "abstract": "Photogrammetric 3D reconstruction has long relied on traditional Structure-from-Motion (SfM) and Multi-View Stereo (MVS) methods, which provide high accuracy but face challenges in speed and scalability. Recently, learning-based MVS methods have emerged, aiming for faster and more efficient reconstruction. This work presents a comparative evaluation between a representative traditional MVS pipeline (COLMAP) and state-of-the-art learning-based approaches, including geometry-guided methods (MVSNet, PatchmatchNet, MVSAnywhere, MVSFormer++) and end-to-end frameworks (Stereo4D, FoundationStereo, DUSt3R, MASt3R, Fast3R, VGGT). Two experiments were conducted on different aerial scenarios. The first experiment used the MARS-LVIG dataset, where ground-truth 3D reconstruction was provided by LiDAR point clouds. The second experiment used a public scene from the Pix4D official website, with ground truth generated by Pix4Dmapper. We evaluated accuracy, coverage, and runtime across all methods. Experimental results show that although COLMAP can provide reliable and geometrically consistent reconstruction results, it requires more computation time. In cases where traditional methods fail in image registration, learning-based approaches exhibit stronger feature-matching capability and greater robustness. Geometry-guided methods usually require careful dataset preparation and often depend on camera pose or depth priors generated by COLMAP. End-to-end methods such as DUSt3R and VGGT achieve competitive accuracy and reasonable coverage while offering substantially faster reconstruction. However, they exhibit relatively large residuals in 3D reconstruction, particularly in challenging scenarios.", "date": "2026-04-11", "categories": []}
{"id": "2604.10554", "title": "Spatio-Temporal Difference Guided Motion Deblurring with the Complementary Vision Sensor", "abstract": "Motion blur arises when rapid scene changes occur during the exposure period, collapsing rich intra-exposure motion into a single RGB frame. Without explicit structural or temporal cues, RGB-only deblurring is highly ill-posed and often fails under extreme motion. Inspired by the human visual system, brain-inspired vision sensors introduce temporally dense information to alleviate this problem. However, event cameras still suffer from event rate saturation under rapid motion, while the event modality entangles edge features and motion cues, which limits their effectiveness. As a recent breakthrough, the complementary vision sensor (CVS), Tianmouc, captures synchronized RGB frames together with high-frame-rate, multi-bit spatial difference (SD, encoding structural edges) and temporal difference (TD, encoding motion cues) data within a single RGB exposure, offering a promising solution for RGB deblurring under extreme dynamic scenes. To fully leverage these complementary modalities, we propose Spatio-Temporal Difference Guided Deblur Net (STGDNet), which adopts a recurrent multi-branch architecture that iteratively encodes and fuses SD and TD sequences to restore structure and color details lost in blurry RGB inputs. Our method outperforms current RGB or event-based approaches in both synthetic CVS dataset and real-world evaluations. Moreover, STGDNet exhibits strong generalization capability across over 100 extreme real-world scenarios. Project page: https://tmcDeblur.github.io/", "date": "2026-04-12", "categories": []}
{"id": "2604.10610", "title": "Physics-Informed Synthetic Dataset and Denoising TIE-Reconstructed Phase Maps in Transient Flows Using Deep Learning", "abstract": "High-speed quantitative phase imaging enables non-intrusive visualization of transient compressible gas flows and energetic phenomena. However, phase maps reconstructed via the transport of intensity equation (TIE) suffer from spatially correlated low-frequency artifacts introduced by the inverse Laplacian solver, which obscure meaningful flow structures such as jet plumes, shockwave fronts, and density gradients. Conventional filtering approaches fail because signal and noise occupy overlapping spatial frequency bands, and no paired ground truth exists since every frame represents a physically unique, non-repeatable flow state. We address this by developing a physics-informed synthetic training dataset where clean targets are procedurally generated using physically plausible gas flow morphologies, including compressible jet plumes, turbulent eddy fields, density fronts, periodic air pockets, and expansion fans, and passed through a forward TIE simulation followed by inverse Laplacian reconstruction to produce realistic noisy phase maps. A U-Net-based convolutional denoising network trained solely on this synthetic data is evaluated on real phase maps acquired at 25,000 fps, demonstrating zero-shot generalization to real parallel TIE recordings, with a 13,260% improvement in signal-to-background ratio and 100.8% improvement in jet-region structural sharpness across 20 evaluated frames.", "date": "2026-04-12", "categories": []}
{"id": "2604.10619", "title": "How to Design a Compact High-Throughput Video Camera?", "abstract": "High throughput video acquisition is a challenging problem and has been drawing increasing attention. Existing high throughput imaging systems splice hundreds of sub-images/videos into high throughput videos, suffering from extremely high system complexity. Alternatively, with pixel sizes reducing to sub-micrometer levels, integrating ultra-high throughput on a single chip is becoming feasible. Nevertheless, the readout and output transmission speed cannot keep pace with the increasing pixel numbers. To this end, this paper analyzes the strength of gradient cameras in fast readout and efficient representation, and proposes a low-bit gradient camera scheme based on existing technologies that can resolve the readout and transmission bottlenecks for high throughput video imaging. A multi-scale reconstruction CNN is proposed to reconstruct high-resolution images. Extensive experiments on both simulated and real data are conducted to demonstrate the promising quality and feasibility of the proposed method.", "date": "2026-04-12", "categories": []}
{"id": "2604.10777", "title": "Uncertainty-quantified Pulse Signal Recovery from Facial Video using Regularized Stochastic Interpolants", "abstract": "Imaging Photoplethysmography (iPPG), an optical procedure which recovers a human's blood volume pulse (BVP) waveform using pixel readout from a camera, is an exciting research field with many researchers performing clinical studies of iPPG algorithms. While current algorithms to solve the iPPG task have shown outstanding performance on benchmark datasets, no state-of-the art algorithms, to the best of our knowledge, performs test-time sampling of solution space, precluding an uncertainty analysis that is critical for clinical applications. We address this deficiency though a new paradigm named Regularized Interpolants with Stochasticity for iPPG (RIS-iPPG). Modeling iPPG recovery as an inverse problem, we build probability paths that evolve the camera pixel distribution to the ground-truth signal distribution by predicting the instantaneous flow and score vectors of a time-dependent stochastic process; and at test-time, we sample the posterior distribution of the correct BVP waveform given the camera pixel intensity measurements by solving a stochastic differential equation. Given that physiological changes are slowly varying, we show that iPPG recovery can be improved through regularization that maximizes the correlation between the residual flow vector predictions of two adjacent time windows. Experimental results on three datasets show that RIS-iPPG provides superior reconstruction quality and uncertainty estimates of the reconstruction, a critical tool for the widespread adoption of iPPG algorithms in clinical and consumer settings.", "date": "2026-04-12", "categories": []}
{"id": "2604.10983", "title": "Energy-oriented Diffusion Bridge for Image Restoration with Foundational Diffusion Models", "abstract": "Diffusion bridge models have shown great promise in image restoration by explicitly connecting clean and degraded image distributions. However, they often rely on complex and high-cost trajectories, which limit both sampling efficiency and final restoration quality. To address this, we propose an Energy-oriented diffusion Bridge (E-Bridge) framework to approximate a set of low-cost manifold geodesic trajectories to boost the performance of the proposed method. We achieve this by designing a novel bridge process that evolves over a shorter time horizon and makes the reverse process start from an entropy-regularized point that mixes the degraded image and Gaussian noise, which theoretically reduces the required trajectory energy. To solve this process efficiently, we draw inspiration from consistency models to learn a single-step mapping function, optimized via a continuous-time consistency objective tailored for our trajectory, so as to analytically map any state on the trajectory to the target image. Notably, the trajectory length in our framework becomes a tunable task-adaptive knob, allowing the model to adaptively balance information preservation against generative power for tasks of varying degradation, such as denoising versus super-resolution. Extensive experiments demonstrate that our E-Bridge achieves state-of-the-art performance across various image restoration tasks while enabling high-quality recovery with a single or fewer sampling steps. Our project page is https://jinnh.github.io/E-Bridge/.", "date": "2026-04-13", "categories": []}
{"id": "2604.10994", "title": "LumiMotion: Improving Gaussian Relighting with Scene Dynamics", "abstract": "In 3D reconstruction, the problem of inverse rendering, namely recovering the illumination of the scene and the material properties, is fundamental. Existing Gaussian Splatting-based methods primarily target static scenes and often assume simplified or moderate lighting to avoid entangling shadows with surface appearance. This limits their ability to accurately separate lighting effects from material properties, particularly in real-world conditions. We address this limitation by leveraging dynamic elements - regions of the scene that undergo motion - as a supervisory signal for inverse rendering. Motion reveals the same surfaces under varying lighting conditions, providing stronger cues for disentangling material and illumination. This thesis is supported by our experimental results which show we improve LPIPS by 23% for albedo estimation and by 15% for scene relighting relative to next-best baseline. To this end, we introduce LumiMotion, the first Gaussian-based approach that leverages dynamics for inverse rendering and operates in arbitrary dynamic scenes. Our method learns a dynamic 2D Gaussian Splatting representation that employs a set of novel constraints which encourage the dynamic regions of the scene to deform, while keeping static regions stable. As we demonstrate, this separation is crucial for correct optimization of the albedo. Finally, we release a new synthetic benchmark comprising five scenes under four lighting conditions, each in both static and dynamic variants, for the first time enabling systematic evaluation of inverse rendering methods in dynamic environments and challenging lighting. Link to project page: https://joaxkal.github.io/LumiMotion/", "date": "2026-04-13", "categories": []}
{"id": "2604.11014", "title": "UHD-GPGNet: UHD Video Denoising via Gaussian-Process-Guided Local Spatio-Temporal Modeling", "abstract": "Ultra-high-definition (UHD) video denoising requires simultaneously suppressing complex spatio-temporal degradations, preserving fine textures and chromatic stability, and maintaining efficient full-resolution 4K deployment. In this paper, we propose UHD-GPGNet, a Gaussian-process-guided local spatio-temporal denoising framework that addresses these requirements jointly. Rather than relying on implicit feature learning alone, the method estimates sparse GP posterior statistics over compact spatio-temporal descriptors to explicitly characterize local degradation response and uncertainty, which then guide adaptive temporal-detail fusion. A structure-color collaborative reconstruction head decouples luminance, chroma, and high-frequency correction, while a heteroscedastic objective and overlap-tiled inference further stabilize optimization and enable memory-bounded 4K deployment. Experiments on UVG and RealisVideo-4K show that UHD-GPGNet achieves competitive restoration fidelity with substantially fewer parameters than existing methods, enables real-time full-resolution 4K inference with significant speedup over the closest quality competitor, and maintains robust performance across a multi-level mixed-degradation schedule.A real-world study on phone-captured 4K video further confirms that the model, trained entirely on synthetic degradation, generalizes to unseen real sensor noise and improves downstream object detection under challenging conditions.", "date": "2026-04-13", "categories": []}
{"id": "2604.11279", "title": "A Deep Equilibrium Network for Hyperspectral Unmixing", "abstract": "Hyperspectral unmixing (HU) is crucial for analyzing hyperspectral imagery, yet achieving accurate unmixing remains challenging. While traditional methods struggle to effectively model complex spectral-spatial features, deep learning approaches often lack physical interpretability. Unrolling-based methods, despite offering network interpretability, inadequately exploit spectral-spatial information and incur high memory costs and numerical precision issues during backpropagation. To address these limitations, we propose DEQ-Unmix, which reformulates abundance estimation as a deep equilibrium model, enabling efficient constant-memory training via implicit differentiation. It replaces the gradient operator of the data reconstruction term with a trainable convolutional network to capture spectral-spatial information. By leveraging implicit differentiation, DEQ-Unmix enables efficient and constant-memory backpropagation. Experiments on synthetic and two real-world datasets demonstrate that DEQ-Unmix achieves superior unmixing performance while maintaining constant memory cost.", "date": "2026-04-13", "categories": []}
{"id": "2505.18600", "title": "Chain-of-Zoom: Extreme Super-Resolution via Scale Autoregression and Preference Alignment", "abstract": "Modern single-image super-resolution (SISR) models deliver photo-realistic results at the scale factors on which they are trained, but collapse when asked to magnify far beyond that regime. We address this scalability bottleneck with Chain-of-Zoom (CoZ), a model-agnostic framework that factorizes SISR into an autoregressive chain of intermediate scale-states with multi-scale-aware prompts. CoZ repeatedly re-uses a backbone SR model, decomposing the conditional probability into tractable sub-problems to achieve extreme resolutions without additional training. Because visual cues diminish at high magnifications, we augment each zoom step with multi-scale-aware text prompts generated by a vision-language model (VLM). The prompt extractor itself is fine-tuned using Generalized Reward Policy Optimization (GRPO) with a critic VLM, aligning text guidance towards human preference. Experiments show that a standard 4x diffusion SR model wrapped in CoZ attains beyond 256x enlargement with high perceptual quality and fidelity. Project Page: https://bryanswkim.github.io/chain-of-zoom/.", "date": "2025-05-24", "categories": []}
{"id": "2604.07248", "title": "TurPy: a physics-based and differentiable optical turbulence simulator for algorithmic development and system optimization", "abstract": "Developing optical systems for free-space applications requires simulation tools that accurately capture turbulence-induced wavefront distortions and support gradient-based optimization. Here we introduce TurPy, a GPU-accelerated, fully differentiable wave optics turbulence simulator to bridge high fidelity simulation with end-to-end optical system design. TurPy incorporates subharmonic phase screen generation, autoregressive temporal evolution, and an automated screen placement routine balancing Fourier aliasing constraints and weak-turbulence approximations into a unified, user-ready framework. Because TurPy's phase screen generation is parameterized through a media-specific power spectral density, the framework extends to atmospheric, oceanic, and biological propagation environments with minimal modification. We validate TurPy against established atmospheric turbulence theory by matching 2nd order Gaussian beam broadening and 4th order plane wave scintillation to closed-form models with 98% accuracy across weak to strong turbulence regimes, requiring only the medium's refractive index structure constant and power spectral density as inputs. To demonstrate TurPy as a gradient-based training platform, we optimize a dual-domain diffractive deep neural network (D2NN) in a two-mask dual-domain architecture to recover a Gaussian beam from a weakly turbulent path and achieving over 20x reduction in scintillation relative to an uncompensated receiver in simulation. TurPy is released as an open-source package to support synthetic data generation, turbulence-informed algorithm development, and the end-to-end design of optical platforms operating in turbulent environments.", "date": "2026-04-08", "categories": []}
{"id": "2604.08781", "title": "PSIRNet: Deep Learning-based Free-breathing Rapid Acquisition Late Enhancement Imaging", "abstract": "Purpose: To develop and evaluate a deep learning (DL) method for free-breathing phase-sensitive inversion recovery (PSIR) late gadolinium enhancement (LGE) cardiac MRI that produces diagnostic-quality images from a single acquisition over two heartbeats, eliminating the need for 8 to 24 motion-corrected (MOCO) signal averages.\n  Materials and Methods: Raw data comprising 800,653 slices from 55,917 patients, acquired on 1.5T and 3T scanners across multiple sites from 2016 to 2024, were used in this retrospective study. Data were split by patient: 640,000 slices (42,822 patients) for training and the remainder for validation and testing, without overlap. The training and testing data were from different institutions. PSIRNet, a physics-guided DL network with 845 million parameters, was trained end-to-end to reconstruct PSIR images with surface coil correction from a single interleaved IR/PD acquisition over two heartbeats. Reconstruction quality was evaluated using SSIM, PSNR, and NRMSE against MOCO PSIR references. Two expert cardiologists performed an independent qualitative assessment, scoring image quality on a 5-point Likert scale across bright blood, dark blood, and wideband LGE variants. Paired superiority and equivalence (margin = 0.25 Likert points) were tested using exact Wilcoxon signed-rank tests at a significance level of 0.05 using R version 4.5.2.\n  Results: Both readers rated single-average PSIRNet reconstructions superior to MOCO PSIR for dark blood LGE (conservative P = .002); for bright blood and wideband, one reader rated it superior and the other confirmed equivalence (all P < .001). Inference required approximately 100 msec per slice versus more than 5 sec for MOCO PSIR.\n  Conclusion: PSIRNet produces diagnostic-quality free-breathing PSIR LGE images from a single acquisition, enabling 8- to 24-fold reduction in acquisition time.", "date": "2026-04-09", "categories": []}
{"id": "2604.08924", "title": "Customized Fusion: A Closed-Loop Dynamic Network for Adaptive Multi-Task-Aware Infrared-Visible Image Fusion", "abstract": "Infrared-visible image fusion aims to integrate complementary information for robust visual understanding, but existing fusion methods struggle with simultaneously adapting to multiple downstream tasks. To address this issue, we propose a Closed-Loop Dynamic Network (CLDyN) that can adaptively respond to the semantic requirements of diverse downstream tasks for task-customized image fusion. Specifically, CLDyN introduces a closed-loop optimization mechanism that establishes a semantic transmission chain to achieve explicit feedback from downstream tasks to the fusion network through a Requirement-driven Semantic Compensation (RSC) module. The RSC module leverages a Basis Vector Bank (BVB) and an Architecture-Adaptive Semantic Injection (A2SI) block to customize the network architecture according to task requirements, thereby enabling task-specific semantic compensation and allowing the fusion network to actively adapt to diverse tasks without retraining. To promote semantic compensation, a reward-penalty strategy is introduced to reward or penalize the RSC module based on task performance variations. Experiments on the M3FD, FMB, and VT5000 datasets demonstrate that CLDyN not only maintains high fusion quality but also exhibits strong multi-task adaptability. The code is available at https://github.com/YR0211/CLDyN.", "date": "2026-04-10", "categories": []}
{"id": "2604.09030", "title": "NTIRE 2026 The 3rd Restore Any Image Model (RAIM) Challenge: Multi-Exposure Image Fusion in Dynamic Scenes (Track 2)", "abstract": "This paper presents NTIRE 2026, the 3rd Restore Any Image Model (RAIM) challenge on multi-exposure image fusion in dynamic scenes. We introduce a benchmark that targets a practical yet difficult HDR imaging setting, where exposure bracketing must be fused under scene motion, illumination variation, and handheld camera jitter. The challenge data contains 100 training sequences with 7 exposure levels and 100 test sequences with 5 exposure levels, reflecting real-world scenarios that frequently cause misalignment and ghosting artefacts. We evaluate submissions with a leaderboard score derived from PSNR, SSIM, and LPIPS, while also considering perceptual quality, efficiency, and reproducibility during the final review. This track attracted 114 participating teams and received 987 submissions. The winning methods significantly improved the ability to remove artifacts from multi-exposure fusion and recover fine details. The dataset and the code of each team can be found at the repository: https://github.com/qulishen/RAIM-HDR.", "date": "2026-04-10", "categories": []}
{"id": "2604.09145", "title": "Deep Light Pollution Removal in Night Cityscape Photographs", "abstract": "Nighttime photography is severely degraded by light pollution induced by pervasive artificial lighting in urban environments. After long-range scattering and spatial diffusion, unwanted artificial light overwhelms natural night luminance, generates skyglow that washes out the view of stars and celestial objects and produces halos and glow artifacts around light sources. Unlike nighttime dehazing, which aims to improve detail legibility through thick air, the objective of light pollution removal is to restore the pristine night appearance by neutralizing the radiative footprint of ground lighting. In this paper we introduce a physically-based degradation model that adds to the previous ones for nighttime dehazing two critical aspects; (i) anisotropic spread of directional light sources, and (ii) skyglow caused by invisible surface lights behind skylines. In addition, we construct a training strategy that leverages large generative model and synthetic-real coupling to compensate for the scarcity of paired real data and enhance generalization. Extensive experiments demonstrate that the proposed formulation and learning framework substantially reduce light pollution artifacts and better recover authentic night imagery than prior nighttime restoration methods.", "date": "2026-04-10", "categories": []}
{"id": "2604.09511", "title": "RIRF: Reasoning Image Restoration Framework", "abstract": "Universal image restoration (UIR) aims to recover clean images from diverse and unknown degradations using a unified model. Existing UIR methods primarily focus on pixel reconstruction and often lack explicit diagnostic reasoning over degradation composition, severity, and scene semantics prior to restoration. We propose Reason and Restore (R\\&R), a novel framework that integrates structured Chain-of-Thought (CoT) reasoning into the image restoration pipeline. R\\&R introduces an explicit reasoner, implemented by fine-tuning Qwen3-VL, to diagnose degradation types, quantify degradation severity, infer key degradation-related factors, and describe relevant scene and object semantics. The resulting structured reasoning provides interpretable and fine-grained diagnostic priors for the restorer. To further improve restoration quality, the quantified degradation severity produced by the reasoner is leveraged as reinforcement learning (RL) signals to guide and strengthen the restorer. Unlike existing multimodal LLM-based agentic systems that decouple reasoning from low-level vision tasks, R\\&R tightly couples semantic diagnostic reasoning with pixel-level restoration in a unified framework. Extensive experiments across diverse UIR benchmarks demonstrate that R\\&R achieves state-of-the-art performance while offering unique interpretability into the restoration process.", "date": "2026-04-10", "categories": []}
{"id": "2508.21618", "title": "Physics-Informed Spectral Modeling for Hyperspectral Imaging", "abstract": "We present PhISM, a physics-informed deep learning architecture that learns without supervision to explicitly disentangle hyperspectral observations and model them with continuous basis functions. PhISM outperforms prior methods on several classification and regression benchmarks, requires limited labeled data, and provides additional insights thanks to interpretable latent representation.", "date": "2025-08-29", "categories": []}
{"id": "2604.06561", "title": "Accelerating 4D Hyperspectral Imaging through Physics-Informed Neural Representation and Adaptive Sampling", "abstract": "High-dimensional hyperspectral imaging (HSI) enables the visualization of ultrafast molecular dynamics and complex, heterogeneous spectra. However, applying this capability to resolve spatially varying vibrational couplings in two-dimensional infrared (2DIR) spectroscopy, a type of coherent multidimensional spectroscopy (CMDS), necessitates prohibitively long data acquisition, driven by dense Nyquist sampling requirements and the need for extensive signal accumulation. To address this challenge, we introduce a physics-informed neural representation approach that efficiently reconstructs dense spatially-resolved 2DIR hyperspectral images from sparse experimental measurements. In particular, we used a multilayer perceptron (MLP) to model the relationship between the sub-sampled 4D coordinates and their corresponding spectral intensities, and recover densely sampled 4D spectra from limited observations. The reconstruction results demonstrate that our method, using a fraction of the samples, faithfully recovers both oscillatory and non-oscillatory spectral dynamics in experimental measurement. Moreover, we develop a loss-aware adaptive sampling method to progressively introduce potentially informative samples for iterative data collection while conducting experiments. Experimental results show that the proposed approach achieves high-fidelity spectral recovery using only $1/32$ of the sampling budget, as opposed to exhaustive sampling, effectively reducing total experiment time by up to 32-fold. This framework offers a scalable solution for accelerating any experiments with hypercube data, including multidimensional spectroscopy and hyperspectral imaging, paving the way for rapid chemical imaging of transient biological and material systems.", "date": "2026-04-08", "categories": []}
{"id": "2604.06739", "title": "DOC-GS: Dual-Domain Observation and Calibration for Reliable Sparse-View Gaussian Splatting", "abstract": "Sparse-view reconstruction with 3D Gaussian Splatting (3DGS) is fundamentally ill-posed due to insufficient geometric supervision, often leading to severe overfitting and the emergence of structural distortions and translucent haze-like artifacts. While existing approaches attempt to alleviate this issue via dropout-based regularization, they are largely heuristic and lack a unified understanding of artifact formation. In this paper, we revisit sparse-view 3DGS reconstruction from a new perspective and identify the core challenge as the unobservability of Gaussian primitive reliability. Unreliable Gaussians are insufficiently constrained during optimization and accumulate as haze-like degradations in rendered images. Motivated by this observation, we propose a unified Dual-domain Observation and Calibration (DOC-GS) framework that models and corrects Gaussian reliability through the synergy of optimization-domain inductive bias and observation-domain evidence. Specifically, in the optimization domain, we characterize Gaussian reliability by the degree to which each primitive is constrained during training, and instantiate this signal via a Continuous Depth-Guided Dropout (CDGD) strategy, where the dropout probability serves as an explicit proxy for primitive reliability. This imposes a smooth depth-aware inductive bias to suppress weakly constrained Gaussians and improve optimization stability. In the observation domain, we establish a connection between floater artifacts and atmospheric scattering, and leverage the Dark Channel Prior (DCP) as a structural consistency cue to identify and accumulate anomalous regions. Based on cross-view aggregated evidence, we further design a reliability-driven geometric pruning strategy to remove low-confidence Gaussians.", "date": "2026-04-08", "categories": []}
{"id": "2506.22397", "title": "HazeMatching: Dehazing Light Microscopy Images with Guided Conditional Flow Matching", "abstract": "Fluorescence microscopy is a major driver of scientific progress in the life sciences. Although high-end confocal microscopes are capable of filtering out-of-focus light, cheaper and more accessible microscopy modalities, such as widefield microscopy, can not, which consequently leads to hazy image data. Computational dehazing is trying to combine the best of both worlds, leading to cheap microscopy but crisp-looking images. The perception-distortion trade-off tells us that we can optimize either for data fidelity, e.g. low MSE or high PSNR, or for data realism, measured by perceptual metrics such as LPIPS or FID. Existing methods either prioritize fidelity at the expense of realism, or produce perceptually convincing results that lack quantitative accuracy. In this work, we propose HazeMatching, a novel iterative method for dehazing light microscopy images, which effectively balances these objectives. Our goal was to find a balanced trade-off between the fidelity of the dehazing results and the realism of individual predictions (samples). We achieve this by adapting the conditional flow matching framework by guiding the generative process with a hazy observation in the conditional velocity field. We evaluate HazeMatching on 5 datasets, covering both synthetic and real data, assessing both distortion and perceptual quality. Our method is compared against 12 baselines, achieving a consistent balance between fidelity and realism on average. Additionally, with calibration analysis, we show that HazeMatching produces well-calibrated predictions. Note that our method does not need an explicit degradation operator to exist, making it easily applicable on real microscopy data. All data used for training and evaluation and our code will be publicly available under a permissive license.", "date": "2025-06-27", "categories": []}
{"id": "2508.08900", "title": "DSER: Spectral Epipolar Representation for Efficient Light Field Depth Estimation", "abstract": "Dense light field depth estimation remains challenging due to sparse angular sampling, occlusion boundaries, textureless regions, and the cost of exhaustive multi-view matching. We propose \\emph{Deep Spectral Epipolar Representation} (DSER), a geometry-aware framework that introduces spectral regularization in the epipolar domain for dense disparity reconstruction. DSER models frequency-consistent EPI structure to constrain correspondence estimation and couples this prior with a hybrid inference pipeline that combines least squares gradient initialization, plane-sweeping cost aggregation, and multiscale EPI refinement. An occlusion-aware directed random walk further propagates reliable disparity along edge-consistent paths, improving boundary sharpness and weak-texture stability. Experiments on benchmark and real-world light field datasets show that DSER achieves a strong accuracy-efficiency trade-off, producing more structurally consistent depth maps than representative classical and hybrid baselines. These results establish spectral epipolar regularization as an effective inductive bias for scalable and noise-robust light field depth estimation.", "date": "2025-08-12", "categories": []}
{"id": "2604.05581", "title": "High-Resolution Single-Shot Polarimetric Imaging Made Easy", "abstract": "Polarization-based vision has gained increasing attention for providing richer physical cues beyond RGB images. While achieving single-shot capture is highly desirable for practical applications, existing Division-of-Focal-Plane (DoFP) sensors inherently suffer from reduced spatial resolution and artifacts due to their spatial multiplexing mechanism. To overcome these limitations without sacrificing the snapshot capability, we propose EasyPolar, a multi-view polarimetric imaging framework. Our system is grounded in the physical insight that three independent intensity measurements are sufficient to fully characterize linear polarization. Guided by this, we design a triple-camera setup consisting of three synchronized RGB cameras that capture one unpolarized view and two polarized views with distinct orientations. Building upon this hardware design, we further propose a confidence-guided polarization reconstruction network to address the potential misalignment in multi-view fusion. The network performs multi-modal feature fusion under a confidence-aware physical guidance mechanism, which effectively suppresses warping-induced artifacts and enforces explicit geometric constraints on the solution space. Experimental results demonstrate that our method achieves high-quality results and benefits various downstream tasks.", "date": "2026-04-07", "categories": []}
{"id": "2603.27158", "title": "Weakly Convex Ridge Regularization for 3D Non-Cartesian MRI Reconstruction", "abstract": "While highly accelerated non-Cartesian acquisition protocols significantly reduce scan time, they often entail long reconstruction delays. Deep learning based reconstruction methods can alleviate this, but often lack stability and robustness to distribution shifts. As an alternative, we train a rotation invariant weakly convex ridge regularizer (WCRR). The resulting variational reconstruction approach is benchmarked against state of the art methods on retrospectively simulated data and (out of distribution) on prospective GoLF SPARKLING and CAIPIRINHA acquisitions. Our approach consistently outperforms widely used baselines and achieves performance comparable to Plug and Play reconstruction with a state of the art 3D DRUNet denoiser, while offering substantially improved computational efficiency and robustness to acquisition changes. In summary, WCRR unifies the strengths of principled variational methods and modern deep learning based approaches.", "date": "2026-03-28", "categories": []}
{"id": "2604.03572", "title": "Physics-Informed Untrained Learning for RGB-Guided Superresolution Single-Pixel Hyperspectral Imaging", "abstract": "Single-pixel imaging (SPI) offers a cost-effective route to hyperspectral acquisition but struggles to recover high-fidelity spatial and spectral details under extremely low sampling rates, a severely ill-posed inverse problem. While deep learning has shown potential, existing data-driven methods demand large-scale pretraining datasets that are often impractical in hyperspectral imaging. To overcome this limitation, we propose an end-to-end physics-informed framework that leverages untrained neural networks and RGB guidance for joint hyperspectral reconstruction and super-resolution without any external training data. The framework comprises three physically grounded stages: (1) a Regularized Least-Squares method with RGB-derived Grayscale Priors (LS-RGP) that initializes the solution by exploiting cross-modal structural correlations; (2) an Untrained Hyperspectral Recovery Network (UHRNet) that refines the reconstruction through measurement consistency and hybrid regularization; and (3) a Transformer-based Untrained Super-Resolution Network (USRNet) that upsamples the spatial resolution via cross-modal attention, transferring high-frequency details from the RGB guide. Extensive experiments on benchmark datasets demonstrate that our approach significantly surpasses state-of-the-art algorithms in both reconstruction accuracy and spectral fidelity. Moreover, a proof-of-concept experiment using a physical single-pixel imaging system validates the framework's practical applicability, successfully reconstructing a 144-band hyperspectral data cube at a mere 6.25% sampling rate. The proposed method thus provides a robust, data-efficient solution for computational hyperspectral imaging.", "date": "2026-04-04", "categories": []}
{"id": "2604.03603", "title": "Stochastic Generative Plug-and-Play Priors", "abstract": "Plug-and-play (PnP) methods are widely used for solving imaging inverse problems by incorporating a denoiser into optimization algorithms. Score-based diffusion models (SBDMs) have recently demonstrated strong generative performance through a denoiser trained across a wide range of noise levels. Despite their shared reliance on denoisers, it remains unclear how to systematically use SBDMs as priors within the PnP framework without relying on reverse diffusion sampling. In this paper, we establish a score-based interpretation of PnP that justifies using pretrained SBDMs directly within PnP algorithms. Building on this connection, we introduce a stochastic generative PnP (SGPnP) framework that injects noise to better leverage the expressive generative SBDM priors, thereby improving robustness in severely ill-posed inverse problems. We provide a new theory showing that this noise injection induces optimization on a Gaussian-smoothed objective and promotes escape from strict saddle points. Experiments on challenging inverse tasks, such as multi-coil MRI reconstruction and large-mask natural image inpainting, demonstrate consistent improvement over conventional PnP methods and achieve performance competitive with diffusion-based solvers.", "date": "2026-04-04", "categories": []}
{"id": "2604.04407", "title": "NAIMA: Semantics Aware RGB Guided Depth Super-Resolution", "abstract": "Guided depth super-resolution (GDSR) is a multi-modal approach for depth map super-resolution that relies on a low-resolution depth map and a high-resolution RGB image to restore finer structural details. However, the misleading color and texture cues indicating depth discontinuities in RGB images often lead to artifacts and blurred depth boundaries in the generated depth map. We propose a solution that introduces global contextual semantic priors, generated from pretrained vision transformer token embeddings. Our approach to distilling semantic knowledge from pretrained token embeddings is motivated by their demonstrated effectiveness in related monocular depth estimation tasks. We introduce a Guided Token Attention (GTA) module, which iteratively aligns encoded RGB spatial features with depth encodings, using cross-attention for selectively injecting global semantic context extracted from different layers of a pretrained vision transformer. Additionally, we present an architecture called Neural Attention for Implicit Multi-token Alignment (NAIMA), which integrates DINOv2 with GTA blocks for a semantics-aware GDSR. Our proposed architecture, with its ability to distill semantic knowledge, achieves significant improvements over existing methods across multiple scaling factors and datasets.", "date": "2026-04-06", "categories": []}
{"id": "2604.04693", "title": "3D Gaussian Splatting for Annular Dark Field Scanning Transmission Electron Microscopy Tomography Reconstruction", "abstract": "Analytical Dark Field Scanning Transmission Electron Microscopy (ADF-STEM) tomography reconstructs nanoscale materials in 3D by integrating multi-view tilt-series images, enabling precise analysis of their structural and compositional features. Although integrating more tilt views improves 3D reconstruction, it requires extended electron exposure that risks damaging dose-sensitive materials and introduces drift and misalignment, making it difficult to balance reconstruction fidelity with sample preservation. In practice, sparse-view acquisition is frequently required, yet conventional ADF-STEM methods degrade under limited views, exhibiting artifacts and reduced structural fidelity. To resolve these issues, in this paper, we adapt 3D GS to this domain with three key components. We first model the local scattering strength as a learnable scalar field, denza, to address the mismatch between 3DGS and ADF-STEM imaging physics. Then we introduce a coefficient $\\gamma$ to stabilize scattering across tilt angles, ensuring consistent denza via scattering view normalization. Finally, We incorporate a loss function that includes a 2D Fourier amplitude term to suppress missing wedge artifacts in sparse-view reconstruction. Experiments on 45-view and 15-view tilt series show that DenZa-Gaussian produces high-fidelity reconstructions and 2D projections that align more closely with original tilts, demonstrating superior robustness under sparse-view conditions.", "date": "2026-04-06", "categories": []}
{"id": "2604.02742", "title": "Task-Guided Prompting for Unified Remote Sensing Image Restoration", "abstract": "Remote sensing image restoration (RSIR) is essential for recovering high-fidelity imagery from degraded observations, enabling accurate downstream analysis. However, most existing methods focus on single degradation types within homogeneous data, restricting their practicality in real-world scenarios where multiple degradations often across diverse spectral bands or sensor modalities, creating a significant operational bottleneck. To address this fundamental gap, we propose TGPNet, a unified framework capable of handling denoising, cloud removal, shadow removal, deblurring, and SAR despeckling within a single, unified architecture. The core of our framework is a novel Task-Guided Prompting (TGP) strategy. TGP leverages learnable, task-specific embeddings to generate degradation-aware cues, which then hierarchically modulate features throughout the decoder. This task-adaptive mechanism allows the network to precisely tailor its restoration process for distinct degradation patterns while maintaining a single set of shared weights. To validate our framework, we construct a unified RSIR benchmark covering RGB, multispectral, SAR, and thermal infrared modalities for five aforementioned restoration tasks. Experimental results demonstrate that TGPNet achieves state-of-the-art performance on both unified multi-task scenarios and unseen composite degradations, surpassing even specialized models in individual domains such as cloud removal. By successfully unifying heterogeneous degradation removal within a single adaptive framework, this work presents a significant advancement for multi-task RSIR, offering a practical and scalable solution for operational pipelines. The code and benchmark will be released at https://github.com/huangwenwenlili/TGPNet.", "date": "2026-04-03", "categories": []}
{"id": "2604.02787", "title": "LumaFlux: Lifting 8-Bit Worlds to HDR Reality with Physically-Guided Diffusion Transformers", "abstract": "The rapid adoption of HDR-capable devices has created a pressing need to convert the 8-bit Standard Dynamic Range (SDR) content into perceptually and physically accurate 10-bit High Dynamic Range (HDR). Existing inverse tone-mapping (ITM) methods often rely on fixed tone-mapping operators that struggle to generalize to real-world degradations, stylistic variations, and camera pipelines, frequently producing clipped highlights, desaturated colors, or unstable tone reproduction. We introduce LumaFlux, a first physically and perceptually guided diffusion transformer (DiT) for SDR-to-HDR reconstruction by adapting a large pretrained DiT. Our LumaFlux introduces (1) a Physically-Guided Adaptation (PGA) module that injects luminance, spatial descriptors, and frequency cues into attention through low-rank residuals; (2) a Perceptual Cross-Modulation (PCM) layer that stabilizes chroma and texture via FiLM conditioning from vision encoder features; and (3) an HDR Residual Coupler that fuses physical and perceptual signals under a timestep- and layer-adaptive modulation schedule. Finally, a lightweight Rational-Quadratic Spline decoder reconstructs smooth, interpretable tone fields for highlight and exposure expansion, enhancing the output of the VAE decoder to generate HDR. To enable robust HDR learning, we curate the first large-scale SDR-HDR training corpus. For fair and reproducible comparison, we further establish a new evaluation benchmark, comprising HDR references and corresponding expert-graded SDR versions. Across benchmarks, LumaFlux outperforms state-of-the-art baselines, achieving superior luminance reconstruction and perceptual color fidelity with minimal additional parameters.", "date": "2026-04-03", "categories": []}
{"id": "2604.03198", "title": "The Eleventh NTIRE 2026 Efficient Super-Resolution Challenge Report", "abstract": "This paper reviews the NTIRE 2026 challenge on efficient single-image super-resolution with a focus on the proposed solutions and results. The aim of this challenge is to devise a network that reduces one or several aspects, such as runtime, parameters, and FLOPs, while maintaining PSNR of around 26.90 dB on the DIV2K_LSDIR_valid dataset, and 26.99 dB on the DIV2K_LSDIR_test dataset. The challenge had 95 registered participants, and 15 teams made valid submissions. They gauge the state-of-the-art results for efficient single-image super-resolution.", "date": "2026-04-03", "categories": []}
{"id": "2512.10498", "title": "Robust Shape from Focus via Multiscale Directional Dilated Laplacian and Recurrent Network", "abstract": "Shape-from-Focus (SFF) is a passive depth estimation technique that infers scene depth by analyzing focus variations in a focal stack. Most recent deep learning-based SFF methods typically operate in two stages: first, they extract focus volumes (a per pixel representation of focus likelihood across the focal stack) using heavy feature encoders; then, they estimate depth via a simple one-step aggregation technique that often introduces artifacts and amplifies noise in the depth map. To address these issues, we propose a hybrid framework. Our method computes multi-scale focus volumes traditionally using handcrafted Directional Dilated Laplacian (DDL) kernels, which capture long-range and directional focus variations to form robust focus volumes. These focus volumes are then fed into a lightweight, multi-scale GRU-based depth extraction module that iteratively refines an initial depth estimate at a lower resolution for computational efficiency. Finally, a learned convex upsampling module within our recurrent network reconstructs high-resolution depth maps while preserving fine scene details and sharp boundaries. Extensive experiments on both synthetic and real-world datasets demonstrate that our approach outperforms state-of-the-art deep learning and traditional methods, achieving superior accuracy and generalization across diverse focal conditions.", "date": "2025-12-11", "categories": []}
{"id": "2604.01542", "title": "Universal computational thermal imaging overcoming the ghosting effect", "abstract": "Thermal imaging is crucial for night vision but fundamentally hampered by the ghosting effect, a loss of detailed texture in cluttered photon streams. While conventional ghosting mitigation has relied on data post-processing, the recent breakthrough in heat-assisted detection and ranging (HADAR) opens a promising frontier for hyperspectral computational thermal imaging that produces night vision with day-like visibility. However, universal anti-ghosting imaging remains elusive, as state-of-the-art HADAR applies only to limited scenes with uniform materials, whereas material non-uniformity is ubiquitous in the real world. Here, we propose a universal computational thermal imaging framework, TAG (thermal anti-ghosting), to address material non-uniformity and overcome ghosting for high-fidelity night vision. TAG takes hyperspectral photon streams for nonparametric texture recovery, enabling our experimental demonstration of unprecedented expression recovery in thus-far-elusive ghostly human faces -- the archetypal, long-recognized ghosting phenomenon. Strikingly, TAG not only universally outperforms HADAR across various scenes, but also reveals the influence of material non-uniformity, shedding light on HADAR's effectiveness boundary. We extensively test facial texture and expression recovery across day and night, and demonstrate, for the first time, thermal 3D topological alignment and mood detection. This work establishes a universal foundation for high-fidelity computational night vision, with potential applications in autonomous navigation, reconnaissance, healthcare, and wildlife monitoring.", "date": "2026-04-02", "categories": []}
{"id": "2604.01603", "title": "Towards Minimal Focal Stack in Shape from Focus", "abstract": "Shape from Focus (SFF) is a depth reconstruction technique that estimates scene structure from focus variations observed across a focal stack, that is, a sequence of images captured at different focus settings. A key limitation of SFF methods is their reliance on densely sampled, large focal stacks, which limits their practical applicability. In this study, we propose a focal stack augmentation that enables SFF methods to estimate depth using a reduced stack of just two images, without sacrificing precision. We introduce a simple yet effective physics-based focal stack augmentation that enriches the stack with two auxiliary cues: an all-in-focus (AiF) image estimated from two input images, and Energy-of-Difference (EOD) maps, computed as the energy of differences between the AiF and input images. Furthermore, we propose a deep network that computes a deep focus volume from the augmented focal stacks and iteratively refines depth using convolutional Gated Recurrent Units (ConvGRUs) at multiple scales. Extensive experiments on both synthetic and real-world datasets demonstrate that the proposed augmentation benefits existing state-of-the-art SFF models, enabling them to achieve comparable accuracy. The results also show that our approach maintains state-of-the-art performance with a minimal stack size.", "date": "2026-04-02", "categories": []}
{"id": "2604.00225", "title": "Pupil Design for Computational Wavefront Estimation", "abstract": "Establishing a precise connection between imaged intensity and the incident wavefront is essential for emerging applications in adaptive optics, holography, computational microscopy, and non-line-of-sight imaging. While prior work has shown that breaking symmetries in pupil design enables wavefront recovery from a single intensity measurement, there is little guidance on how to design a pupil that improves wavefront estimation. In this work we introduce a quantitative asymmetry metric to bridge this gap and, through an extensive empirical study and supporting analysis, demonstrate that increasing asymmetry enhances wavefront recoverability. We analyze the trade-offs in pupil design, and the impact on light throughput along with performance in noise. Both large-scale simulations and optical bench experiments are carried out to support our findings.", "date": "2026-03-31", "categories": []}
{"id": "2604.00250", "title": "PRISM: Differentiable Analysis-by-Synthesis for Fixel Recovery in Diffusion MRI", "abstract": "Diffusion MRI microstructure fitting is nonconvex and often performed voxelwise, which limits fiber peak recovery in narrow crossings. This work introduces PRISM, a differentiable analysis-by-synthesis framework that fits an explicit multi-compartment forward model end-to-end over spatial patches. The model combines cerebrospinal fluid (CSF), gray matter, up to K white-matter fiber compartments (stick-and-zeppelin), and a restricted compartment, with explicit fiber directions and soft model selection via repulsion and sparsity priors. PRISM supports a fast MSE objective and a Rician negative log-likelihood (NLL) that jointly learns sigma without oracle information. A lightweight nuisance calibration module (smooth bias field and per-measurement scale/offset) is included for robustness and regularized to identity in clean-data tests. On synthetic crossing-fiber data (SNR=30; five methods, 16 crossing angles), PRISM achieves 3.5 degrees best-match angular error with 95% recall, which is 1.9x lower than the best baseline (MSMT-CSD, 6.8 degrees, 83% recall); in NLL mode with learned sigma, error drops to 2.3 degrees with 99% recall, resolving crossings down to 20 degrees. On the DiSCo1 phantom (NLL mode), PRISM improves connectivity correlation over CSD baselines at all four tracking angles (best r=.934 at 25 degrees vs. .920 for MSMT-CSD). Whole-brain HCP fitting (~741k voxels, MSE mode) completes in ~12 min on a single GPU with near-identical results across random seeds.", "date": "2026-03-31", "categories": []}
{"id": "2508.09655", "title": "Noise-adapted Neural Operator for Robust Non-Line-of-Sight Imaging", "abstract": "This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. Computational imaging, especially non-line-of-sight (NLOS) imaging, the extraction of information from obscured or hidden scenes is achieved through the utilization of indirect light signals resulting from multiple reflections or scattering. The inherently weak nature of these signals, coupled with their susceptibility to noise, necessitates the integration of physical processes to ensure accurate reconstruction. This paper presents a parameterized inverse problem framework tailored for large-scale linear problems in 3D imaging reconstruction. Initially, a noise estimation module is employed to adaptively assess the noise levels present in transient data. Subsequently, a parameterized neural operator is developed to approximate the inverse mapping, facilitating end-to-end rapid image reconstruction. Our 3D image reconstruction framework, grounded in operator learning, is constructed through deep algorithm unfolding, which not only provides commendable model interpretability but also enables dynamic adaptation to varying noise levels in the acquired data, thereby ensuring consistently robust and accurate reconstruction outcomes. Furthermore, we introduce a novel method for the fusion of global and local spatiotemporal data features. By integrating structural and detailed information, this method significantly enhances both accuracy and robustness. Comprehensive numerical experiments conducted on both simulated and real datasets substantiate the efficacy of the proposed method. It demonstrates remarkable performance with fast scanning data and sparse illumination point data, offering a viable solution for NLOS imaging in complex scenarios.", "date": "2025-08-13", "categories": []}
{"id": "2509.10388", "title": "VT-Intrinsic: Physics-Based Decomposition of Reflectance and Shading using a Single Visible-Thermal Image Pair", "abstract": "Decomposing a scene into its reflectance and shading is a challenge due to the lack of extensive ground-truth data for real-world scenes. We introduce a novel physics-based approach for intrinsic image decomposition using a pair of visible and thermal images. We leverage the principle that light not reflected from an opaque surface is absorbed and detected as heat by a thermal camera. This allows us to relate the ordinalities (or relative magnitudes) between visible and thermal image intensities to the ordinalities of shading and reflectance. The ordinalities enable dense self-supervision of an optimizing neural network to recover shading and reflectance. We perform quantitative evaluations with known reflectance and shading under natural and artificial lighting, and qualitative experiments across diverse scenes. The results demonstrate superior performance over both physics-based and recent learning-based methods, providing a path toward scalable real-world data curation with supervision.", "date": "2025-09-12", "categories": []}
{"id": "2603.27516", "title": "SGS-Intrinsic: Semantic-Invariant Gaussian Splatting for Sparse-View Indoor Inverse Rendering", "abstract": "We present SGS-Intrinsic, an indoor inverse rendering framework that works well for sparse-view images. Unlike existing 3D Gaussian Splatting (3DGS) based methods that focus on object-centric reconstruction and fail to work under sparse view settings, our method allows to achieve high-quality geometry reconstruction and accurate disentanglement of material and illumination. The core idea is to construct a dense and geometry-consistent Gaussian semantic field guided by semantic and geometric priors, providing a reliable foundation for subsequent inverse rendering. Building upon this, we perform material-illumination disentanglement by combining a hybrid illumination model and material prior to effectively capture illumination-material interactions. To mitigate the impact of cast shadows and enhance the robustness of material recovery, we introduce illumination-invariant material constraint together with a deshadowing model. Extensive experiments on benchmark datasets show that our method consistently improves both reconstruction fidelity and inverse rendering quality over existing 3DGS-based inverse rendering approaches. Our code is available at https://github.com/GrumpySloths/SGS_Intrinsic.github.io.", "date": "2026-03-29", "categories": []}
{"id": "2603.29438", "title": "Polyhedral Unmixing: Bridging Semantic Segmentation with Hyperspectral Unmixing via Polyhedral-Cone Partitioning", "abstract": "Semantic segmentation and hyperspectral unmixing are two central problems in spectral image analysis. The former assigns each pixel a discrete label corresponding to its material class, whereas the latter estimates pure material spectra, called endmembers, and, for each pixel, a vector representing material abundances in the observed scene. Despite their complementarity, these two problems are usually addressed independently. This paper aims to bridge these two lines of work by formally showing that, under the linear mixing model, pixel classification by dominant materials induces polyhedral-cone regions in the spectral space. We leverage this fundamental property to propose a direct segmentation-to-unmixing pipeline that performs blind hyperspectral unmixing from any semantic segmentation by constructing a polyhedral-cone partition of the space that best fits the labeled pixels. Signed distances from pixels to the estimated regions are then computed, linearly transformed via a change of basis in the distance space, and projected onto the probability simplex, yielding an initial abundance estimate. This estimate is used to extract endmembers and recover final abundances via matrix pseudo-inversion. Because the segmentation method can be freely chosen, the user gains explicit control over the unmixing process, while the rest of the pipeline remains essentially deterministic and lightweight. Beyond improving interpretability, experiments on three real datasets demonstrate the effectiveness of the proposed approach when associated with appropriate clustering algorithms, and show consistent improvements over recent deep and non-deep state-of-the-art methods. The code is available at: https://github.com/antoine-bottenmuller/polyhedral-unmixing", "date": "2026-03-31", "categories": []}
{"id": "2603.29507", "title": "Transmittance-Guided Structure-Texture Decomposition for Nighttime Image Dehazing", "abstract": "Nighttime images captured under hazy conditions suffer from severe quality degradation, including low visibility, color distortion, and reduced contrast, caused by the combined effects of atmospheric scattering, absorption by suspended particles, and non-uniform illumination from artificial light sources. While existing nighttime dehazing methods have achieved partial success, they typically address only a subset of these issues, such as glow suppression or brightness enhancement, without jointly tackling the full spectrum of degradation factors. In this paper, we propose a two-stage nighttime image dehazing framework that integrates transmittance correction with structure-texture layered optimization. In the first stage, we introduce a novel transmittance correction method that establishes boundary-constrained initial transmittance maps and subsequently applies region-adaptive compensation and normalization based on whether image regions correspond to light source areas. A quadratic Gaussian filtering scheme operating in the YUV color space is employed to estimate the spatially varying atmospheric light map. The corrected transmittance map and atmospheric light map are then used in conjunction with an improved nighttime imaging model to produce the initial dehazed image. In the second stage, we propose a STAR-YUV decomposition model that separates the dehazed image into structure and texture layers within the YUV color space. Gamma correction and MSRCR-based color restoration are applied to the structure layer for illumination compensation and color bias correction, while Laplacian-of-Gaussian filtering is applied to the texture layer for detail enhancement. A novel two-phase fusion strategy, comprising nonlinear Retinex-based fusion of the enhanced layers followed by linear blending with the initial dehazing result, yields the final output.", "date": "2026-03-31", "categories": []}
{"id": "2603.29732", "title": "Compressive sensing inspired self-supervised single-pixel imaging", "abstract": "Single-pixel imaging (SPI) is a promising imaging modality with distinctive advantages in strongly perturbed environments. Existing SPI methods lack physical sparsity constraints and overlook the integration of local and global features, leading to severe noise vulnerability, structural distortions and blurred details. To address these limitations, we propose SISTA-Net, a compressive sensing-inspired self-supervised method for single-pixel imaging. SISTA-Net unfolds the Iterative Shrinkage-Thresholding Algorithm (ISTA) into an interpretable network consisting of a data fidelity module and a proximal mapping module. The fidelity module adopts a hybrid CNN-Visual State Space Model (VSSM) architecture to integrate local and global feature modeling, enhancing reconstruction integrity and fidelity. We leverage deep nonlinear networks as adaptive sparse transforms combined with a learnable soft-thresholding operator to impose explicit physical sparsity in the latent domain, enabling noise suppression and robustness to interference even at extremely low sampling rates. Extensive experiments on multiple simulation scenarios demonstrate that SISTA-Net outperforms state-of-the-art methods by 2.6 dB in PSNR. Real-world far-field underwater tests yield a 3.4 dB average PSNR improvement, validating its robust anti-interference capability.", "date": "2026-03-31", "categories": []}
{"id": "2503.08915", "title": "Reconstruct Anything Model: a lightweight general model for computational imaging", "abstract": "Most existing learning-based methods for solving imaging inverse problems can be roughly divided into two classes: iterative algorithms, such as plug-and-play and diffusion methods leveraging pretrained denoisers, and unrolled architectures that are trained end-to-end for specific imaging problems. Iterative methods in the first class are computationally costly and often yield suboptimal reconstruction performance, whereas unrolled architectures are generally problem-specific and require expensive training. In this work, we propose a novel non-iterative, lightweight architecture that incorporates knowledge about the forward operator (acquisition physics and noise parameters) without relying on unrolling. Our model is trained to solve a wide range of inverse problems, such as deblurring, magnetic resonance imaging, computed tomography, inpainting, and super-resolution, and handles arbitrary image sizes and channels, such as grayscale, complex, and color data. The proposed model can be easily adapted to unseen inverse problems or datasets with a few fine-tuning steps (up to a few images) in a self-supervised way, without ground-truth references. Throughout a series of experiments, we demonstrate state-of-the-art performance from medical imaging to low-photon imaging and microscopy. Our code is available at https://github.com/matthieutrs/ram.", "date": "2025-03-11", "categories": []}
{"id": "2509.11334", "title": "Dual Band Thermal Videography: Separating Time-Varying Reflection and Emission Near Ambient Conditions", "abstract": "Long-wave infrared radiation captured by a thermal camera includes (a) emission from an object governed by its temperature and emissivity, and (b) reflected radiation from the surrounding environment. Separating these components is a long-standing challenge in thermography. Even when using multiple bands, the problem is under-determined without priors on emissivity. This difficulty is amplified in near ambient conditions, where emitted and reflected signals are of comparable magnitude. We present a dual-band thermal videography framework that reduces this ambiguity by combining two complementary ideas at a per-pixel level: (i) spectral cues (ratio of emissivity between bands is unknown but fixed), and (ii) temporal cues (object radiation changes smoothly while background radiation changes rapidly). We derive an image formation model and an algorithm to jointly estimate the object's emissivity at each band, and the time-varying object and background temperatures. Experiments with calibrated and uncalibrated emissivities in everyday scenes (e.g., coffee pot heating up, palm print on mirrors, reflections of moving people), demonstrate robust separation and recovery of temperature fields.", "date": "2025-09-14", "categories": []}
{"id": "2511.18471", "title": "Jacobian-aware Posterior Sampling for Inverse Problems", "abstract": "Diffusion models provide powerful generative priors for solving inverse problems by sampling from a posterior distribution conditioned on corrupted measurements. Existing methods primarily follow two paradigms: direct methods, which approximate the likelihood term, and proximal methods, which incorporate intermediate solutions satisfying measurement constraints into the sampling process. We demonstrate that these approaches differ fundamentally in their treatment of the diffusion denoiser's Jacobian within the likelihood term. While this Jacobian encodes critical prior knowledge of the data distribution, training-induced non-idealities can degrade performance in zero-shot settings. In this work, we bridge direct and proximal approaches by proposing a principled Jacobian-Aware Posterior Sampler (JAPS). JAPS leverages the Jacobian's prior knowledge while mitigating its detrimental effects through a corresponding proximal solution, requiring no additional computational cost. Our method enhances reconstruction quality across diverse linear and nonlinear noisy imaging tasks, outperforming existing diffusion-based baselines in perceptual quality while maintaining or improving distortion metrics.", "date": "2025-11-23", "categories": []}
{"id": "2603.26777", "title": "BHCast: Unlocking Black Hole Plasma Dynamics from a Single Blurry Image with Long-Term Forecasting", "abstract": "The Event Horizon Telescope (EHT) delivered the first image of a black hole by capturing the light from its surrounding accretion flow, revealing structure but not dynamics. Simulations of black hole accretion dynamics are essential for interpreting EHT images but costly to generate and impractical for inference. Motivated by this bottleneck, BHCast presents a framework for forecasting black hole plasma dynamics from a single, blurry snapshot, such as those captured by the EHT. At its core, BHCast is a neural model that transforms a static image into forecasted future frames, revealing the underlying dynamics hidden within one snapshot. With a multi-scale pyramid loss, we demonstrate how autoregressive forecasting can simultaneously super-resolve and evolve a blurry frame into a coherent, high-resolution movie that remains stable over long time horizons. From forecasted dynamics, we can then extract interpretable spatio-temporal features, such as pattern speed (rotation rate) and pitch angle. Finally, BHCast uses gradient-boosting trees to recover black hole properties from these plasma features, including the spin and viewing inclination angle. The separation between forecasting and inference provides modular flexibility, interpretability, and robust uncertainty quantification. We demonstrate the effectiveness of BHCast on simulations of two distinct black hole accretion systems, Sagittarius A* and M87*, by testing on simulated frames blurred to EHT resolution and real EHT images of M87*. Ultimately, our methodology establishes a scalable paradigm for solving inverse problems, demonstrating the potential of learned dynamics to unlock insights from resolution-limited scientific data.", "date": "2026-03-24", "categories": []}
{"id": "2603.27357", "title": "Guided Lensless Polarization Imaging", "abstract": "Polarization imaging captures the polarization state of light, revealing information invisible to the human eye yet valuable in domains such as biomedical diagnostics, autonomous driving, and remote sensing. However, conventional polarization cameras are often expensive, bulky, or both, limiting their practical use. Lensless imaging offers a compact, low-cost alternative by replacing the lens with a simple optical element like a diffuser and performing computational reconstruction, but existing lensless polarization systems suffer from limited reconstruction quality. To overcome these limitations, we introduce a RGB-guided lensless polarization imaging system that combines a compact polarization-RGB sensor with an auxiliary, widely available conventional RGB camera providing structural guidance. We reconstruct multi-angle polarization images for each RGB color channel through a two-stage pipeline: a physics-based inversion recovers an initial polarization image, followed by a Transformer-based fusion network that refines this reconstruction using the RGB guidance image from the conventional RGB camera. Our two-stage method significantly improves reconstruction quality and fidelity over lensless-only baselines, generalizes across datasets and imaging conditions, and achieves high-quality real-world results on our physical prototype lensless camera without any fine-tuning.", "date": "2026-03-28", "categories": []}
{"id": "2603.27832", "title": "3-D Representations for Hyperspectral Flame Tomography", "abstract": "Flame tomography is a compelling approach for extracting large amounts of data from experiments via 3-D thermochemical reconstruction. Recent efforts employing neural-network flame representations have suggested improved reconstruction quality compared with classical tomography approaches, but a rigorous quantitative comparison with the same algorithm using a voxel-grid representation has not been conducted. Here, we compare a classical voxel-grid representation with varying regularizers to a continuous neural representation for tomographic reconstruction of a simulated pool fire. The representations are constructed to give temperature and composition as a function of location, and a subsequent ray-tracing step is used to solve the radiative transfer equation to determine the spectral intensity incident on hyperspectral infrared cameras, which is then convolved with an instrument lineshape function. We demonstrate that the voxel-grid approach with a total-variation regularizer reproduces the ground-truth synthetic flame with the highest accuracy for reduced memory intensity and runtime. Future work will explore more representations and under experimental configurations.", "date": "2026-03-29", "categories": []}
{"id": "2603.28658", "title": "Divide and Restore: A Modular Task-Decoupled Framework for Universal Image Restoration", "abstract": "Restoring images affected by various types of degradation, such as noise, blur, or improper exposure, remains a significant challenge in computer vision. While recent trends favor complex monolithic all-in-one architectures, these models often suffer from negative task interference and require extensive joint training cycles on high-end computing clusters. In this paper, we propose a modular, task-decoupled image restoration framework based on an explicit diagnostic routing mechanism. The architecture consists of a lightweight Convolutional Neural Network (CNN) classifier that evaluates the input image and dynamically directs it to a specialized restoration node. A key advantage of this framework is its model-agnostic extensibility: while we demonstrate it using three independent U-Net experts, the system allows for the integration of any restoration method tailored to specific tasks. By isolating reconstruction paths, the framework prevents feature conflicts and significantly reduces training overhead. Unlike monolithic models, adding new degradation types in our framework only requires training a single expert and updating the router, rather than a full system retraining. Experimental results demonstrate that this computationally accessible approach offers a scalable and efficient solution for multi-degradation restoration on standard local hardware. The code will be published upon paper acceptance.", "date": "2026-03-30", "categories": []}
{"id": "2603.25869", "title": "Learning to Recorrupt: Noise Distribution Agnostic Self-Supervised Image Denoising", "abstract": "Self-supervised image denoising methods have traditionally relied on either architectural constraints or specialized loss functions that require prior knowledge of the noise distribution to avoid the trivial identity mapping. Among these, approaches such as Noisier2Noise or Recorrupted2Recorrupted, create training pairs by adding synthetic noise to the noisy images. While effective, these recorruption-based approaches require precise knowledge of the noise distribution, which is often not available. We present Learning to Recorrupt (L2R), a noise distribution-agnostic denoising technique that eliminates the need for knowledge of the noise distribution. Our method introduces a learnable monotonic neural network that learns the recorruption process through a min-max saddle-point objective. The proposed method achieves state-of-the-art performance across unconventional and heavy-tailed noise distributions, such as log-gamma, Laplace, and spatially correlated noise, as well as signal-dependent noise models such as Poisson-Gaussian noise.", "date": "2026-03-26", "categories": []}
{"id": "2603.26181", "title": "GLINT: Modeling Scene-Scale Transparency via Gaussian Radiance Transport", "abstract": "While 3D Gaussian splatting has emerged as a powerful paradigm, it fundamentally fails to model transparency such as glass panels. The core challenge lies in decoupling the intertwined radiance contributions from transparent interfaces and the transmitted geometry observed through the glass. We present GLINT, a framework that models scene-scale transparency through explicit decomposed Gaussian representation. GLINT reconstructs the primary interface and models reflected and transmitted radiance separately, enabling consistent radiance transport. During optimization, GLINT bootstraps transparency localization from geometry-separation cues induced by the decomposition, together with geometry and material priors from a pre-trained video relighting model. Extensive experiments demonstrate consistent improvements over prior methods for reconstructing complex transparent scenes.", "date": "2026-03-27", "categories": []}
{"id": "2603.14023", "title": "High-speed Imaging through Turbulence with Event-based Light Fields", "abstract": "This work introduces and demonstrates the first system capable of imaging fast-moving extended non-rigid objects through strong atmospheric turbulence at high frame rate. Event cameras are a novel sensing architecture capable of estimating high-speed imagery at thousands of frames per second. However, on their own event cameras are unable to disambiguate scene motion from turbulence. In this work, we overcome this limitation using event-based light field cameras: By simultaneously capturing multiple views of a scene, event-based light field cameras and machine learning-based reconstruction algorithms are able to disambiguate motion-induced dynamics, which produce events that are strongly correlated across views, from turbulence-induced dynamics, which produce events that are weakly correlated across view. Tabletop experiments demonstrate event-based light field can overcome strong turbulence while imaging high-speed objects traveling at up to 16,000 pixels per second.", "date": "2026-03-14", "categories": []}
{"id": "2603.24969", "title": "PASDiff: Physics-Aware Semantic Guidance for Joint Real-world Low-Light Face Enhancement and Restoration", "abstract": "Face images captured in real-world low light suffer multiple degradations-low illumination, blur, noise, and low visibility, etc. Existing cascaded solutions often suffer from severe error accumulation, while generic joint models lack explicit facial priors and struggle to resolve clear face structures. In this paper, we propose PASDiff, a Physics-Aware Semantic Diffusion with a training-free manner. To achieve a plausible illumination and color distribution, we leverage inverse intensity weighting and Retinex theory to introduce photometric constraints, thereby reliably recovering visibility and natural chromaticity. To faithfully reconstruct facial details, our Style-Agnostic Structural Injection (SASI) extracts structures from an off-the-shelf facial prior while filtering out its intrinsic photometric biases, seamlessly harmonizing identity features with physical constraints. Furthermore, we construct WildDark-Face, a real-world benchmark of 700 low-light facial images with complex degradations. Extensive experiments demonstrate that PASDiff significantly outperforms existing methods, achieving a superior balance among natural illumination, color recovery, and identity consistency.", "date": "2026-03-26", "categories": []}
{"id": "2603.25054", "title": "Synergistic Event-SVE Imaging for Quantitative Propellant Combustion Diagnostics", "abstract": "Real-time monitoring of high-energy propellant combustion is difficult. Extreme high dynamic range (HDR), microsecond-scale particle motion, and heavy smoke often occur together. These conditions drive saturation, motion blur, and unstable particle extraction in conventional imaging. We present a closed-loop Event--SVE measurement system that couples a spatially variant exposure (SVE) camera with a stereo pair of neuromorphic event cameras. The SVE branch produces HDR maps with an explicit smoke-aware fusion strategy. A multi-cue smoke-likelihood map is used to separate particle emission from smoke scattering, yielding calibrated intensity maps for downstream analysis. The resulting HDR maps also provide the absolute-intensity reference missing in event cameras. This reference is used to suppress smoke-driven event artifacts and to improve particle-state discrimination. Based on the cleaned event observations, a stereo event-based 3D pipeline estimates separation height and equivalent particle size through feature extraction and triangulation (maximum calibration error 0.56%). Experiments on boron-based propellants show multimodal equivalent-radius statistics. The system also captures fast separation transients that are difficult to observe with conventional sensors. Overall, the proposed framework provides a practical, calibration-consistent route to microsecond-resolved 3D combustion measurement under smoke-obscured HDR conditions.", "date": "2026-03-26", "categories": []}
{"id": "2603.25502", "title": "RealRestorer: Towards Generalizable Real-World Image Restoration with Large-Scale Image Editing Models", "abstract": "Image restoration under real-world degradations is critical for downstream tasks such as autonomous driving and object detection. However, existing restoration models are often limited by the scale and distribution of their training data, resulting in poor generalization to real-world scenarios. Recently, large-scale image editing models have shown strong generalization ability in restoration tasks, especially for closed-source models like Nano Banana Pro, which can restore images while preserving consistency. Nevertheless, achieving such performance with those large universal models requires substantial data and computational costs. To address this issue, we construct a large-scale dataset covering nine common real-world degradation types and train a state-of-the-art open-source model to narrow the gap with closed-source alternatives. Furthermore, we introduce RealIR-Bench, which contains 464 real-world degraded images and tailored evaluation metrics focusing on degradation removal and consistency preservation. Extensive experiments demonstrate our model ranks first among open-source methods, achieving state-of-the-art performance.", "date": "2026-03-26", "categories": []}
{"id": "2603.25636", "title": "Designing Any Imaging System from Natural Language: Agent-Constrained Composition over a Finite Primitive Basis", "abstract": "Designing a computational imaging system -- selecting operators, setting parameters, validating consistency -- requires weeks of specialist effort per modality, creating an expertise bottleneck that excludes the\n  broader scientific community from prototyping imaging instruments. We introduce spec.md, a structured specification format, and three autonomous agents -- Plan, Judge, and Execute -- that translate a one-sentence\n  natural-language description into a validated forward model with bounded reconstruction error. A design-to-real error theorem decomposes total reconstruction error into five independently bounded terms, each linked\n  to a corrective action. On 6 real-data modalities spanning all 5 carrier families, the automated pipeline matches expert-library quality (98.1 +/- 4.2%). Ten novel designs -- composing primitives into chains from 3D\n  to 5D -- demonstrate compositional reach beyond any single-modality tool.", "date": "2026-03-26", "categories": []}
{"id": "2510.06020", "title": "RamPINN: Recovering Raman Spectra From Coherent Anti-Stokes Spectra Using Embedded Physics", "abstract": "Transferring the recent advancements in deep learning into scientific disciplines is hindered by the lack of the required large-scale datasets for training. We argue that in these knowledge-rich domains, the established body of scientific theory provides reliable inductive biases in the form of governing physical laws. We address the ill-posed inverse problem of recovering Raman spectra from noisy Coherent Anti-Stokes Raman Scattering (CARS) measurements, as the true Raman signal here is suppressed by a dominating non-resonant background. We propose RamPINN, a model that learns to recover Raman spectra from given CARS spectra. Our core methodological contribution is a physics-informed neural network that utilizes a dual-decoder architecture to disentangle resonant and non-resonant signals. This is done by enforcing the Kramers-Kronig causality relations via a differentiable Hilbert transform loss on the resonant and a smoothness prior on the non-resonant part of the signal. Trained entirely on synthetic data, RamPINN demonstrates strong zero-shot generalization to real-world experimental data, explicitly closing this gap and significantly outperforming existing baselines. Furthermore, we show that training with these physics-based losses alone, without access to any ground-truth Raman spectra, still yields competitive results. This work highlights a broader concept: formal scientific rules can act as a potent inductive bias, enabling robust, self-supervised learning in data-limited scientific domains.", "date": "2025-10-07", "categories": []}
{"id": "2310.03956", "title": "Gradient Descent Provably Solves Nonlinear Tomographic Reconstruction", "abstract": "In computed tomography (CT), the forward model consists of a linear Radon transform followed by an exponential nonlinearity based on the attenuation of light according to the Beer-Lambert Law. Conventional reconstruction often involves inverting this nonlinearity and then solving a linear inverse problem. However, this nonlinear measurement preprocessing is poorly conditioned in the vicinity of high-density materials, such as metal. This preprocessing makes CT reconstruction methods numerically sensitive and susceptible to artifacts near high-density regions. In this paper, we study a technique where the signal is directly reconstructed from raw measurements through the nonlinear forward model. Though this optimization is nonconvex, we show that gradient descent provably converges to the global optimum at a geometric rate, perfectly reconstructing the underlying signal with a near minimal number of random measurements. We also prove similar results in the under-determined setting where the number of measurements is significantly smaller than the dimension of the signal. This is achieved by enforcing prior structural information about the signal through constraints on the optimization variables. We illustrate the benefits of direct nonlinear CT reconstruction with cone-beam CT experiments on synthetic and real 3D volumes, in which metal artifacts are reduced compared to standard linear reconstruction methods. Our experiments also demonstrate that logarithmic preprocessing alone is sufficient to produce metal artifacts, even in the absence of other causes such as beam hardening.", "date": "2023-10-06", "categories": []}
{"id": "2506.16210", "title": "From Coarse to Continuous: Progressive Refinement Implicit Neural Representation for Motion-Robust Anisotropic MRI Reconstruction", "abstract": "In motion-robust magnetic resonance imaging (MRI), slice-to-volume reconstruction is critical for recovering anatomically consistent 3D brain volumes from 2D slices, especially under accelerated acquisitions or patient motion. However, this task remains challenging due to hierarchical structural disruptions. It includes local detail loss from k-space undersampling, global structural aliasing caused by motion, and volumetric anisotropy. Therefore, we propose a progressive refinement implicit neural representation (PR-INR) framework. Our PR-INR unifies motion correction, structural refinement, and volumetric synthesis within a geometry-aware coordinate space. Specifically, a motion-aware diffusion module is first employed to generate coarse volumetric reconstructions that suppress motion artifacts and preserve global anatomical structures. Then, we introduce an implicit detail restoration module that performs residual refinement by aligning spatial coordinates with visual features. It corrects local structures and enhances boundary precision. Further, a voxel continuous-aware representation module represents the image as a continuous function over 3D coordinates. It enables accurate inter-slice completion and high-frequency detail recovery. We evaluate PR-INR on five public MRI datasets under various motion conditions (3% and 5% displacement), undersampling rates (4x and 8x) and slice resolutions (scale = 5). Experimental results demonstrate that PR-INR outperforms state-of-the-art methods in both quantitative reconstruction metrics and visual quality. It further shows generalization and robustness across diverse unseen domains.", "date": "2025-06-19", "categories": []}
{"id": "2603.23071", "title": "PolarAPP: Beyond Polarization Demosaicking for Polarimetric Applications", "abstract": "Polarimetric imaging enables advanced vision applications such as normal estimation and de-reflection by capturing unique surface-material interactions. However, existing applications (alternatively called downstream tasks) rely on datasets constructed by naively regrouping raw measurements from division-of-focal-plane sensors, where pixels of the same polarization angle are extracted and aligned into sparse images without proper demosaicking. This reconstruction strategy results in suboptimal, incomplete targets that limit downstream performance. Moreover, current demosaicking methods are task-agnostic, optimizing only for photometric fidelity rather than utility in downstream tasks. Towards this end, we propose PolarAPP, the first framework to jointly optimize demosaicking and its downstream tasks. PolarAPP introduces a feature alignment mechanism that semantically aligns the representations of demosaicking and downstream networks via meta-learning, guiding the reconstruction to be task-aware. It further employs an equivalent imaging constraint for demosaicking training, enabling direct regression to physically meaningful outputs without relying on rearranged data. Finally, a task-refinement stage fine-tunes the task network using the stable demosaicking front-end to further enhance accuracy. Extensive experimental results demonstrate that PolarAPP outperforms existing methods in both demosaicking quality and downstream performance. Code is available upon acceptance.", "date": "2026-03-24", "categories": []}
{"id": "2401.04317", "title": "WiFi-GEN: High-Resolution Indoor Imaging from WiFi Signals Using Generative AI", "abstract": "Indoor imaging is a critical task for robotics and internet-ofthings. WiFi as an omnipresent signal is a promising candidate for carrying out passive imaging and synchronizing the up-to-date information to all connected devices. This is the first research work to consider WiFi indoor imaging as a multi-modal image generation task that converts the measured WiFi power into a high-resolution indoor image. Our proposedWiFi-GEN network achieves a shape reconstruction accuracy that is 275% of that achieved by physical model-based inversion methods. Additionally, the Frechet Inception Distance score has been significantly reduced by 82%. To examine the effectiveness of models for this task, the first large-scale dataset is released containing 80,000 pairs of WiFi signal and imaging target. Our model absorbs challenges for the model-based methods including the nonlinearity, ill-posedness and non-certainty into massive parameters of our generative AI network. The network is also designed to best fit measured WiFi signals and the desired imaging output. Code: https://github.com/CNFightingSjy/WiFiGEN", "date": "2024-01-09", "categories": []}
{"id": "2407.03239", "title": "Solving the inverse problem of microscopy deconvolution with a residual Beylkin-Coifman-Rokhlin neural network", "abstract": "Optic deconvolution in light microscopy (LM) refers to recovering the object details from images, revealing the ground truth of samples. Traditional explicit methods in LM rely on the point spread function (PSF) during image acquisition. Yet, these approaches often fall short due to inaccurate PSF models and noise artifacts, hampering the overall restoration quality. In this paper, we approached the optic deconvolution as an inverse problem. Motivated by the nonstandard-form compression scheme introduced by Beylkin, Coifman, and Rokhlin (BCR), we proposed an innovative physics-informed neural network Multi-Stage Residual-BCR Net (m-rBCR) to approximate the optic deconvolution. We validated the m-rBCR model on four microscopy datasets - two simulated microscopy datasets from ImageNet and BioSR, real dSTORM microscopy images, and real widefield microscopy images. In contrast to the explicit deconvolution methods (e.g. Richardson-Lucy) and other state-of-the-art NN models (U-Net, DDPM, CARE, DnCNN, ESRGAN, RCAN, Noise2Noise, MPRNet, and MIMO-U-Net), the m-rBCR model demonstrates superior performance to other candidates by PSNR and SSIM in two real microscopy datasets and the simulated BioSR dataset. In the simulated ImageNet dataset, m-rBCR ranks the second-best place (right after MIMO-U-Net). With the backbone from the optical physics, m-rBCR exploits the trainable parameters with better performances (from ~30 times fewer than the benchmark MIMO-U-Net to ~210 times than ESRGAN). This enables m-rBCR to achieve a shorter runtime (from ~3 times faster than MIMO-U-Net to ~300 times faster than DDPM). To summarize, by leveraging physics constraints our model reduced potentially redundant parameters significantly in expertise-oriented NN candidates and achieved high efficiency with superior performance.", "date": "2024-07-03", "categories": []}
{"id": "2512.08441", "title": "Leveraging Multispectral Sensors for Color Correction in Mobile Cameras", "abstract": "Recent advances in snapshot multispectral (MS) imaging have enabled compact, low-cost spectral sensors for consumer and mobile devices. By capturing richer spectral information than conventional RGB sensors, these systems can enhance key imaging tasks, including color correction. However, most existing methods treat the color correction pipeline in separate stages, often discarding MS data early in the process. We propose a unified, learning-based framework that performs end-to-end color correction and jointly leverages data from a high-resolution RGB sensor and an auxiliary low-resolution MS sensor. Our approach integrates the full pipeline within a single model, producing coherent and color-accurate outputs. We demonstrate the flexibility and generality of our framework by refactoring two different state-of-the-art image-to-image architectures. To support training and evaluation, we construct a dedicated dataset by aggregating and repurposing publicly available spectral datasets, rendering under multiple RGB camera sensitivities. Extensive experiments show that our approach improves color accuracy and stability, reducing error by up to 50% compared to RGB-only and MS-driven baselines. Code, models and dataset available at: https://lucacogo.github.io/Mobile-Spectral-CC/.", "date": "2025-12-09", "categories": []}
{"id": "2601.02763", "title": "ClearAIR: A Human-Visual-Perception-Inspired All-in-One Image Restoration", "abstract": "All-in-One Image Restoration (AiOIR) has advanced significantly, offering promising solutions for complex real-world degradations. However, most existing approaches rely heavily on degradation-specific representations, often resulting in oversmoothing and artifacts. To address this, we propose ClearAIR, a novel AiOIR framework inspired by Human Visual Perception (HVP) and designed with a hierarchical, coarse-to-fine restoration strategy. First, leveraging the global priority of early HVP, we employ a Multimodal Large Language Model (MLLM)-based Image Quality Assessment (IQA) model for overall evaluation. Unlike conventional IQA, our method integrates cross-modal understanding to more accurately characterize complex, composite degradations. Building upon this overall assessment, we then introduce a region awareness and task recognition pipeline. A semantic cross-attention, leveraging semantic guidance unit, first produces coarse semantic prompts. Guided by this regional context, a degradation-aware module implicitly captures region-specific degradation characteristics, enabling more precise local restoration. Finally, to recover fine details, we propose an internal clue reuse mechanism. It operates in a self-supervised manner to mine and leverage the intrinsic information of the image itself, substantially enhancing detail restoration. Experimental results show that ClearAIR achieves superior performance across diverse synthetic and real-world datasets.", "date": "2026-01-06", "categories": []}
{"id": "2603.03710", "title": "MPFlow: Multi-modal Posterior-Guided Flow Matching for Zero-Shot MRI Reconstruction", "abstract": "Zero-shot MRI reconstruction relies on generative priors, but single-modality unconditional priors produce hallucinations under severe ill-posedness. In many clinical workflows, complementary MRI acquisitions (e.g. high-quality structural scans) are routinely available, yet existing reconstruction methods lack mechanisms to leverage this additional information. We propose MPFlow, a zero-shot multi-modal reconstruction framework built on rectified flow that incorporates auxiliary MRI modalities at inference time without retraining the generative prior to improve anatomical fidelity. Cross-modal guidance is enabled by our proposed self-supervised pretraining strategy, Patch-level Multi-modal MR Image Pretraining (PAMRI), which learns shared representations across modalities. Sampling is jointly guided by data consistency and cross-modal feature alignment using pre-trained PAMRI, systematically suppressing intrinsic and extrinsic hallucinations. Extensive experiments on HCP and BraTS show that MPFlow matches diffusion baselines on image quality using only 20% of sampling steps while reducing tumor hallucinations by more than 15% (segmentation dice score). This demonstrates that cross-modal guidance enables more reliable and efficient zero-shot MRI reconstruction.", "date": "2026-03-04", "categories": []}
{"id": "2603.20611", "title": "GaussianPile: A Unified Sparse Gaussian Splatting Framework for Slice-based Volumetric Reconstruction", "abstract": "Slice-based volumetric imaging is widely applied and it demands representations that compress aggressively while preserving internal structure for analysis. We introduce GaussianPile, unifying 3D Gaussian splatting with an imaging system-aware focus model to address this challenge. Our proposed method introduces three key innovations: (i) a slice-aware piling strategy that positions anisotropic 3D Gaussians to model through-slice contributions, (ii) a differentiable projection operator that encodes the finite-thickness point spread function of the imaging acquisition system, and (iii) a compact encoding and joint optimization pipeline that simultaneously reconstructs and compresses the Gaussian sets. Our CUDA-based design retains the compression and real-time rendering efficiency of Gaussian primitives while preserving high-frequency internal volumetric detail. Experiments on microscopy and ultrasound datasets demonstrate that our method reduces storage and reconstruction cost, sustains diagnostic fidelity, and enables fast 2D visualization, along with 3D voxelization. In practice, it delivers high-quality results in as few as 3 minutes, up to 11x faster than NeRF-based approaches, and achieves consistent 16x compression over voxel grids, offering a practical path to deployable compression and exploration of slice-based volumetric datasets.", "date": "2026-03-21", "categories": []}
{"id": "2603.20669", "title": "ToFormer: Towards Large-scale Scenario Depth Completion for Lightweight ToF Camera", "abstract": "Time-of-Flight (ToF) cameras possess compact design and high measurement precision to be applied to various robot tasks. However, their limited sensing range restricts deployment in large-scale scenarios. Depth completion has emerged as a potential solution to expand the sensing range of ToF cameras, but existing research lacks dedicated datasets and struggles to generalize to ToF measurements. In this paper, we propose a full-stack framework that enables depth completion in large-scale scenarios for short-range ToF cameras. First, we construct a multi-sensor platform with a reconstruction-based pipeline to collect real-world ToF samples with dense large-scale ground truth, yielding the first LArge-ScalE scenaRio ToF depth completion dataset (LASER-ToF). Second, we propose a sensor-aware depth completion network that incorporates a novel 3D branch with a 3D-2D Joint Propagation Pooling (JPP) module and Multimodal Cross-Covariance Attention (MXCA), enabling effective modeling of long-range relationships and efficient 3D-2D fusion under non-uniform ToF depth sparsity. Moreover, our network can utilize the sparse point cloud from visual SLAM as a supplement to ToF depth to further improve prediction accuracy. Experiments show that our method achieves an 8.6% lower mean absolute error than the second-best method, while maintaining lightweight design to support onboard deployment. Finally, to verify the system's applicability on real robots, we deploy proposed method on a quadrotor at a 10Hz runtime, enabling reliable large-scale mapping and long-range planning in challenging environments for short-range ToF cameras.", "date": "2026-03-21", "categories": []}
{"id": "2603.20690", "title": "MFSR: MeanFlow Distillation for One Step Real-World Image Super Resolution", "abstract": "Diffusion- and flow-based models have advanced Real-world Image Super-Resolution (Real-ISR), but their multi-step sampling makes inference slow and hard to deploy. One-step distillation alleviates the cost, yet often degrades restoration quality and removes the option to refine with more steps. We present Mean Flows for Super-Resolution (MFSR), a new distillation framework that produces photorealistic results in a single step while still allowing an optional few-step path for further improvement. Our approach uses MeanFlow as the learning target, enabling the student to approximate the average velocity between arbitrary states of the Probability Flow ODE (PF-ODE) and effectively capture the teacher's dynamics without explicit rollouts. To better leverage pretrained generative priors, we additionally improve original MeanFlow's Classifier-Free Guidance (CFG) formulation with teacher CFG distillation strategy, which enhances restoration capability and preserves fine details. Experiments on both synthetic and real-world benchmarks demonstrate that MFSR achieves efficient, flexible, and high-quality super-resolution, delivering results on par with or even better than multi-step teachers while requiring much lower computational cost.", "date": "2026-03-21", "categories": []}
{"id": "2603.21192", "title": "DSCSNet: A Dynamic Sparse Compression Sensing Network for Closely-Spaced Infrared Small Target Unmixing", "abstract": "Due to the limitations of optical lens focal length and detector resolution, distant clustered infrared small targets often appear as mixed spots. The Close Small Object Unmixing (CSOU) task aims to recover the number, sub-pixel positions, and radiant intensities of individual targets from these spots, which is a highly ill-posed inverse problem. Existing methods struggle to balance the rigorous sparsity guarantees of model-driven approaches and the dynamic scene adaptability of data-driven methods. To address this dilemma, this paper proposes a Dynamic Sparse Compressed Sensing Network (DSCSNet), a deep-unfolded network that couples the Alternating Direction Method of Multipliers (ADMM) with learnable parameters. Specifically, we embed a strict $\\ell_1$-norm sparsity constraint into the auxiliary variable update step of ADMM to replace the traditional $\\ell_2$-norm smoothness-promoting terms, which effectively preserves the discrete energy peaks of small targets. We also integrate a self-attention-based dynamic thresholding mechanism into the reconstruction stage, which adaptively adjusts the sparsification intensity using the sparsity-enhanced information from the iterative process. These modules are jointly optimized end-to-end across the three iterative steps of ADMM. Retaining the physical logic of compressed sensing, DSCSNet achieves robust sparsity induction and scene adaptability, thus enhancing the unmixing accuracy and generalization in complex infrared scenarios. Extensive experiments on the synthetic infrared dataset CSIST-100K demonstrate that DSCSNet outperforms state-of-the-art methods in key metrics such as CSO-mAP and sub-pixel localization error.", "date": "2026-03-22", "categories": []}
{"id": "2603.21784", "title": "Dynamic Exposure Burst Image Restoration", "abstract": "Burst image restoration aims to reconstruct a high-quality image from burst images, which are typically captured using manually designed exposure settings. Although these exposure settings significantly influence the final restoration performance, the problem of finding optimal exposure settings has been overlooked. In this paper, we present Dynamic Exposure Burst Image Restoration (DEBIR), a novel burst image restoration pipeline that enhances restoration quality by dynamically predicting exposure times tailored to the shooting environment. In our pipeline, Burst Auto-Exposure Network (BAENet) estimates the optimal exposure time for each burst image based on a preview image, as well as motion magnitude and gain. Subsequently, a burst image restoration network reconstructs a high-quality image from burst images captured using these optimal exposure times. For training, we introduce a differentiable burst simulator and a three-stage training strategy. Our experiments demonstrate that our pipeline achieves state-of-the-art restoration quality. Furthermore, we validate the effectiveness of our approach on a real-world camera system, demonstrating its practicality.", "date": "2026-03-23", "categories": []}
{"id": "2506.21349", "title": "Electromagnetic Inverse Scattering from a Single Transmitter", "abstract": "Electromagnetic Inverse Scattering Problems (EISP) seek to reconstruct relative permittivity from scattered fields and are fundamental to applications like medical imaging. This inverse process is inherently ill-posed and highly nonlinear, making it particularly challenging, especially under sparse transmitter setups, e.g., with only one transmitter. While recent machine learning-based approaches have shown promising results, they often rely on time-consuming, case-specific optimization and perform poorly under sparse transmitter setups. To address these limitations, we revisit EISP from a data-driven perspective. The scarcity of transmitters leads to an insufficient amount of measured data, which fails to capture adequate physical information for stable inversion. Accordingly, we propose a fully end-to-end and data-driven framework that predicts the relative permittivity of scatterers from measured fields, leveraging data distribution priors to compensate for the incomplete information from sparse measurements. This design enables data-driven training and feed-forward prediction of relative permittivity while maintaining strong robustness to transmitter sparsity. Extensive experiments show that our method outperforms state-of-the-art approaches in reconstruction accuracy and robustness. Notably, we demonstrate, for the first time, high-quality reconstruction from a single transmitter. This work advances practical electromagnetic imaging by providing a new, cost-effective paradigm to inverse scattering. Code and models are released at https://gomenei.github.io/SingleTX-EISP/.", "date": "2025-06-26", "categories": []}
{"id": "2412.09465", "title": "OFTSR: One-Step Flow for Image Super-Resolution with Tunable Fidelity-Realism Trade-offs", "abstract": "Recent advances in diffusion and flow-based generative models have demonstrated remarkable success in image restoration tasks, achieving superior perceptual quality compared to traditional deep learning approaches. However, these methods either require numerous sampling steps to generate high-quality images, resulting in significant computational overhead, or rely on common model distillation, which usually imposes a fixed fidelity-realism trade-off and thus lacks flexibility. In this paper, we introduce OFTSR, a novel flow-based framework for one-step image super-resolution that can produce outputs with tunable levels of fidelity and realism. Our approach first trains a conditional flow-based super-resolution model to serve as a teacher model. We then distill this teacher model by applying a specialized constraint. Specifically, we force the predictions from our one-step student model for same input to lie on the same sampling ODE trajectory of the teacher model. This alignment ensures that the student model's single-step predictions from initial states match the teacher's predictions from a closer intermediate state. Through extensive experiments on datasets including FFHQ (256$\\times$256), DIV2K, and ImageNet (256$\\times$256), we demonstrate that OFTSR achieves state-of-the-art performance for one-step image super-resolution, while having the ability to flexibly tune the fidelity-realism trade-off. Codes: \\href{https://github.com/yuanzhi-zhu/OFTSR}{https://github.com/yuanzhi-zhu/OFTSR}.", "date": "2024-12-12", "categories": []}
{"id": "2508.05321", "title": "Unsupervised Learning for Inverse Problems in Computed Tomography", "abstract": "Assume you encounter an inverse problem that shall be solved for a large number of data, but no ground-truth data is available. To emulate this encounter, in this study, we assume it is unknown how to solve the imaging problem of Computed Tomography (CT). An unsupervised deep learning approach is introduced, that leverages the inherent similarities between deep neural network training, deep image prior (DIP) and unrolled optimization schemes. We demonstrate the feasibility of reconstructing images from measurement data by pure network inference, without relying on ground-truth images in the training process or additional gradient steps for unseen samples. Our method is evaluated on the two-dimensional 2DeteCT dataset, showcasing superior performance in terms of mean squared error (MSE) and structural similarity index (SSIM) compared to traditional filtered backprojection (FBP) and maximum likelihood (ML) reconstruction techniques as well as similar performance compared to a supervised DL reconstruction. Additionally, our approach significantly reduces reconstruction time, making it a promising alternative for real-time medical imaging applications. Future work will focus on extending this methodology for adaptability of the projection geometry and other use-cases in medical imaging.", "date": "2025-08-07", "categories": []}
{"id": "2603.16549", "title": "Bridging the Simulation-to-Reality Gap in Electron Microscope Calibration via VAE-EM Estimation", "abstract": "Electron microscopy has enabled many scientific breakthroughs across multiple fields. A key challenge is the tuning of microscope parameters based on images to overcome optical aberrations that deteriorate image quality. This calibration problem is challenging due to the high-dimensional and noisy nature of the diagnostic images, and the fact that optimal parameters cannot be identified from a single image. We tackle the calibration problem for Scanning Transmission Electron Microscopes (STEM) by employing variational autoencoders (VAEs), trained on simulated data, to learn low-dimensional representations of images, whereas most existing methods extract only scalar values. We then simultaneously estimate the model that maps calibration parameters to encoded representations and the optimal calibration parameters using an expectation maximization (EM) approach. This joint estimation explicitly addresses the simulation-to-reality gap inherent in data-driven methods that train on simulated data from a digital twin. We leverage the known symmetry property of the optical system to establish global identifiability of the joint estimation problem, ensuring that a unique optimum exists. We demonstrate that our approach is substantially faster and more consistent than existing methods on a real STEM, achieving a 2x reduction in estimation error while requiring fewer observations. This represents a notable advance in automated STEM calibration and demonstrates the potential of VAEs for information compression in images. Beyond microscopy, the VAE-EM framework applies to inverse problems where simulated training data introduces a reality gap and where non-injective mappings would otherwise prevent unique solutions.", "date": "2026-03-17", "categories": []}
{"id": "2603.18572", "title": "UEPS: Robust and Efficient MRI Reconstruction", "abstract": "Deep unrolled models (DUMs) have become the state of the art for accelerated MRI reconstruction, yet their robustness under domain shift remains a critical barrier to clinical adoption. In this work, we identify coil sensitivity map (CSM) estimation as the primary bottleneck limiting generalization. To address this, we propose UEPS, a novel DUM architecture featuring three key innovations: (i) an Unrolled Expanded (UE) design that eliminates CSM dependency by reconstructing each coil independently; (ii) progressive resolution, which leverages k-space-to-image mapping for efficient coarse-to-fine refinement; and (iii) sparse attention tailored to MRI's 1D undersampling nature. These physics-grounded designs enable simultaneous gains in robustness and computational efficiency. We construct a large-scale zero-shot transfer benchmark comprising 10 out-of-distribution test sets spanning diverse clinical shifts -- anatomy, view, contrast, vendor, field strength, and coil configurations. Extensive experiments demonstrate that UEPS consistently and substantially outperforms existing DUM, end-to-end, diffusion, and untrained methods across all OOD tests, achieving state-of-the-art robustness with low-latency inference suitable for real-time deployment.", "date": "2026-03-19", "categories": []}
{"id": "2410.12346", "title": "Efficient Diffusion as Low Light Enhancer", "abstract": "The computational burden of the iterative sampling process remains a major challenge in diffusion-based Low-Light Image Enhancement (LLIE). Current acceleration methods, whether training-based or training-free, often lead to significant performance degradation, highlighting the trade-off between performance and efficiency. In this paper, we identify two primary factors contributing to performance degradation: fitting errors and the inference gap. Our key insight is that fitting errors can be mitigated by linearly extrapolating the incorrect score functions, while the inference gap can be reduced by shifting the Gaussian flow to a reflectance-aware residual space. Based on the above insights, we design Reflectance-Aware Trajectory Refinement (RATR) module, a simple yet effective module to refine the teacher trajectory using the reflectance component of images. Following this, we introduce \\textbf{Re}flectance-aware \\textbf{D}iffusion with \\textbf{Di}stilled \\textbf{T}rajectory (\\textbf{ReDDiT}), an efficient and flexible distillation framework tailored for LLIE. Our framework achieves comparable performance to previous diffusion-based methods with redundant steps in just 2 steps while establishing new state-of-the-art (SOTA) results with 8 or 4 steps. Comprehensive experimental evaluations on 10 benchmark datasets validate the effectiveness of our method, consistently outperforming existing SOTA methods.", "date": "2024-10-16", "categories": []}
{"id": "2512.16923", "title": "Generative Refocusing: Flexible Defocus Control from a Single Image", "abstract": "Depth-of-field control is essential in photography, but achieving perfect focus often requires multiple attempts or specialized equipment. Single-image refocusing is still difficult. It involves recovering sharp content and creating realistic bokeh. Current methods have significant drawbacks. They require all-in-focus inputs, rely on synthetic data from simulators, and have limited control over the aperture. We introduce Generative Refocusing, a two-step process that uses DeblurNet to recover all-in-focus images from diverse inputs and BokehNet to create controllable bokeh. This method combines synthetic and real bokeh images to achieve precise control while preserving authentic optical characteristics. Our experiments show we achieve top performance in defocus deblurring, bokeh synthesis, and refocusing benchmarks. Additionally, our Generative Refocusing allows custom aperture shapes. Project page: https://generative-refocusing.github.io/", "date": "2025-12-18", "categories": []}
{"id": "2601.21592", "title": "Unifying Heterogeneous Degradations: Uncertainty-Aware Diffusion Bridge Model for All-in-One Image Restoration", "abstract": "All-in-One Image Restoration (AiOIR) faces the fundamental challenge in reconciling conflicting optimization objectives across heterogeneous degradations. Existing methods are often constrained by coarse-grained control mechanisms or fixed mapping schedules, yielding suboptimal adaptation. To address this, we propose an Uncertainty-Aware Diffusion Bridge Model (UDBM), which innovatively reformulates AiOIR as a stochastic transport problem steered by pixel-wise uncertainty. By introducing a relaxed diffusion bridge formulation which replaces the strict terminal constraint with a relaxed constraint, we model the uncertainty of degradations while theoretically resolving the drift singularity inherent in standard diffusion bridges. Furthermore, we devise a dual modulation strategy: the noise schedule aligns diverse degradations into a shared high-entropy latent space, while the path schedule adaptively regulates the transport trajectory motivated by the viscous dynamics of entropy regularization. By effectively rectifying the transport geometry and dynamics, UDBM achieves state-of-the-art performance across diverse restoration tasks within a single inference step.", "date": "2026-01-29", "categories": []}
{"id": "2603.17156", "title": "A Lensless Polarization Camera", "abstract": "Polarization imaging is a technique that creates a pixel map of the polarization state in a scene. Although invisible to the human eye, polarization can assist various sensing and computer vision tasks. Existing polarization cameras use spatial or temporal multiplexing, which increases the camera volume, weight, cost, or all of the above. Recent lensless imaging approaches, such as DiffuserCam, have demonstrated that compact imaging systems can be realized by replacing the lens with a coding element and performing computational reconstruction. In this work, we propose a compact lensless polarization camera composed of a diffuser and a simple striped polarization mask. By combining this optical design with a reconstruction algorithm that explicitly models the polarization-encoded lensless measurements, four linear polarization images are recovered from a single snapshot. Our results demonstrate the potential of lensless approaches for polarization imaging and reveal the physical factors that govern reconstruction quality, guiding the development of high-quality practical systems.", "date": "2026-03-17", "categories": []}
{"id": "2511.19117", "title": "3M-TI: High-Quality Mobile Thermal Imaging via Calibration-free Multi-Camera Cross-Modal Diffusion", "abstract": "The miniaturization of thermal sensors for mobile platforms inherently limits their spatial resolution and textural fidelity, leading to blurry and less informative images. Existing thermal super-resolution (SR) methods can be grouped into single-image and RGB-guided approaches: the former struggles to recover fine structures from limited information, while the latter relies on accurate and laborious cross-camera calibration, which hinders practical deployment and robustness. Here, we propose 3M-TI, a calibration-free Multi-camera cross-Modality diffusion framework for Mobile Thermal Imaging. At its core, 3M-TI integrates a cross-modal self-attention module (CSM) into the diffusion UNet, replacing the original self-attention layers to adaptively align thermal and RGB features throughout the denoising process, without requiring explicit camera calibration. This design enables the diffusion network to leverage its generative prior to enhance spatial resolution, structural fidelity, and texture detail in the super-resolved thermal images. Extensive evaluations on real-world mobile thermal cameras and public benchmarks validate our superior performance, achieving state-of-the-art results in both visual quality and quantitative metrics. More importantly, the thermal images enhanced by 3M-TI lead to substantial gains in critical downstream tasks like object detection and segmentation, underscoring its practical value for robust mobile thermal perception systems. More materials: https://github.com/work-submit/3MTI.", "date": "2025-11-24", "categories": []}
{"id": "2512.11237", "title": "WildCap: Facial Albedo Capture in the Wild via Hybrid Inverse Rendering", "abstract": "Existing methods achieve high-quality facial albedo capture under controllable lighting, which increases capture cost and limits usability. We propose WildCap, a novel method for high-quality facial albedo capture from a smartphone video recorded in the wild. To disentangle high-quality albedo from complex lighting effects in in-the-wild captures, we propose a novel hybrid inverse rendering framework. We first apply a data-driven method, i.e., SwitchLight, to convert the captured images into more constrained conditions and then adopt model-based inverse rendering. However, unavoidable local artifacts in network predictions, such as shadow-baking, are non-physical and thus hinder accurate inverse rendering of lighting and material. To address this, we propose a novel texel grid lighting model to explain non-physical effects as clean albedo illuminated by local physical lighting. During optimization, we jointly sample a diffusion prior for the albedo map and optimize the lighting, effectively resolving scale ambiguity between local lights and albedo. Other reflectance maps are then predicted from the albedo. Our method achieves significantly better results than prior arts in the same capture setup, closing the quality gap between in-the-wild and controllable recordings by a large margin.", "date": "2025-12-12", "categories": []}
{"id": "2603.16243", "title": "RASLF: Representation-Aware State Space Model for Light Field Super-Resolution", "abstract": "Current SSM-based light field super-resolution (LFSR) methods often fail to fully leverage the complementarity among various LF representations, leading to the loss of fine textures and geometric misalignments across views. To address these issues, we propose RASLF, a representation-aware state-space framework that explicitly models structural correlations across multiple LF representations. Specifically, a Progressive Geometric Refinement (PGR) block is created that uses a panoramic epipolar representation to explicitly encode multi-view parallax differences, thereby enabling integration across different LF representations. Furthermore, we introduce a Representation Aware Asymmetric Scanning (RAAS) mechanism that dynamically adjusts scanning paths based on the physical properties of different representation spaces, optimizing the balance between performance and efficiency through path pruning. Additionally, a Dual-Anchor Aggregation (DAA) module improves hierarchical feature flow, reducing redundant deeplayer features and prioritizing important reconstruction information. Experiments on various public benchmarks show that RASLF achieves the highest reconstruction accuracy while remaining highly computationally efficient.", "date": "2026-03-17", "categories": []}
{"id": "2603.16393", "title": "Robust Physics-Guided Diffusion for Full-Waveform Inversion", "abstract": "We develop a robust physics-guided diffusion framework for full-waveform inversion that combines a score-based generative prior with likelihood guidance computed through wave-equation simulations. We adopt a transport-based data-consistency potential (Wasserstein-2), incorporating wavefield enhancement via bounded weighting and observation-dependent normalization, thereby improving robustness to amplitude imbalance and time/phase misalignment. On the inference side, we introduce a preconditioned guided reverse-diffusion scheme that adapts the guidance strength and spatial scaling throughout the reverse-time dynamics, yielding a more stable and effective data-consistency guidance step than standard diffusion posterior sampling (DPS). Numerical experiments on OpenFWI datasets demonstrate improved reconstruction quality over deterministic optimization baselines and standard DPS under comparable computational budgets.", "date": "2026-03-17", "categories": []}
{"id": "2603.16404", "title": "Near-light Photometric Stereo with Symmetric Lights", "abstract": "This paper describes a linear solution method for near-light photometric stereo by exploiting symmetric light source arrangements. Unlike conventional non-convex optimization approaches, by arranging multiple sets of symmetric nearby light source pairs, our method derives a closed-form solution for surface normal and depth without requiring initialization. In addition, our method works as long as the light sources are symmetrically distributed about an arbitrary point even when the entire spatial offset is uncalibrated. Experiments showcase the accuracy of shape recovery accuracy of our method, achieving comparable results to the state-of-the-art calibrated near-light photometric stereo method while significantly reducing requirements of careful depth initialization and light calibration.", "date": "2026-03-17", "categories": []}
{"id": "2410.12424", "title": "Nonlinear Bayesian Doppler Tomography for Simultaneous Reconstruction of Flow and Temperature", "abstract": "We present a nonlinear Bayesian tomographic framework for Doppler spectral imaging that enables simultaneous reconstruction of emissivity, ion temperature, and flow velocity from line-integrated spectra. The method employs nonlinear Gaussian process tomography (GPT) with a Laplace approximation while retaining the full Doppler forward model. A log-Gaussian process prior stabilizes the velocity reconstruction in low-emissivity regions where Doppler information becomes weak, preventing the unphysical divergence of velocity estimates commonly encountered in conventional spectral tomography. The reconstruction method is verified using synthetic phantom data and applied to coherence imaging spectroscopy (CIS) measurements in the RT-1 device, resolving spatial structures of ion temperature and toroidal ion flow characteristic of magnetospheric plasma in the RT-1 device. The framework extends existing CIS tomography to regimes with strong flows and large temperature variations and provides a general Bayesian approach for Doppler spectral tomography that can be integrated with complementary spectroscopic diagnostics.", "date": "2024-10-16", "categories": []}
{"id": "2512.03216", "title": "Kaleidoscopic Scintillation Event Imaging", "abstract": "Scintillators are transparent materials that interact with high-energy particles and emit visible light as a result. They are used in state of the art methods of measuring high-energy particles and radiation sources. Most existing methods use fast single-pixel detectors to detect and time scintillation events. Cameras provide spatial resolution but can only capture an average over many events, making it difficult to image the events associated with an individual particle. Emerging single-photon avalanche diode cameras combine speed and spatial resolution to enable capturing images of individual events. This allows us to use machine vision techniques to analyze events, enabling new types of detectors. The main challenge is the very low brightness of the events. Techniques have to work with a very limited number of photons.\n  We propose a kaleidoscopic scintillator to increase light collection in a single-photon camera while preserving the event's spatial information. The kaleidoscopic geometry creates mirror reflections of the event in known locations for a given event location that are captured by the camera. We introduce theory for imaging an event in a kaleidoscopic scintillator and an algorithm to estimate the event's 3D position. We find that the kaleidoscopic scintillator design provides sufficient light collection to perform high-resolution event measurements for advanced radiation imaging techniques using a commercial CMOS single-photon camera.", "date": "2025-12-02", "categories": []}
{"id": "2601.23276", "title": "Denoising the Deep Sky: Physics-Based CCD Noise Formation for Astronomical Imaging", "abstract": "Astronomical imaging remains noise-limited under practical observing conditions. Standard calibration pipelines remove structured artifacts but largely leave stochastic noise unresolved. Although learning-based denoising has shown strong potential, progress is constrained by scarce paired training data and the requirement for physically interpretable models in scientific workflows. We propose a physics-based noise synthesis framework tailored to CCD noise formation in the telescope. The pipeline models photon shot noise, photo-response non-uniformity, dark-current noise, readout effects, and localized outliers arising from cosmic-ray hits and hot pixels. To obtain low-noise inputs for synthesis, we stack multiple unregistered exposures to produce high-SNR bases. Realistic noisy counterparts synthesized from these bases using our noise model enable the construction of abundant paired datasets for supervised learning. Extensive experiments on our real-world multi-band dataset curated from two ground-based telescopes demonstrate the effectiveness of our framework in both photometric and scientific accuracy.", "date": "2026-01-30", "categories": []}
{"id": "2603.07234", "title": "Single Image Super-Resolution via Bivariate `A Trous Wavelet Diffusion", "abstract": "The effectiveness of super resolution (SR) models hinges on their ability to recover high frequency structure without introducing artifacts. Diffusion based approaches have recently advanced the state of the art in SR. However, most diffusion based SR pipelines operate purely in the spatial domain, which may yield high frequency details that are not well supported by the underlying low resolution evidence. On the other hand, unlike supervised SR models that may inject dataset specific textures, single image SR relies primarily on internal image statistics and can therefore be less prone to dataset-driven hallucinations; nevertheless, ambiguity in the LR observation can still lead to inconsistent high frequency details. To tackle this problem, we introduce BATDiff, an unsupervised Bivariate A trous Wavelet Diffusion model designed to provide structured cross scale guidance during the generative process. BATDiff employs an a Trous wavelet transform that constructs an undecimated multiscale representation in which high frequency components are progressively revealed while the full spatial resolution is preserved. As the core inference mechanism, BATDiff includes a bivariate cross scale module that models parent child dependencies between adjacent scales. It improves high frequency coherence and reduces mismatch artifacts in diffusion based SR. Experiments on standard benchmarks demonstrate that BATDiff produces sharper and more structurally consistent reconstructions than existing diffusion and non diffusion baselines, achieving improvements in fidelity and perceptual quality.", "date": "2026-03-07", "categories": []}
{"id": "2603.07489", "title": "RobustSCI: Beyond Reconstruction to Restoration for Snapshot Compressive Imaging under Real-World Degradations", "abstract": "Deep learning algorithms for video Snapshot Compressive Imaging (SCI) have achieved great success, yet they predominantly focus on reconstructing from clean measurements. This overlooks a critical real-world challenge: the captured signal itself is often severely degraded by motion blur and low light. Consequently, existing models falter in practical applications. To break this limitation, we pioneer the first study on robust video SCI restoration, shifting the goal from \"reconstruction\" to \"restoration\"--recovering the underlying pristine scene from a degraded measurement. To facilitate this new task, we first construct a large-scale benchmark by simulating realistic, continuous degradations on the DAVIS 2017 dataset. Second, we propose RobustSCI, a network that enhances a strong encoder-decoder backbone with a novel RobustCFormer block. This block introduces two parallel branches--a multi-scale deblur branch and a frequency enhancement branch--to explicitly disentangle and remove degradations during the recovery process. Furthermore, we introduce RobustSCI-C (RobustSCI-Cascade), which integrates a pre-trained Lightweight Post-processing Deblurring Network to significantly boost restoration performance with minimal overhead. Extensive experiments demonstrate that our methods outperform all SOTA models on the new degraded testbeds, with additional validation on real-world degraded SCI data confirming their practical effectiveness, elevating SCI from merely reconstructing what is captured to restoring what truly happened.", "date": "2026-03-08", "categories": []}
{"id": "2603.08582", "title": "Online Sparse Synthetic Aperture Radar Imaging", "abstract": "With modern defense applications increasingly relying on inexpensive, autonomous drones, lies the major challenge of designing computationally and memory-efficient onboard algorithms to fulfill mission objectives. This challenge is particularly significant in Synthetic Aperture Radar (SAR), where large volumes of data must be collected and processed for downstream tasks. We propose an online reconstruction method, the Online Fast Iterative Shrinkage-Thresholding Algorithm (Online FISTA), which incrementally reconstructs a scene with limited data through sparse coding. Rather than requiring storage of all received signal data, the algorithm recursively updates storage matrices for each iteration, greatly reducing memory demands. Online SAR image reconstruction facilitates more complex downstream tasks, such as Automatic Target Recognition (ATR), in an online manner, resulting in a more versatile and integrated framework compared to existing post-collection reconstruction and ATR approaches.", "date": "2026-03-09", "categories": []}
{"id": "2603.13425", "title": "Self-Flow-Matching assisted Full Waveform Inversion", "abstract": "Full-waveform inversion (FWI) is a high-resolution seismic imaging method that estimates subsurface velocity by matching simulated and recorded waveforms. However, FWI is highly nonlinear, prone to cycle skipping, and sensitive to noise, particularly when low frequencies are missing or the initial model is poor, leading to failures under imperfect acquisition. Diffusion-regularized FWI introduces generative priors to encourage geologically realistic models, but these priors typically require costly offline pretraining and can deteriorate under distribution shift. Moreover, they assume Gaussian initialization and a fixed noise schedule, in which it is unclear how to map a deterministic FWI iterate and its starting model to a well-defined diffusion time or noise level. To address these limitations, we introduce Self-Flow-Matching assisted Full-Waveform Inversion (SFM-FWI), a physics-driven framework that eliminates the need for large-scale offline pretraining while avoiding the noise-level alignment ambiguity. SFM-FWI leverages flow matching to learn a transport field without assuming Gaussian initialization or a predefined noise schedule, so the initial model can be used directly as the starting point of the dynamics. Our approach trains a single flow network online using the governing physics and observed data. At each outer iteration, we build an interpolated model and update the flow by backpropagating the FWI data misfit, providing self-supervision without external training pairs. Experiments on challenging synthetic benchmarks show that SFM-FWI delivers more accurate reconstructions, greater noise robustness, and more stable convergence than standard FWI and pretraining-free regularization methods.", "date": "2026-03-13", "categories": []}
{"id": "2603.13466", "title": "Open World MRI Reconstruction with Bias-Calibrated Adaptation", "abstract": "Real-world MRI reconstruction systems face the open-world challenge: test data from unseen imaging centers, anatomical structures, or acquisition protocols can differ drastically from training data, causing severe performance degradation. Existing methods struggle with this challenge. To address this, we propose BiasRecon, a bias-calibrated adaptation framework grounded in the minimal intervention principle: preserve what transfers, calibrate what does not. Concretely, BiasRecon formulates open-world adaptation as an alternating optimization framework that jointly optimizes three components: (1) frequency-guided prior calibration that introduces layer-wise calibration variables to selectively modulate frequency-specific features of the pre-trained score network via self-supervised k-space signals, (2) score-based denoising that leverages the calibrated generative prior for high-fidelity image reconstruction, and (3) adaptive regularization that employs Stein's Unbiased Risk Estimator to dynamically balance the prior-measurement trade-off, matching test-time noise characteristics without requiring ground truth. By intervening minimally and precisely through this alternating scheme, BiasRecon achieves robust adaptation with fewer than 100 tunable parameters. Extensive experiments across four datasets demonstrate state-of-the-art performance on open-world reconstruction tasks.", "date": "2026-03-13", "categories": []}
{"id": "2603.13521", "title": "Eleven Primitives and Three Gates: The Universal Structure of Computational Imaging", "abstract": "Computational imaging systems -- from coded-aperture cameras to cryo-electron microscopes -- span five carrier families yet share a hidden structural simplicity. We prove that every imaging forward model decomposes into a directed acyclic graph over exactly 11 physically typed primitives (Finite Primitive Basis Theorem) -- a sufficient and minimal basis that provides a compositional language for designing any imaging modality. We further prove that every reconstruction failure has exactly three independent root causes: information deficiency, carrier noise, and operator mismatch (Triad\n  Decomposition). The three gates map to the system lifecycle: Gates 1 and 2 guide design (sampling geometry, carrier\n  selection); Gate 3 governs deployment-stage calibration and drift correction. Validation across 12 modalities and all\n  five carrier families confirms both results, with +0.8 to +13.9 dB recovery on deployed instruments. Together, the 11\n  primitives and 3 gates establish the first universal grammar for designing, diagnosing, and correcting computational\n  imaging systems.", "date": "2026-03-13", "categories": []}
{"id": "2603.14031", "title": "Intrinsic Tolerance in C-Arm Imaging: How Extrinsic Re-optimization Preserves 3D Reconstruction Accuracy", "abstract": "\\textbf{Purpose:} C-arm fluoroscopy's 3D reconstruction relies on accurate intrinsic calibration, which is often challenging in clinical practice. This study ensures high-precision reconstruction accuracy by re-optimizing the extrinsic parameters to compensate for intrinsic calibration errors.\n  \\noindent\\textbf{Methods:} We conducted both simulation and real-world experiments using five commercial C-arm systems. Intrinsic parameters were perturbed in controlled increments. Focal length was increased by 100 to 700 pixels ($\\approx$20 mm to 140 mm) and principal point by 20 to 200 pixels. For each perturbation, we (1) reconstructed 3D points from known phantom geometries, (2) re-estimated extrinsic poses using standard optimization, and (3) measured reconstruction and reprojection errors relative to ground truth.\n  \\noindent\\textbf{Results:} Even with focal length errors up to 500 pixels ($\\approx$100 mm, assuming a nominal focal length of $\\sim$1000 mm), mean 3D reconstruction error remained under 0.2 mm. Larger focal length deviations (700 pixels) elevated error to only $\\approx$0.3 mm. Principal point shifts up to 200 pixels introduced negligible reconstruction error once extrinsic parameters were re-optimized, with reprojection error increases below 0.5 pixels.\n  \\noindent\\textbf{Conclusion:} Moderate errors in intrinsic calibration can be effectively mitigated by extrinsic re-optimization, preserving submillimeter 3D reconstruction accuracy. This intrinsic tolerance suggests a practical pathway to relax calibration precision requirements, thereby simplifying C-arm system setup and reducing clinical workflow burden without compromising performance.", "date": "2026-03-14", "categories": []}
{"id": "2603.14074", "title": "Self-Supervised Uncertainty Estimation For Super-Resolution of Satellite Images", "abstract": "Super-resolution (SR) of satellite imagery is challenging due to the lack of paired low-/high-resolution data. Recent self-supervised SR methods overcome this limitation by exploiting the temporal redundancy in burst observations, but they lack a mechanism to quantify uncertainty in the reconstruction. In this work, we introduce a novel self-supervised loss that allows to estimate uncertainty in image super-resolution without ever accessing the ground-truth high-resolution data. We adopt a decision-theoretic perspective and show that minimizing the corresponding Bayesian risk yields the posterior mean and variance as optimal estimators. We validate our approach on a synthetic SkySat L1B dataset and demonstrate that it produces calibrated uncertainty estimates comparable to supervised methods. Our work bridges self-supervised restoration with uncertainty quantification, making a practical framework for uncertainty-aware image reconstruction.", "date": "2026-03-14", "categories": []}
{"id": "2603.14084", "title": "Bootstrapped Physically-Primed Neural Networks for Robust T2 Distribution Estimation in Low-SNR Pancreatic MRI", "abstract": "Estimating multi-component T2 relaxation distributions from Multi-Echo Spin Echo (MESE) MRI is a severely ill-posed inverse problem, traditionally solved using regularized non-negative least squares (NNLS). In abdominal imaging, particularly the pancreas, low SNR and residual uncorrelated noise challenge classical solvers and deterministic deep learning models. We introduce a bootstrap-based inference framework for robust distributional T2 estimation that performs stochastic resampling of the echo train and aggregates predictions across multiple subsets. This treats the acquisition as a distribution rather than a fixed input, yielding variance-reduced, physically consistent estimates and converting deterministic relaxometry networks into probabilistic ensemble predictors. Applied to the P2T2 architecture, our method uses inference-time bootstrapping to smooth noise artifacts and enhance fidelity to the underlying relaxation distribution.\n  Noninvasive pancreatic evaluation is limited by location and biopsy risks, highlighting the need for biomarkers capable of capturing early pathophysiological changes. In type 1 diabetes (T1DM), progressive beta-cell destruction begins years before overt hyperglycemia, yet current imaging cannot assess early islet decline.\n  We evaluate clinical utility via a test-retest reproducibility study (N=7) and a T1DM versus healthy differentiation task (N=8). Our approach achieves the lowest Wasserstein distances across repeated scans and superior sensitivity to physiology-driven shifts in the relaxation-time distribution, outperforming NNLS and deterministic deep learning baselines. These results establish inference-time bootstrapping as an effective enhancement for quantitative T2 relaxometry in low-SNR abdominal imaging.", "date": "2026-03-14", "categories": []}
{"id": "2603.14816", "title": "M2IR: Proactive All-in-One Image Restoration via Mamba-style Modulation and Mixture-of-Experts", "abstract": "While Transformer-based architectures have dominated recent advances in all-in-one image restoration, they remain fundamentally reactive: propagating degradations rather than proactively suppressing them. In the absence of explicit suppression mechanisms, degraded signals interfere with feature learning, compelling the decoder to balance artifact removal and detail preservation, thereby increasing model complexity and limiting adaptability. To address these challenges, we propose M2IR, a novel restoration framework that proactively regulates degradation propagation during the encoding stage and efficiently eliminates residual degradations during decoding. Specifically, the Mamba-Style Transformer (MST) block performs pixel-wise selective state modulation to mitigate degradations while preserving structural integrity. In parallel, the Adaptive Degradation Expert Collaboration (ADEC) module utilizes degradation-specific experts guided by a DA-CLIP-driven router and complemented by a shared expert to eliminate residual degradations through targeted and cooperative restoration. By integrating the MST block and ADEC module, M2IR transitions from passive reaction to active degradation control, effectively harnessing learned representations to achieve superior generalization, enhanced adaptability, and refined recovery of fine-grained details across diverse all-in-one image restoration benchmarks. Our source codes are available at https://github.com/Im34v/M2IR.", "date": "2026-03-16", "categories": []}
{"id": "2603.14850", "title": "From Artefact to Insight: Efficient Low-Rank Adaptation of BrushNet for Scanning Probe Microscopy Image Restoration", "abstract": "Scanning Probe Microscopy or SPM offers nanoscale resolution but is frequently marred by structured artefacts such as line scan dropout, gain induced noise, tip convolution, and phase hops. While most available methods treat SPM artefact removal as isolated denoising or interpolation tasks, the generative inpainting perspective remains largely unexplored. In this work, we introduce a diffusion based inpainting framework tailored to scientific grayscale imagery. By fine tuning less than 0.2 percent of BrushNet weights with rank constrained low rank adaptation (LoRA), we adapt a pretrained diffusion model using only 7390 artefact, clean pairs distilled from 739 experimental scans. On our forthcoming public SPM InpBench benchmark, the LoRA enhanced model lifts the Peak Signal to Noise Ratio or PSNR by 6.61 dB and halves the Learned Perceptual Image Patch Similarity or LPIPS relative to zero-shot inference, while matching or slightly surpassing the accuracy of full retraining, trainable on a single GPU instead of four high-memory cards. The approach generalizes across various SPM image channels including height, amplitude and phase, faithfully restores subtle structural details, and suppresses hallucination artefacts inherited from natural image priors. This lightweight framework enables efficient, scalable recovery of irreplaceable SPM images and paves the way for a broader diffusion model adoption in nanoscopic imaging analysis.", "date": "2026-03-16", "categories": []}
{"id": "2603.14885", "title": "SpiralDiff: Spiral Diffusion with LoRA for RGB-to-RAW Conversion Across Cameras", "abstract": "RAW images preserve superior fidelity and rich scene information compared to RGB, making them essential for tasks in challenging imaging conditions. To alleviate the high cost of data collection, recent RGB-to-RAW conversion methods aim to synthesize RAW images from RGB. However, they overlook two key challenges: (i) the reconstruction difficulty varies with pixel intensity, and (ii) multi-camera conversion requires camera-specific adaptation. To address these issues, we propose SpiralDiff, a diffusion-based framework tailored for RGB-to-RAW conversion with a signal-dependent noise weighting strategy that adapts reconstruction fidelity across intensity levels. In addition, we introduce CamLoRA, a camera-aware lightweight adaptation module that enables a unified model to adapt to different camera-specific ISP characteristics. Extensive experiments on four benchmark datasets demonstrate the superiority of SpiralDiff in RGB-to-RAW conversion quality and its downstream benefits in RAW-based object detection. Our code and model are available at https://github.com/Chuancy-TJU/SpiralDiff.", "date": "2026-03-16", "categories": []}
{"id": "2603.14915", "title": "ILV: Iterative Latent Volumes for Fast and Accurate Sparse-View CT Reconstruction", "abstract": "A long-term goal in CT imaging is to achieve fast and accurate 3D reconstruction from sparse-view projections, thereby reducing radiation exposure, lowering system cost, and enabling timely imaging in clinical workflows. Recent feed-forward approaches have shown strong potential toward this overarching goal, yet their results still suffer from artifacts and loss of fine details. In this work, we introduce Iterative Latent Volumes (ILV), a feed-forward framework that integrates data-driven priors with classical iterative reconstruction principles to overcome key limitations of prior feed-forward models in sparse-view CBCT reconstruction. At its core, ILV constructs an explicit 3D latent volume that is repeatedly updated by conditioning on multi-view X-ray features and the learned anatomical prior, enabling the recovery of fine structural details beyond the reach of prior feed-forward models. In addition, we develop and incorporate several key architectural components, including an X-ray feature volume, group cross-attention, efficient self-attention, and view-wise feature aggregation, that efficiently realize its core latent volume refinement concept. Extensive experiments on a large-scale dataset of approximately 14,000 CT volumes demonstrate that ILV significantly outperforms existing feed-forward and optimization-based methods in both reconstruction quality and speed. These results show that ILV enables fast and accurate sparse-view CBCT reconstruction suitable for clinical use. The project page is available at: https://sngryonglee.github.io/ILV/.", "date": "2026-03-16", "categories": []}
{"id": "2603.15131", "title": "Low-light Image Enhancement with Retinex Decomposition in Latent Space", "abstract": "Retinex theory provides a principled foundation for low-light image enhancement, inspiring numerous learning-based methods that integrate its principles. However, existing methods exhibits limitations in accurately decomposing reflectance and illumination components. To address this, we propose a Retinex-Guided Transformer~(RGT) model, which is a two-stage model consisting of decomposition and enhancement phases. First, we propose a latent space decomposition strategy to separate reflectance and illumination components. By incorporating the log transformation and 1-pixel offset, we convert the intrinsically multiplicative relationship into an additive formulation, enhancing decomposition stability and precision. Subsequently, we construct a U-shaped component refiner incorporating the proposed guidance fusion transformer block. The component refiner refines reflectance component to preserve texture details and optimize illumination distribution, effectively transforming low-light inputs to normal-light counterparts. Experimental evaluations across four benchmark datasets validate that our method achieves competitive performance in low-light enhancement and a more stable training process.", "date": "2026-03-16", "categories": []}
{"id": "2603.15348", "title": "Oscillating Dispersion for Maximal Light-throughput Spectral Imaging", "abstract": "Existing computational spectral imaging systems typically rely on coded aperture and beam splitters that block a substantial fraction of incident light, degrading reconstruction quality under light-starved conditions. To address this limitation, we develop the Oscillating Dispersion Imaging Spectrometer (ODIS), which for the first time achieves near-full light throughput by axially translating a disperser between the conjugate image plane and a defocused position, sequentially capturing a panchromatic (PAN) image and a dispersed measurement along a single optical path. We further propose a PAN-guided Dispersion-Aware Deep Unfolding Network (PDAUN) that recovers high-fidelity spectral information from maskless dispersion under PAN structural guidance. Its data-fidelity step derives an FFT-Woodbury preconditioned solver by exploiting the cyclic-convolution property of the ODIS forward model, while a Dispersion-Aware Deformable Convolution module (DADC) corrects sub-pixel spectral misalignment using PAN features. Experiments show state-of-the-art performance on standard benchmarks, and cross-system comparisons confirm that ODIS yields decisive gains under low illumination. High-fidelity reconstruction is validated on a physical prototype.", "date": "2026-03-16", "categories": []}
{"id": "2603.15555", "title": "Learning Latent Proxies for Controllable Single-Image Relighting", "abstract": "Single-image relighting is highly under-constrained: small illumination changes can produce large, nonlinear variations in shading, shadows, and specularities, while geometry and materials remain unobserved. Existing diffusion-based approaches either rely on intrinsic or G-buffer pipelines that require dense and fragile supervision, or operate purely in latent space without physical grounding, making fine-grained control of direction, intensity, and color unreliable. We observe that a full intrinsic decomposition is unnecessary and redundant for accurate relighting. Instead, sparse but physically meaningful cues, indicating where illumination should change and how materials should respond, are sufficient to guide a diffusion model. Based on this insight, we introduce LightCtrl that integrates physical priors at two levels: a few-shot latent proxy encoder that extracts compact material-geometry cues from limited PBR supervision, and a lighting-aware mask that identifies sensitive illumination regions and steers the denoiser toward shading relevant pixels. To compensate for scarce PBR data, we refine the proxy branch using a DPO-based objective that enforces physical consistency in the predicted cues. We also present ScaLight, a large-scale object-level dataset with systematically varied illumination and complete camera-light metadata, enabling physically consistent and controllable training. Across object and scene level benchmarks, our method achieves photometrically faithful relighting with accurate continuous control, surpassing prior diffusion and intrinsic-based baselines, including gains of up to +2.4 dB PSNR and 35% lower RMSE under controlled lighting shifts.", "date": "2026-03-16", "categories": []}
{"id": "2601.12551", "title": "PISE: Physics-Anchored Semantically-Enhanced Deep Computational Ghost Imaging for Robust Low-Bandwidth Machine Perception", "abstract": "We propose PISE, a physics-informed deep ghost imaging framework for low-bandwidth edge perception. By combining adjoint operator initialization with semantic guidance, PISE improves classification accuracy by 2.57% and reduces variance by 9x at 5% sampling.", "date": "2026-01-18", "categories": []}
{"id": "2603.12493", "title": "RAW-Domain Degradation Models for Realistic Smartphone Super-Resolution", "abstract": "Digital zoom on smartphones relies on learning-based super-resolution (SR) models that operate on RAW sensor images, but obtaining sensor-specific training data is challenging due to the lack of ground-truth images. Synthetic data generation via ``unprocessing'' pipelines offers a potential solution by simulating the degradations that transform high-resolution (HR) images into their low-resolution (LR) counterparts. However, these pipelines can introduce domain gaps due to incomplete or unrealistic degradation modeling. In this paper, we demonstrate that principled and carefully designed degradation modeling can enhance SR performance in real-world conditions. Instead of relying on generic priors for camera blur and noise, we model device-specific degradations through calibration and unprocess publicly available rendered images into the RAW domain of different smartphones. Using these image pairs, we train a single-image RAW-to-RGB SR model and evaluate it on real data from a held-out device. Our experiments show that accurate degradation modeling leads to noticeable improvements, with our SR model outperforming baselines trained on large pools of arbitrarily chosen degradations.", "date": "2026-03-12", "categories": []}
{"id": "2511.17895", "title": "Radiative-Structured Neural Operator for Continuous Spectral Super-Resolution", "abstract": "Spectral super-resolution (SSR) aims to reconstruct hyperspectral images (HSIs) from multispectral observations, with broad applications in computer vision and remote sensing. Deep learning-based methods have been widely used, but they often treat spectra as discrete vectors learned from data, rather than continuous curves constrained by physics principles, leading to unrealistic predictions and limited applicability. To address this challenge, we propose the Radiative-Structured Neural Operator (RSNO), which learns a continuous mapping for spectral super-resolution while enforcing physical consistency under the radiative prior. The proposed RSNO consists of three stages: upsampling, reconstruction, and refinement. In the upsampling stage, we leverage prior information to expand the input multispectral image, producing a physically plausible hyperspectral estimate. Subsequently, we adopt a neural operator backbone in the reconstruction stage to learn a continuous mapping across the spectral domain. Finally, the refinement stage imposes a hard constraint on the output HSI to eliminate color distortion. The upsampling and refinement stages are implemented via the proposed angular-consistent projection (ACP), which is derived from a non-convex optimization problem. Moreover, we theoretically demonstrated the optimality of ACP by null-space decomposition. Various experiments validate the effectiveness of the proposed approach in both discrete and continuous spectral super-resolution.", "date": "2025-11-22", "categories": []}
{"id": "2603.11316", "title": "MRI2Qmap: multi-parametric quantitative mapping with MRI-driven denoising priors", "abstract": "Magnetic Resonance Fingerprinting (MRF) and other highly accelerated transient-state parameter mapping techniques enable simultaneous quantification of multiple tissue properties, but often suffer from aliasing artifacts due to compressed sampling. Incorporating spatial image priors can mitigate these artifacts, and deep learning has shown strong potential when large training datasets are available. However, extending this paradigm to MRF-type sequences remains challenging due to the scarcity of quantitative imaging data for training. Can this limitation be overcome by leveraging sources of training data from clinically-routine weighted MRI images? To this end, we introduce MRI2Qmap, a plug-and-play quantitative reconstruction framework that integrates the physical acquisition model with priors learned from deep denoising autoencoders pretrained on large multimodal weighted-MRI datasets. MRI2Qmap demonstrates that spatial-domain structural priors learned from independently acquired datasets of routine weighted-MRI images can be effectively used for quantitative MRI reconstruction. The proposed method is validated on highly accelerated 3D whole-brain MRF data from both in-vivo and simulated acquisitions, achieving competitive or superior performance relative to existing baselines without requiring ground-truth quantitative imaging data for training. By decoupling quantitative reconstruction from the need for ground-truth MRF training data, this framework points toward a scalable paradigm for quantitative MRI that can capitalize on the large and growing repositories of routine clinical MRI.", "date": "2026-03-11", "categories": []}
{"id": "2603.12036", "title": "Single Pixel Image Classification using an Ultrafast Digital Light Projector", "abstract": "Pattern recognition and image classification are essential tasks in machine vision. Autonomous vehicles, for example, require being able to collect the complex information contained in a changing environment and classify it in real time. Here, we experimentally demonstrate image classification at multi-kHz frame rates combining the technique of single pixel imaging (SPI) with a low complexity machine learning model. The use of a microLED-on-CMOS digital light projector for SPI enables ultrafast pattern generation for sub-ms image encoding. We investigate the classification accuracy of our experimental system against the broadly accepted benchmarking task of the MNIST digits classification. We compare the classification performance of two machine learning models: An extreme learning machine (ELM) and a backpropagation trained deep neural network. The complexity of both models is kept low so the overhead added to the inference time is comparable to the image generation time. Crucially, our single pixel image classification approach is based on a spatiotemporal transformation of the information, entirely bypassing the need for image reconstruction. By exploring the performance of our SPI based ELM as binary classifier we demonstrate its potential for efficient anomaly detection in ultrafast imaging scenarios.", "date": "2026-03-12", "categories": []}
{"id": "2603.12083", "title": "Towards Universal Computational Aberration Correction in Photographic Cameras: A Comprehensive Benchmark Analysis", "abstract": "Prevalent Computational Aberration Correction (CAC) methods are typically tailored to specific optical systems, leading to poor generalization and labor-intensive re-training for new lenses. Developing CAC paradigms capable of generalizing across diverse photographic lenses offers a promising solution to these challenges. However, efforts to achieve such cross-lens universality within consumer photography are still in their early stages due to the lack of a comprehensive benchmark that encompasses a sufficiently wide range of optical aberrations. Furthermore, it remains unclear which specific factors influence existing CAC methods and how these factors affect their performance. In this paper, we present comprehensive experiments and evaluations involving 24 image restoration and CAC algorithms, utilizing our newly proposed UniCAC, a large-scale benchmark for photographic cameras constructed via automatic optical design. The Optical Degradation Evaluator (ODE) is introduced as a novel framework to objectively assess the difficulty of CAC tasks, offering credible quantification of optical aberrations and enabling reliable evaluation. Drawing on our comparative analysis, we identify three key factors -- prior utilization, network architecture, and training strategy -- that most significantly influence CAC performance, and further investigate their respective effects. We believe that our benchmark, dataset, and observations contribute foundational insights to related areas and lay the groundwork for future investigations. Benchmarks, codes, and Zemax files will be available at https://github.com/XiaolongQian/UniCAC.", "date": "2026-03-12", "categories": []}
{"id": "2208.01702", "title": "Non-Line-of-Sight Tracking and Mapping with an Active Corner Camera", "abstract": "The ability to form non-line-of-sight (NLOS) images of changing scenes could be transformative in a variety of fields, including search and rescue, autonomous vehicle navigation, and reconnaissance. Most existing active NLOS methods illuminate the hidden scene using a pulsed laser directed at a relay surface and collect time-resolved measurements of returning light. The prevailing approaches include raster scanning of a rectangular grid on a vertical wall opposite the volume of interest to generate a collection of confocal measurements. These are inherently limited by the need for laser scanning. Methods that avoid laser scanning track the moving parts of the hidden scene as one or two point targets. In this work, based on more complete optical response modeling yet still without multiple illumination positions, we demonstrate accurate reconstructions of objects in motion and a 'map' of the stationary scenery behind them. The ability to count, localize, and characterize the sizes of hidden objects in motion, combined with mapping of the stationary hidden scene, could greatly improve indoor situational awareness in a variety of applications.", "date": "2022-08-02", "categories": []}
{"id": "2308.05818", "title": "Absorption-Based, Passive Range Imaging from Hyperspectral Thermal Measurements", "abstract": "Passive hyperspectral longwave infrared measurements are remarkably informative about the surroundings. Remote object material and temperature determine the spectrum of thermal radiance, and range, air temperature, and gas concentrations determine how this spectrum is modified by propagation to the sensor. We introduce a passive range imaging method based on computationally separating these phenomena. Previous methods assume hot and highly emitting objects; ranging is more challenging when objects' temperatures do not deviate greatly from air temperature. Our method jointly estimates range and intrinsic object properties, with explicit consideration of air emission, though reflected light is assumed negligible. Inversion being underdetermined is mitigated by using a parametric model of atmospheric absorption and regularizing for smooth emissivity estimates. To assess where our estimate is likely accurate, we introduce a technique to detect which scene pixels are significantly influenced by reflected downwelling. Monte Carlo simulations demonstrate the importance of regularization, temperature differentials, and availability of many spectral bands. We apply our method to longwave infrared (8--13 $\\mu$m) hyperspectral image data acquired from natural scenes with no active illumination. Range features from 15m to 150m are recovered, with good qualitative match to lidar data for pixels classified as having negligible reflected downwelling.", "date": "2023-08-10", "categories": []}
{"id": "2603.10456", "title": "LCAMV: High-Accuracy 3D Reconstruction of Color-Varying Objects Using LCA Correction and Minimum-Variance Fusion in Structured Light", "abstract": "Accurate 3D reconstruction of colored objects with structured light (SL) is hindered by lateral chromatic aberration (LCA) in optical components and uneven noise characteristics across RGB channels. This paper introduces lateral chromatic aberration correction and minimum-variance fusion (LCAMV), a robust 3D reconstruction method that operates with a single projector-camera pair without additional hardware or acquisition constraints. LCAMV analytically models and pixel-wise compensates LCA in both the projector and camera, then adaptively fuses multi-channel phase data using a Poisson-Gaussian noise model and minimum-variance estimation. Unlike existing methods that require extra hardware or multiple exposures, LCAMV enables fast acquisition. Experiments on planar and non-planar colored surfaces show that LCAMV outperforms grayscale conversion and conventional channel-weighting, reducing depth error by up to 43.6\\%. These results establish LCAMV as an effective solution for high-precision 3D reconstruction of nonuniformly colored objects.", "date": "2026-03-11", "categories": []}
{"id": "2603.10551", "title": "P-GSVC: Layered Progressive 2D Gaussian Splatting for Scalable Image and Video", "abstract": "Gaussian splatting has emerged as a competitive explicit representation for image and video reconstruction. In this work, we present P-GSVC, the first layered progressive 2D Gaussian splatting framework that provides a unified solution for scalable Gaussian representation in both images and videos. P-GSVC organizes 2D Gaussian splats into a base layer and successive enhancement layers, enabling coarse-to-fine reconstructions. To effectively optimize this layered representation, we propose a joint training strategy that simultaneously updates Gaussians across layers, aligning their optimization trajectories to ensure inter-layer compatibility and a stable progressive reconstruction. P-GSVC supports scalability in terms of both quality and resolution. Our experiments show that the joint training strategy can gain up to 1.9 dB improvement in PSNR for video and 2.6 dB improvement in PSNR for image when compared to methods that perform sequential layer-wise training. Project page: https://longanwang-cs.github.io/PGSVC-webpage/", "date": "2026-03-11", "categories": []}
{"id": "2603.09125", "title": "QUSR: Quality-Aware and Uncertainty-Guided Image Super-Resolution Diffusion Model", "abstract": "Diffusion-based image super-resolution (ISR) has shown strong potential, but it still struggles in real-world scenarios where degradations are unknown and spatially non-uniform, often resulting in lost details or visual artifacts. To address this challenge, we propose a novel super-resolution diffusion model, QUSR, which integrates a Quality-Aware Prior (QAP) with an Uncertainty-Guided Noise Generation (UNG) module. The UNG module adaptively adjusts the noise injection intensity, applying stronger perturbations to high-uncertainty regions (e.g., edges and textures) to reconstruct complex details, while minimizing noise in low-uncertainty regions (e.g., flat areas) to preserve original information. Concurrently, the QAP leverages an advanced Multimodal Large Language Model (MLLM) to generate reliable quality descriptions, providing an effective and interpretable quality prior for the restoration process. Experimental results confirm that QUSR can produce high-fidelity and high-realism images in real-world scenarios. The source code is available at https://github.com/oTvTog/QUSR.", "date": "2026-03-10", "categories": []}
{"id": "2603.09171", "title": "Progressive Split Mamba: Effective State Space Modelling for Image Restoration", "abstract": "Image restoration requires simultaneously preserving fine-grained local structures and maintaining long-range spatial coherence. While convolutional networks struggle with limited receptive fields, and Transformers incur quadratic complexity for global attention, recent State Space Models (SSMs), such as Mamba, provide an appealing linear-time alternative for long-range dependency modelling. However, naively extending Mamba to 2D images exposes two intrinsic shortcomings. First, flattening 2D feature maps into 1D sequences disrupts spatial topology, leading to locality distortion that hampers precise structural recovery. Second, the stability-driven recurrent dynamics of SSMs induce long-range decay, progressively attenuating information across distant spatial positions and weakening global consistency. Together, these effects limit the effectiveness of state-space modelling in high-fidelity restoration. We propose Progressive Split-Mamba (PS-Mamba), a topology-aware hierarchical state-space framework designed to reconcile locality preservation with efficient global propagation. Instead of sequentially flattening entire feature maps, PS-Mamba performs geometry-consistent partitioning, maintaining neighbourhood integrity prior to state-space processing. A progressive split hierarchy (halves, quadrants, octants) enables structured multi-scale modelling while retaining linear complexity. To counteract long-range decay, we introduce symmetric cross-scale shortcut pathways that directly transmit low-frequency global context across hierarchical levels, stabilising information flow over large spatial extents. Extensive experiments on super-resolution, denoising, and JPEG artifact reduction show consistent improvements over recent Mamba-based and attention-based models with a clear margin.", "date": "2026-03-10", "categories": []}
{"id": "2603.09548", "title": "A comprehensive study of time-of-flight non-line-of-sight imaging", "abstract": "Time-of-Flight non-line-of-sight (ToF NLOS) imaging techniques provide state-of-the-art reconstructions of scenes hidden around corners by inverting the optical path of indirect photons scattered by visible surfaces and measured by picosecond resolution sensors. The emergence of a wide range of ToF NLOS imaging methods with heterogeneous formulae and hardware implementations obscures the assessment of both their theoretical and experimental aspects. We present a comprehensive study of a representative set of ToF NLOS imaging methods by discussing their similarities and differences under common formulation and hardware. We first outline the problem statement under a common general forward model for ToF NLOS measurements, and the typical assumptions that yield tractable inverse models. We discuss the relationship of the resulting simplified forward and inverse models to a family of Radon transforms, and how migrating these to the frequency domain relates to recent phasor-based virtual line-of-sight imaging models for NLOS imaging that obey the constraints of conventional lens-based imaging systems. We then evaluate performance of the selected methods on hidden scenes captured under the same hardware setup and similar photon counts. Our experiments show that existing methods share similar limitations on spatial resolution, visibility, and sensitivity to noise when operating under equal hardware constraints, with particular differences that stem from method-specific parameters. We expect our methodology to become a reference in future research on ToF NLOS imaging to obtain objective comparisons of existing and new methods.", "date": "2026-03-10", "categories": []}
{"id": "2506.18882", "title": "Light of Normals: Unified Feature Representation for Universal Photometric Stereo", "abstract": "Universal photometric stereo (PS) is defined by two factors: it must (i) operate under arbitrary, unknown lighting conditions and (ii) avoid reliance on specific illumination models. Despite progress (e.g., SDM UniPS), two challenges remain. First, current encoders cannot guarantee that illumination and normal information are decoupled. To enforce decoupling, we introduce LINO UniPS with two key components: (i) Light Register Tokens with light alignment supervision to aggregate point, direction, and environment lights; (ii) Interleaved Attention Block featuring global cross-image attention that takes all lighting conditions together so the encoder can factor out lighting while retaining normal-related evidence. Second, high-frequency geometric details are easily lost. We address this with (i) a Wavelet-based Dual-branch Architecture and (ii) a Normal-gradient Perception Loss. These techniques yield a unified feature space in which lighting is explicitly represented by register tokens, while normal details are preserved via wavelet branch. We further introduce PS-Verse, a large-scale synthetic dataset graded by geometric complexity and lighting diversity, and adopt curriculum training from simple to complex scenes. Extensive experiments show new state-of-the-art results on public benchmarks (e.g., DiLiGenT, Luces), stronger generalization to real materials, and improved efficiency; ablations confirm that Light Register Tokens + Interleaved Attention Block drive better feature decoupling, while Wavelet-based Dual-branch Architecture + Normal-gradient Perception Loss recover finer details.", "date": "2025-06-23", "categories": []}
{"id": "2507.00790", "title": "LD-RPS: Zero-Shot Unified Image Restoration via Latent Diffusion Recurrent Posterior Sampling", "abstract": "Unified image restoration is a significantly challenging task in low-level vision. Existing methods either make tailored designs for specific tasks, limiting their generalizability across various types of degradation, or rely on training with paired datasets, thereby suffering from closed-set constraints. To address these issues, we propose a novel, dataset-free, and unified approach through recurrent posterior sampling utilizing a pretrained latent diffusion model. Our method incorporates the multimodal understanding model to provide sematic priors for the generative model under a task-blind condition. Furthermore, it utilizes a lightweight module to align the degraded input with the generated preference of the diffusion model, and employs recurrent refinement for posterior sampling. Extensive experiments demonstrate that our method outperforms state-of-the-art methods, validating its effectiveness and robustness. Our code and data are available at https://github.com/AMAP-ML/LD-RPS.", "date": "2025-07-01", "categories": []}
{"id": "2509.09894", "title": "Physics-Aware Neural Operators for Direct Inversion in 3D Photoacoustic Tomography", "abstract": "Learning physics-constrained inverse operators-rather than post-processing physics-based reconstructions-is a broadly applicable strategy for problems with expensive forward models. We demonstrate this principle in three-dimensional photoacoustic computed tomography (3D PACT), where current systems demand dense transducer arrays and prolonged scans, restricting clinical translation. We introduce PANO (PACT imaging neural operator), an end-to-end physics-aware neural operator-a deep learning architecture that generalizes across input sampling densities without retraining-that directly learns the inverse mapping from raw sensor measurements to a 3D volumetric image. Unlike two-step methods that reconstruct then denoise, PANO performs direct inversion in a single pass, jointly embedding physics and data priors. It employs spherical discrete-continuous convolutions to respect hemispherical sensor geometry and Helmholtz equation constraints to ensure physical consistency. PANO reconstructs high-quality images from both simulated and real data across diverse sparse acquisition settings, achieves real-time inference and outperforms the widely-used UBP algorithm by approximately 33 percentage points in cosine similarity on simulated data and 14 percentage points on real phantom data. These results establish a pathway toward more accessible 3D PACT systems for preclinical research, and motivate future in-vivo validation for clinical translation.", "date": "2025-09-11", "categories": []}
{"id": "2603.06689", "title": "High-Resolution Image Reconstruction with Unsupervised Learning and Noisy Data Applied to Ion-Beam Dynamics for Particle Accelerators", "abstract": "Image reconstruction in the presence of severe degradation remains a challenging inverse problem, particularly in beam diagnostics for high-energy physics accelerators. As modern facilities demand precise detection of beam halo structures to control losses, traditional analysis tools have reached their performance limits. This work reviews existing image-processing techniques for data cleaning, contour extraction, and emittance reconstruction, and introduces a novel approach based on convolutional filtering and neural networks with optimized early-stopping strategies in order to control overfitting. Despite the absence of training datasets, the proposed unsupervised framework achieves robust denoising and high-fidelity reconstruction of beam emittance images under low signal-to-noise conditions. The method extends measurable amplitudes beyond seven standard deviations, enabling unprecedented halo resolution.", "date": "2026-03-04", "categories": []}
{"id": "2603.06852", "title": "Active View Selection with Perturbed Gaussian Ensemble for Tomographic Reconstruction", "abstract": "Sparse-view computed tomography (CT) is critical for reducing radiation exposure to patients. Recent advances in radiative 3D Gaussian Splatting (3DGS) have enabled fast and accurate sparse-view CT reconstruction. Despite these algorithmic advancements, practical reconstruction fidelity remains fundamentally bounded by the quality of the captured data, raising the crucial yet underexplored problem of X-ray active view selection. Existing active view selection methods are primarily designed for natural-light scenes and fail to capture the unique geometric ambiguities and physical attenuation properties inherent in X-ray imaging. In this paper, we present Perturbed Gaussian Ensemble, an active view selection framework that integrates uncertainty modeling with sequential decision-making, tailored for X-ray Gaussian Splatting. Specifically, we identify low-density Gaussian primitives that are likely to be uncertain and apply stochastic density scaling to construct an ensemble of plausible Gaussian density fields. For each candidate projection, we measure the structural variance of the ensemble predictions and select the one with the highest variance as the next best view. Extensive experimental results on arbitrary-trajectory CT benchmarks demonstrate that our density-guided perturbation strategy effectively eliminates geometric artifacts and consistently outperforms existing baselines in progressive tomographic reconstruction under unified view selection protocols.", "date": "2026-03-06", "categories": []}
{"id": "2508.06490", "title": "Multivariate Fields of Experts for Convergent Image Reconstruction", "abstract": "We introduce the multivariate fields of experts, a new framework for the learning of image priors. Our model generalizes existing fields of experts methods by incorporating multivariate potential functions constructed via Moreau envelopes of the $\\ell_\\infty$-norm. We demonstrate the effectiveness of our proposal across a range of inverse problems that include image denoising, deblurring, compressed-sensing magnetic-resonance imaging, and computed tomography. The proposed approach outperforms comparable univariate models and achieves performance close to that of deep-learning-based regularizers while being significantly faster, requiring fewer parameters, and being trained on substantially fewer data. In addition, our model retains a high level of interpretability due to its structured design. It is supported by theoretical convergence guarantees which ensure reliability in sensitive reconstruction tasks.", "date": "2025-08-08", "categories": []}
{"id": "2511.17353", "title": "Learning Latent Transmission and Glare Maps for Lens Veiling Glare Removal", "abstract": "Beyond the commonly recognized optical aberrations, the imaging performance of simplified optical systems--including single-lens and metalens designs--is often further degraded by veiling glare caused by stray-light scattering from non-ideal optical surfaces and coatings, particularly in complex real-world environments. This compound degradation undermines traditional lens aberration correction yet remains underexplored. A major challenge is that conventional scattering models (e.g., for dehazing) fail to fit veiling glare due to its spatial-varying and depth-independent nature. Consequently, paired high-quality data are difficult to prepare via simulation, hindering application of data-driven veiling glare removal models. To this end, we propose VeilGen, a generative model that learns to simulate veiling glare by estimating its underlying optical transmission and glare maps in an unsupervised manner from target images, regularized by Stable Diffusion (SD)-based priors. VeilGen enables paired dataset generation with realistic compound degradation of optical aberrations and veiling glare, while also providing the estimated latent optical transmission and glare maps to guide the veiling glare removal process. We further introduce DeVeiler, a restoration network trained with a reversibility constraint, which utilizes the predicted latent maps to guide an inverse process of the learned scattering model. Extensive experiments on challenging simplified optical systems demonstrate that our approach delivers superior restoration quality and physical fidelity compared with existing methods. These suggest that VeilGen reliably synthesizes realistic veiling glare, and its learned latent maps effectively guide the restoration process in DeVeiler. All code and datasets will be publicly released at https://github.com/XiaolongQian/DeVeiler.", "date": "2025-11-21", "categories": []}
{"id": "2602.21915", "title": "Protein Graph Neural Networks for Heterogeneous Cryo-EM Reconstruction", "abstract": "We present a geometry-aware method for heterogeneous single-particle cryogenic electron microscopy (cryo-EM) reconstruction that predicts atomic backbone conformations. To incorporate protein-structure priors, we represent the backbone as a graph and use a graph neural network (GNN) autodecoder that maps per-image latent variables to 3D displacements of a template conformation. The objective combines a data-discrepancy term based on a differentiable cryo-EM forward model with geometric regularization, and it supports unknown orientations via ellipsoidal support lifting (ESL) pose estimation. On synthetic datasets derived from molecular dynamics trajectories, the proposed GNN achieves higher accuracy compared to a multilayer perceptron (MLP) of comparable size, highlighting the benefits of a geometry-informed inductive bias.", "date": "2026-02-25", "categories": []}
{"id": "2603.05681", "title": "Gabor Primitives for Accelerated Cardiac Cine MRI Reconstruction", "abstract": "Accelerated cardiac cine MRI requires reconstructing spatiotemporal images from highly undersampled k-space data. Implicit neural representations (INRs) enable scan-specific reconstruction without large training datasets, but encode content implicitly in network weights without physically interpretable parameters. Gaussian primitives provide an explicit and geometrically interpretable alternative, but their spectra are confined near the k-space origin, limiting high-frequency representation. We propose Gabor primitives for MRI reconstruction, modulating each Gaussian envelope with a complex exponential to place its spectral support at an arbitrary k-space location, enabling efficient representation of both smooth structures and sharp boundaries. To exploit spatiotemporal redundancy in cardiac cine, we decompose per-primitive temporal variation into a low-rank geometry basis capturing cardiac motion and a signal-intensity basis modeling contrast changes. Experiments on cardiac cine data with Cartesian and radial trajectories show that Gabor primitives consistently outperform compressed sensing, Gaussian primitives, and hash-grid INR baselines, while providing a compact, continuous-resolution representation with physically meaningful parameters.", "date": "2026-03-05", "categories": []}
{"id": "2603.05696", "title": "Autonomous Algorithm Discovery for Ptychography via Evolutionary LLM Reasoning", "abstract": "Ptychography is a computational imaging technique widely used for high-resolution materials characterization, but high-quality reconstructions often require the use of regularization functions that largely remain manually designed. We introduce Ptychi-Evolve, an autonomous framework that uses large language models (LLMs) to discover and evolve novel regularization algorithms. The framework combines LLM-driven code generation with evolutionary mechanisms, including semantically-guided crossover and mutation. Experiments on three challenging datasets (X-ray integrated circuits, low-dose electron microscopy of apoferritin, and multislice imaging with crosstalk artifacts) demonstrate that discovered regularizers outperform conventional reconstructions, achieving up to +0.26 SSIM and +8.3~dB PSNR improvements. Besides, Ptychi-Evolve records algorithm lineage and evolution metadata, enabling interpretable and reproducible analysis of discovered regularizers.", "date": "2026-03-05", "categories": []}
{"id": "2603.05940", "title": "SLER-IR: Spherical Layer-wise Expert Routing for All-in-One Image Restoration", "abstract": "Image restoration under diverse degradations remains challenging for unified all-in-one frameworks due to feature interference and insufficient expert specialization. We propose SLER-IR, a spherical layer-wise expert routing framework that dynamically activates specialized experts across network layers. To ensure reliable routing, we introduce a Spherical Uniform Degradation Embedding with contrastive learning, which maps degradation representations onto a hypersphere to eliminate geometry bias in linear embedding spaces. In addition, a Global-Local Granularity Fusion (GLGF) module integrates global semantics and local degradation cues to address spatially non-uniform degradations and the train-test granularity gap. Experiments on three-task and five-task benchmarks demonstrate that SLER-IR achieves consistent improvements over state-of-the-art methods in both PSNR and SSIM. Code and models will be publicly released.", "date": "2026-03-06", "categories": []}
{"id": "2504.10288", "title": "Noise2Ghost: Self-supervised deep convolutional reconstruction for ghost imaging", "abstract": "We present a new self-supervised deep-learning-based Ghost Imaging (GI) reconstruction method, which provides unparalleled reconstruction quality for noisy acquisitions among unsupervised methods. We present the supporting mathematical framework and results from theoretical and real data use cases. Self-supervision removes the need for clean reference data while offering strong noise reduction. This provides the necessary tools for addressing signal-to-noise ratio concerns for GI acquisitions in emerging and cutting-edge low-light GI scenarios. Notable examples include micro- and nano-scale x-ray emission imaging, e.g., x-ray fluorescence imaging of dose-sensitive samples. Their applications include in-vivo and in-operando case studies for biological samples and batteries.", "date": "2025-04-14", "categories": []}
{"id": "2603.04538", "title": "InverseNet: Benchmarking Operator Mismatch and Calibration Across Compressive Imaging Modalities", "abstract": "State-of-the-art EfficientSCI loses 20.58 dB when its assumed forward operator deviates from physical reality in just eight parameters, yet no existing benchmark quantifies operator mismatch, the default condition in deployed compressive imaging systems. We introduce InverseNet, the first cross-modality benchmark for operator mismatch, spanning CASSI, CACTI, and single-pixel cameras. Evaluating 12 methods under a four-scenario protocol (ideal, mismatched, oracle-corrected, blind calibration) across 27 simulated scenes and 9 real hardware captures, we find: (1) deep learning methods lose 10-21 dB under mismatch, eliminating their advantage over classical baselines; (2) performance and robustness are inversely correlated across modalities (Spearman r_s = -0.71, p < 0.01); (3) mask-oblivious architectures recover 0% of mismatch losses regardless of calibration quality, while operator-conditioned methods recover 41-90%; (4) blind grid-search calibration recovers 85-100% of the oracle bound without ground truth. Real hardware experiments confirm that simulation trends transfer to physical data. Code will be released upon acceptance.", "date": "2026-03-04", "categories": []}
{"id": "2603.04745", "title": "Toward Real-world Infrared Image Super-Resolution: A Unified Autoregressive Framework and Benchmark Dataset", "abstract": "Infrared image super-resolution (IISR) under real-world conditions is a practically significant yet rarely addressed task. Pioneering works are often trained and evaluated on simulated datasets or neglect the intrinsic differences between infrared and visible imaging. In practice, however, real infrared images are affected by coupled optical and sensing degradations that jointly deteriorate both structural sharpness and thermal fidelity. To address these challenges, we propose Real-IISR, a unified autoregressive framework for real-world IISR that progressively reconstructs fine-grained thermal structures and clear backgrounds in a scale-by-scale manner via thermal-structural guided visual autoregression. Specifically, a Thermal-Structural Guidance module encodes thermal priors to mitigate the mismatch between thermal radiation and structural edges. Since non-uniform degradations typically induce quantization bias, Real-IISR adopts a Condition-Adaptive Codebook that dynamically modulates discrete representations based on degradation-aware thermal priors. Also, a Thermal Order Consistency Loss enforces a monotonic relation between temperature and pixel intensity, ensuring relative brightness order rather than absolute values to maintain physical consistency under spatial misalignment and thermal drift. We build FLIR-IISR, a real-world IISR dataset with paired LR-HR infrared images acquired via automated focus variation and motion-induced blur. Extensive experiments demonstrate the promising performance of Real-IISR, providing a unified foundation for real-world IISR and benchmarking. The dataset and code are available at: https://github.com/JZD151/Real-IISR.", "date": "2026-03-05", "categories": []}
{"id": "2603.04817", "title": "Revisiting Shape from Polarization in the Era of Vision Foundation Models", "abstract": "We show that, with polarization cues, a lightweight model trained on a small dataset can outperform RGB-only vision foundation models (VFMs) in single-shot object-level surface normal estimation. Shape from polarization (SfP) has long been studied due to the strong physical relationship between polarization and surface geometry. Meanwhile, driven by scaling laws, RGB-only VFMs trained on large datasets have recently achieved impressive performance and surpassed existing SfP methods. This situation raises questions about the necessity of polarization cues, which require specialized hardware and have limited training data. We argue that the weaker performance of prior SfP methods does not come from the polarization modality itself, but from domain gaps. These domain gaps mainly arise from two sources. First, existing synthetic datasets use limited and unrealistic 3D objects, with simple geometry and random texture maps that do not match the underlying shapes. Second, real-world polarization signals are often affected by sensor noise, which is not well modeled during training. To address the first issue, we render a high-quality polarization dataset using 1,954 3D-scanned real-world objects. We further incorporate pretrained DINOv3 priors to improve generalization to unseen objects. To address the second issue, we introduce polarization sensor-aware data augmentation that better reflects real-world conditions. With only 40K training scenes, our method significantly outperforms both state-of-the-art SfP approaches and RGB-only VFMs. Extensive experiments show that polarization cues enable a 33x reduction in training data or an 8x reduction in model parameters, while still achieving better performance than RGB-only counterparts.", "date": "2026-03-05", "categories": []}
{"id": "2603.04975", "title": "BiEvLight: Bi-level Learning of Task-Aware Event Refinement for Low-Light Image Enhancement", "abstract": "Event cameras, with their high dynamic range, show great promise for Low-light Image Enhancement (LLIE). Existing works primarily focus on designing effective modal fusion strategies. However, a key challenge is the dual degradation from intrinsic background activity (BA) noise in events and low signal-to-noise ratio (SNR) in images, which causes severe noise coupling during modal fusion, creating a critical performance bottleneck. We therefore posit that precise event denoising is the prerequisite to unlocking the full potential of event-based fusion. To this end, we propose BiEvLight, a hierarchical and task-aware framework that collaboratively optimizes enhancement and denoising by exploiting their intrinsic interdependence. Specifically, BiEvLight exploits the strong gradient correlation between images and events to build a gradient-guided event denoising prior that alleviates insufficient denoising in heavily noisy regions. Moreover, instead of treating event denoising as a static pre-processing stage-which inevitably incurs a trade-off between over- and under-denoising and cannot adapt to the requirements of a specific enhancement objective-we recast it as a bilevel optimization problem constrained by the enhancement task. Through cross-task interaction, the upper-level denoising problem learns event representations tailored to the lower-level enhancement objective, thereby substantially improving overall enhancement quality. Extensive experiments on the Real-world noise Dataset SDE demonstrate that our method significantly outperforms state-of-the-art (SOTA) approaches, with average improvements of 1.30dB in PSNR, 2.03dB in PSNR* and 0.047 in SSIM, respectively. The code will be publicly available at https://github.com/iijjlk/BiEvlight.", "date": "2026-03-05", "categories": []}
{"id": "2603.04999", "title": "Physics-consistent deep learning for blind aberration recovery in mobile optics", "abstract": "Mobile photography is often limited by complex, lens-specific optical aberrations. While recent deep learning methods approach this as an end-to-end deblurring task, these \"black-box\" models lack explicit optical modeling and can hallucinate details. Conversely, classical blind deconvolution remains highly unstable. To bridge this gap, we present Lens2Zernike, a deep learning framework that blindly recovers physical optical parameters from a single blurred image. To the best of our knowledge, no prior work has simultaneously integrated supervision across three distinct optical domains. We introduce a novel physics-consistent strategy that explicitly minimizes errors via direct Zernike coefficient regression (z), differentiable physics constraints encompassing both wavefront and point spread function derivations (p), and auxiliary multi-task spatial map predictions (m). Through an ablation study on a ResNet-18 backbone, we demonstrate that our full multi-task framework (z+p+m) yields a 35% improvement over coefficient-only baselines. Crucially, comparative analysis reveals that our approach outperforms two established deep learning methods from previous literature, achieving significantly lower regression errors. Ultimately, we demonstrate that these recovered physical parameters enable stable non-blind deconvolution, providing substantial in-domain improvement on the patented Institute for Digital Molecular Analytics and Science (IDMxS) Mobile Camera Lens Database for restoring diffraction-limited details from severely aberrated mobile captures.", "date": "2026-03-05", "categories": []}
{"id": "2506.03317", "title": "Structural Vibration Monitoring with Diffractive Optical Processors", "abstract": "Structural Health Monitoring (SHM) is vital for maintaining the safety and longevity of civil infrastructure, yet current solutions remain constrained by cost, power consumption, scalability, and the complexity of data processing. Here, we present a diffractive vibration monitoring system, integrating a jointly optimized diffractive layer with a shallow neural network-based backend to remotely extract 3D structural vibration spectra, offering a low-power, cost-effective and scalable solution. This architecture eliminates the need for dense sensor arrays or extensive data acquisition; instead, it uses a spatially-optimized passive diffractive layer that encodes 3D structural displacements into modulated light, captured by a minimal number of detectors and decoded in real-time by shallow and low-power neural networks to reconstruct the 3D displacement spectra of structures. The diffractive system's efficacy was demonstrated both numerically and experimentally using millimeter-wave illumination on a laboratory-scale building model with a programmable shake table. Our system achieves more than an order-of-magnitude improvement in accuracy over conventional optics or separately trained modules, establishing a foundation for high-throughput 3D monitoring of structures. Beyond SHM, the 3D vibration monitoring capabilities of this cost-effective and data-efficient framework establish a new computational sensing modality with potential applications in disaster resilience, aerospace diagnostics, and autonomous navigation, where energy efficiency, low latency, and high-throughput are critical.", "date": "2025-06-03", "categories": []}
{"id": "2603.02887", "title": "Generalized non-exponential Gaussian splatting", "abstract": "In this work we generalize 3D Gaussian splatting (3DGS) to a wider family of physically-based alpha-blending operators. 3DGS has become the standard de-facto for radiance field rendering and reconstruction, given its flexibility and efficiency. At its core, it is based on alpha-blending sorted semitransparent primitives, which in the limit converges to the classic radiative transfer function with exponential transmittance. Inspired by recent research on non-exponential radiative transfer, we generalize the image formation model of 3DGS to non-exponential regimes. Based on this generalization, we use a quadratic transmittance to define sub-linear, linear, and super-linear versions of 3DGS, which exhibit faster-than-exponential decay. We demonstrate that these new non-exponential variants achieve similar quality than the original 3DGS but significantly reduce the number of overdraws, which result on speed-ups of up to $4\\times$ in complex real-world captures, on a ray-tracing-based renderer.", "date": "2026-03-03", "categories": []}
{"id": "2603.03839", "title": "All-in-One Image Restoration via Causal-Deconfounding Wavelet-Disentangled Prompt Network", "abstract": "Image restoration represents a promising approach for addressing the inherent defects of image content distortion. Standard image restoration approaches suffer from high storage cost and the requirement towards the known degradation pattern, including type and degree, which can barely be satisfied in dynamic practical scenarios. In contrast, all-in-one image restoration (AiOIR) eliminates multiple degradations within a unified model to circumvent the aforementioned issues. However, according to our causal analysis, we disclose that two significant defects still exacerbate the effectiveness and generalization of AiOIR models: 1) the spurious correlation between non-degradation semantic features and degradation patterns; 2) the biased estimation of degradation patterns. To obtain the true causation between degraded images and restored images, we propose Causal-deconfounding Wavelet-disentangled Prompt Network (CWP-Net) to perform effective AiOIR. CWP-Net introduces two modules for decoupling, i.e., wavelet attention module of encoder and wavelet attention module of decoder. These modules explicitly disentangle the degradation and semantic features to tackle the issue of spurious correlation. To address the issue stemming from the biased estimation of degradation patterns, CWP-Net leverages a wavelet prompt block to generate the alternative variable for causal deconfounding. Extensive experiments on two all-in-one settings prove the effectiveness and superior performance of our proposed CWP-Net over the state-of-the-art AiOIR methods.", "date": "2026-03-04", "categories": []}
{"id": "2603.02692", "title": "FiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution", "abstract": "Diffusion-based approaches have recently driven remarkable progress in real-world image super-resolution (SR). However, existing methods still struggle to simultaneously preserve fine details and ensure high-fidelity reconstruction, often resulting in suboptimal visual quality. In this paper, we propose FiDeSR, a high-fidelity and detail-preserving one-step diffusion super-resolution framework. During training, we introduce a detail-aware weighting strategy that adaptively emphasizes regions where the model exhibits higher prediction errors. During inference, low- and high-frequency adaptive enhancers further refine the reconstruction without requiring model retraining, enabling flexible enhancement control. To further improve the reconstruction accuracy, FiDeSR incorporates a residual-in-residual noise refinement, which corrects prediction errors in the diffusion noise and enhances fine detail recovery. FiDeSR achieves superior real-world SR performance compared to existing diffusion-based methods, producing outputs with both high perceptual quality and faithful content restoration. The source code will be released at: https://github.com/Ar0Kim/FiDeSR.", "date": "2026-03-03", "categories": []}
{"id": "2603.02710", "title": "MiM-DiT: MoE in MoE with Diffusion Transformers for All-in-One Image Restoration", "abstract": "All-in-one image restoration is challenging because different degradation types, such as haze, blur, noise, and low-light, impose diverse requirements on restoration strategies, making it difficult for a single model to handle them effectively. In this paper, we propose a unified image restoration framework that integrates a dual-level Mixture-of-Experts (MoE) architecture with a pretrained diffusion model. The framework operates at two levels: the Inter-MoE layer adaptively combines expert groups to handle major degradation types, while the Intra-MoE layer further selects specialized sub-experts to address fine-grained variations within each type. This design enables the model to achieve coarse-grained adaptation across diverse degradation categories while performing fine-grained modulation for specific intra-class variations, ensuring both high specialization in handling complex, real-world corruptions. Extensive experiments demonstrate that the proposed method performs favorably against the state-of-the-art approaches on multiple image restoration task.", "date": "2026-03-03", "categories": []}
{"id": "2305.17185", "title": "Task-Driven Lens Design", "abstract": "Classical lens design minimizes optical aberrations to produce sharp images, but is typically decoupled from downstream computer vision tasks. Existing end-to-end optical design learns optical encoding through joint optimization, but often suffers from an unstable training process. We propose task-driven lens design, a new optimization philosophy for joint optics-network systems. We freeze the pretrained vision model and optimize only the lens so that the image formation better fits the model's feature preferences. This network-frozen setting yields a low-dimensional and stable optimization process, enabling lens design from scratch without human intervention, thereby exploring a broader design space. Multiple computer vision experiments show that TaskLenses outperform classical ImagingLenses with the same or even fewer elements. Our analysis reveals that the learned optics exhibit long-tailed point spread functions, better preserving preferred structural cues when aberrations cannot be fully corrected. These results highlight task-driven design as a practical route for optical lenses that are compatible with modern vision models, and also inspire new optical design objectives beyond traditional aberration minimization.", "date": "2023-05-26", "categories": []}
{"id": "2505.22973", "title": "EquiReg: Equivariance Regularized Diffusion for Inverse Problems", "abstract": "Diffusion models represent the state-of-the-art for solving inverse problems such as image restoration tasks. Diffusion-based inverse solvers incorporate a likelihood term to guide prior sampling, generating data consistent with the posterior distribution. However, due to the intractability of the likelihood, most methods rely on isotropic Gaussian approximations, which can push estimates off the data manifold and produce inconsistent, poor reconstructions. We propose Equivariance Regularized (EquiReg) diffusion, a general plug-and-play framework that improves posterior sampling by penalizing trajectories that deviate from the data manifold. EquiReg formalizes manifold-preferential equivariant functions that exhibit low equivariance error for on-manifold samples and high error for off-manifold ones, thereby guiding sampling toward symmetry-preserving regions of the solution space. We highlight that such functions naturally emerge when training non-equivariant models with augmentation or on data with symmetries. EquiReg is particularly effective under reduced sampling and measurement consistency steps, where many methods suffer severe quality degradation. By regularizing trajectories toward the manifold, EquiReg implicitly accelerates convergence and enables high-quality reconstructions. EquiReg consistently improves performance in linear and nonlinear image restoration tasks and solving partial differential equations.", "date": "2025-05-29", "categories": []}
{"id": "2508.02923", "title": "A Morse-Bott Framework for Blind Inverse Problems: Local Recovery Guarantees and the Failure of the MAP", "abstract": "Maximum A Posteriori (MAP) estimation is a cornerstone framework for blind inverse problems, where an image and a forward operator are jointly estimated as the maximizers of a posterior distribution. In this paper, we analyze the recovery guarantees of MAP-based methods by adopting a Morse-Bott framework. We model the image prior potential as a Morse-Bott function, where natural images are modeled as residing locally on a critical submanifold. This means that while the potential is locally flat along the natural directions of the image manifold, it is strictly convex in the directions normal to it. We demonstrate that this Morse-Bott hypothesis aligns with the structural properties of state-of-the-art learned priors, a finding we validate through an experimental analysis of the potential landscape and its Hessian spectrum. Our theoretical results show that, in a neighborhood of the ground-truth image and operator, the posterior admits local minimizers that are stable both with respect to initialization (gradient steps converge to the same minimizer) and to small noise perturbations (solutions vary smoothly). This local stability explains the empirical success of well-designed gradient-based optimization in these settings. However, we also demonstrate that this stability is a local property: the blurry trap, well-known for sparse priors in blind deconvolution, persists even with state-of-the-art learned priors. Our findings demonstrate that the failure of MAP in blind deconvolution is not a limitation of prior quality, but an intrinsic characteristic of the landscape. We conclude that successful recovery of posterior maximization depends on strategic initialization within the basin of favorable local minima, and we validate this with numerical experiments on both synthetic and real-world data.", "date": "2025-08-04", "categories": []}
{"id": "2512.10683", "title": "Optimal transport unlocks end-to-end learning for single-molecule localization", "abstract": "Single-molecule localization microscopy (SMLM) allows reconstructing biology-relevant structures beyond the diffraction limit by detecting and localizing individual fluorophores -- fluorescent molecules stained onto the observed specimen -- over time to reconstruct super-resolved images. Currently, efficient SMLM requires non-overlapping emitting fluorophores, leading to long acquisition times that hinders live-cell imaging. Recent deep-learning approaches can handle denser emissions, but they rely on variants of non-maximum suppression (NMS) layers, which are unfortunately non-differentiable and may discard true positives with their local fusion strategy. In this presentation, we reformulate the SMLM training objective as a set-matching problem, deriving an optimal-transport loss that eliminates the need for NMS during inference and enables end-to-end training. Additionally, we propose an iterative neural network that integrates knowledge of the microscope's optical system inside our model. Experiments on synthetic benchmarks and real biological data show that both our new loss function and architecture surpass the state of the art at moderate and high emitter densities. Code is available at https://github.com/RSLLES/SHOT.", "date": "2025-12-11", "categories": []}
{"id": "2602.02356", "title": "NAB: Neural Adaptive Binning for Sparse-View CT reconstruction", "abstract": "Computed Tomography (CT) plays a vital role in inspecting the internal structures of industrial objects. Furthermore, achieving high-quality CT reconstruction from sparse views is essential for reducing production costs. While classic implicit neural networks have shown promising results for sparse reconstruction, they are unable to leverage shape priors of objects. Motivated by the observation that numerous industrial objects exhibit rectangular structures, we propose a novel Neural Adaptive Binning (NAB) method that effectively integrates rectangular priors into the reconstruction process. Specifically, our approach first maps coordinate space into a binned vector space. This mapping relies on an innovative binning mechanism based on differences between shifted hyperbolic tangent functions, with our extension enabling rotations around the input-plane normal vector. The resulting representations are then processed by a neural network to predict CT attenuation coefficients. This design enables end-to-end optimization of the encoding parameters -- including position, size, steepness, and rotation -- via gradient flow from the projection data, thus enhancing reconstruction accuracy. By adjusting the smoothness of the binning function, NAB can generalize to objects with more complex geometries. This research provides a new perspective on integrating shape priors into neural network-based reconstruction. Extensive experiments demonstrate that NAB achieves superior performance on two industrial datasets. It also maintains robust on medical datasets when the binning function is extended to more general expression. The code is available at https://github.com/Wangduo-Xie/NAB_CT_reconstruction.", "date": "2026-02-02", "categories": []}
{"id": "2603.00145", "title": "M-Gaussian: An Magnetic Gaussian Framework for Efficient Multi-Stack MRI Reconstruction", "abstract": "Magnetic Resonance Imaging (MRI) is a crucial non-invasive imaging modality. In routine clinical practice, multi-stack thick-slice acquisitions are widely used to reduce scan time and motion sensitivity, particularly in challenging scenarios such as fetal brain imaging. However, the resulting severe through-plane anisotropy compromises volumetric analysis and downstream quantitative assessment, necessitating robust reconstruction of isotropic high-resolution volumes. Implicit neural representation methods, while achieving high quality, suffer from computational inefficiency due to complex network structures. We present M-Gaussian, adapting 3D Gaussian Splatting to MRI reconstruction. Our contributions include: (1) Magnetic Gaussian primitives with physics-consistent volumetric rendering, (2) neural residual field for high-frequency detail refinement, and (3) multi-resolution progressive training. Our method achieves an optimal balance between quality and speed. On the FeTA dataset, M-Gaussian achieves 40.31 dB PSNR while being 14 times faster, representing the first successful adaptation of 3D Gaussian Splatting to multi-stack MRI reconstruction.", "date": "2026-02-24", "categories": []}
{"id": "2603.00337", "title": "Diffusion-Based Low-Light Image Enhancement with Color and Luminance Priors", "abstract": "Low-light images often suffer from low contrast, noise, and color distortion, degrading visual quality and impairing downstream vision tasks. We propose a novel conditional diffusion framework for low-light image enhancement that incorporates a Structured Control Embedding Module (SCEM). SCEM decomposes a low-light image into four informative components including illumination, illumination-invariant features, shadow priors, and color-invariant cues. These components serve as control signals that condition a U-Net-based diffusion model trained with a simplified noise-prediction loss. Thus, the proposed SCEM equipped Diffusion method enforces structured enhancement guided by physical priors. In experiments, our model is trained only on the LOLv1 dataset and evaluated without fine-tuning on LOLv2-real, LSRW, DICM, MEF, and LIME. The method achieves state-of-the-art performance in quantitative and perceptual metrics, demonstrating strong generalization across benchmarks. https://casted.github.io/scem/.", "date": "2026-02-27", "categories": []}
{"id": "2603.00382", "title": "DiffSOS: Acoustic Conditional Diffusion Model for Speed-of-Sound Reconstruction in Ultrasound Computed Tomography", "abstract": "Accurate Speed-of-Sound (SoS) reconstruction from acoustic waveforms is a cornerstone of ultrasound computed tomography (USCT), enabling quantitative velocity mapping that reveals subtle anatomical details and pathological variations often invisible in conventional imaging. However, practical utility is hindered by the limitations of existing algorithms; traditional Full Waveform Inversion (FWI) is computationally intensive, while current deep learning approaches tend to produce oversmoothed results lacking fine details. We propose DiffSOS, a conditional diffusion model that directly maps acoustic waveforms to SoS maps. Our framework employs a specialized acoustic ControlNet to strictly ground the denoising process in physical wave measurements. To ensure structural consistency, we optimize a hybrid loss function that integrates noise prediction, spatial reconstruction, and noise frequency content. To accelerate inference, we employ stochastic Denoising Diffusion Implicit Model (DDIM) sampling, achieving near real-time reconstruction with only 10 steps. Crucially, we exploit the stochastic generative nature of our framework to estimate pixel-wise uncertainty, providing a measure of reliability that is often absent in deterministic approaches. Evaluated on the OpenPros USCT benchmark, DiffSOS significantly outperforms state-of-the-art networks, achieving an average Multi-scale Structural Similarity of 0.957. Our approach provides high-fidelity SoS maps with a principled measure of confidence, facilitating safer and faster clinical interpretation.", "date": "2026-02-27", "categories": []}
{"id": "2603.00467", "title": "High Dynamic Range Imaging Based on an Asymmetric Event-SVE Camera System", "abstract": "High dynamic range (HDR) imaging under extreme illumination remains challenging for conventional cameras due to overexposure. Event cameras provide microsecond temporal resolution and high dynamic range, while spatially varying exposure (SVE) sensors offer single-shot radiometric diversity.We present a hardware--algorithm co-designed HDR imaging system that tightly integrates an SVE micro-attenuation camera with an event sensor in an asymmetric dual-modality configuration. To handle non-coaxial geometry and heterogeneous optics, we develop a two-stage cross-modal alignment framework that combines feature-guided coarse homography estimation with a multi-scale refinement module based on spatial pooling and frequency-domain filtering. On top of aligned representations, we develop a cross-modal HDR reconstruction network with convolutional fusion, mutual-information regularization, and a learnable fusion loss that adaptively balances intensity cues and event-derived structural constraints. Comprehensive experiments on both synthetic benchmarks and real captures demonstrate that the proposed system consistently improves highlight recovery, edge fidelity, and robustness compared with frame-only or event-only HDR pipelines. The results indicate that jointly optimizing optical design, cross-modal alignment, and computational fusion provides an effective foundation for reliable HDR perception in highly dynamic and radiometrically challenging environments.", "date": "2026-02-28", "categories": []}
{"id": "2603.00853", "title": "Neural Discrimination-Prompted Transformers for Efficient UHD Image Restoration and Enhancement", "abstract": "We propose a simple yet effective UHDPromer, a neural discrimination-prompted Transformer, for Ultra-High-Definition (UHD) image restoration and enhancement. Our UHDPromer is inspired by an interesting observation that there implicitly exist neural differences between high-resolution and low-resolution features, and exploring such differences can facilitate low-resolution feature representation. To this end, we first introduce Neural Discrimination Priors (NDP) to measure the differences and then integrate NDP into the proposed Neural Discrimination-Prompted Attention (NDPA) and Neural Discrimination-Prompted Network (NDPN). The proposed NDPA re-formulates the attention by incorporating NDP to globally perceive useful discrimination information, while the NDPN explores a continuous gating mechanism guided by NDP to selectively permit the passage of beneficial content. To enhance the quality of restored images, we propose a super-resolution-guided reconstruction approach, which is guided by super-resolving low-resolution features to facilitate final UHD image restoration. Experiments show that UHDPromer achieves the best computational efficiency while still maintaining state-of-the-art performance on $3$ UHD image restoration and enhancement tasks, including low-light image enhancement, image dehazing, and image deblurring. The source codes and pre-trained models will be made available at https://github.com/supersupercong/uhdpromer.", "date": "2026-03-01", "categories": []}
{"id": "2603.00882", "title": "Solving a Nonlinear Blind Inverse Problem for Tagged MRI with Physics and Deep Generative Priors", "abstract": "Tagged MRI enables tracking internal tissue motion non-invasively. It encodes motion by modulating anatomy with periodic tags, which deform along with tissue. However, the entanglement between anatomy, tags and motion poses significant challenges for post-processing. The existence of tags and imaging blur hinders downstream tasks such as segmenting anatomy. Tag fading, due to T1-relaxation, disrupts the brightness constancy assumption for motion tracking. For decades, these challenges have been handled in isolation and sub-optimally. In contrast, we introduce a blind and nonlinear inverse framework for tagged MRI that, for the first time, unifies these tasks: anatomical image recovery, high-resolution cine image synthesis, and motion estimation. At its core, the synergy of MR physics and generative priors enables us to blindly estimate the unknown forward imaging models and high-resolution underlying anatomy, while simultaneously tracking 3D diffeomorphic Lagrangian motion over time. Experiments on tagged brain MRI demonstrate that our approach yields high-resolution anatomy images, cine images, and more accurate motion than specialized methods.", "date": "2026-03-01", "categories": []}
{"id": "2603.00990", "title": "MLRecon: Robust Markerless Freehand 3D Ultrasound Reconstruction via Coarse-to-Fine Pose Estimation", "abstract": "Freehand 3D ultrasound (US) reconstruction promises volumetric imaging with the flexibility of standard 2D probes, yet existing tracking paradigms face a restrictive trilemma: marker-based systems demand prohibitive costs, inside-out methods require intrusive sensor attachment, and sensorless approaches suffer from severe cumulative drift. To overcome these limitations, we present MLRecon, a robust markerless 3D US reconstruction framework delivering drift-resilient 6D probe pose tracking using a single commodity RGB-D camera. Leveraging the generalization power of vision foundation models, our pipeline enables continuous markerless tracking of the probe, augmented by a vision-guided divergence detector that autonomously monitors tracking integrity and triggers failure recovery to ensure uninterrupted scanning. Crucially, we further propose a dual-stage pose refinement network that explicitly disentangles high-frequency jitter from low-frequency bias, effectively denoising the trajectory while maintaining the kinematic fidelity of operator maneuvers. Experiments demonstrate that MLRecon significantly outperforms competing sensorless and sensor-aided methods, achieving average position errors as low as 0.88 mm on complex trajectories and yielding high-quality 3D reconstructions with sub-millimeter mean surface accuracy. This establishes a new benchmark for low-cost, accessible volumetric US imaging in resource-limited clinical settings.", "date": "2026-03-01", "categories": []}
{"id": "2603.01491", "title": "Radiometrically Consistent Gaussian Surfels for Inverse Rendering", "abstract": "Inverse rendering with Gaussian Splatting has advanced rapidly, but accurately disentangling material properties from complex global illumination effects, particularly indirect illumination, remains a major challenge. Existing methods often query indirect radiance from Gaussian primitives pre-trained for novel-view synthesis. However, these pre-trained Gaussian primitives are supervised only towards limited training viewpoints, thus lack supervision for modeling indirect radiances from unobserved views. To address this issue, we introduce radiometric consistency, a novel physically-based constraint that provides supervision towards unobserved views by minimizing the residual between each Gaussian primitive's learned radiance and its physically-based rendered counterpart. Minimizing the residual for unobserved views establishes a self-correcting feedback loop that provides supervision from both physically-based rendering and novel-view synthesis, enabling accurate modeling of inter-reflection. We then propose Radiometrically Consistent Gaussian Surfels (RadioGS), an inverse rendering framework built upon our principle by efficiently integrating radiometric consistency by utilizing Gaussian surfels and 2D Gaussian ray tracing. We further propose a finetuning-based relighting strategy that adapts Gaussian surfel radiances to new illuminations within minutes, achieving low rendering cost (<10ms). Extensive experiments on existing inverse rendering benchmarks show that RadioGS outperforms existing Gaussian-based methods in inverse rendering, while retaining the computational efficiency.", "date": "2026-03-02", "categories": []}
{"id": "2603.01725", "title": "Learning Domain-Aware Task Prompt Representations for Multi-Domain All-in-One Image Restoration", "abstract": "Recently, significant breakthroughs have been made in all-in-one image restoration (AiOIR), which can handle multiple restoration tasks with a single model. However, existing methods typically focus on a specific image domain, such as natural scene, medical imaging, or remote sensing. In this work, we aim to extend AiOIR to multiple domains and propose the first multi-domain all-in-one image restoration method, DATPRL-IR, based on our proposed Domain-Aware Task Prompt Representation Learning. Specifically, we first construct a task prompt pool containing multiple task prompts, in which task-related knowledge is implicitly encoded. For each input image, the model adaptively selects the most relevant task prompts and composes them into an instance-level task representation via a prompt composition mechanism (PCM). Furthermore, to endow the model with domain awareness, we introduce another domain prompt pool and distill domain priors from multimodal large language models into the domain prompts. PCM is utilized to combine the adaptively selected domain prompts into a domain representation for each input image. Finally, the two representations are fused to form a domain-aware task prompt representation which can make full use of both specific and shared knowledge across tasks and domains to guide the subsequent restoration process. Extensive experiments demonstrate that our DATPRL-IR significantly outperforms existing SOTA image restoration methods, while exhibiting strong generalization capabilities. Code is available at https://github.com/GuangluDong0728/DATPRL-IR.", "date": "2026-03-02", "categories": []}
{"id": "2603.01837", "title": "Constrained Particle Seeking: Solving Diffusion Inverse Problems with Just Forward Passes", "abstract": "Diffusion models have gained prominence as powerful generative tools for solving inverse problems due to their ability to model complex data distributions. However, existing methods typically rely on complete knowledge of the forward observation process to compute gradients for guided sampling, limiting their applicability in scenarios where such information is unavailable. In this work, we introduce \\textbf{\\emph{Constrained Particle Seeking (CPS)}}, a novel gradient-free approach that leverages all candidate particle information to actively search for the optimal particle while incorporating constraints aligned with high-density regions of the unconditional prior. Unlike previous methods that passively select promising candidates, CPS reformulates the inverse problem as a constrained optimization task, enabling more flexible and efficient particle seeking. We demonstrate that CPS can effectively solve both image and scientific inverse problems, achieving results comparable to gradient-based methods while significantly outperforming gradient-free alternatives. Code is available at https://github.com/deng-ai-lab/CPS.", "date": "2026-03-02", "categories": []}
{"id": "2603.01890", "title": "Resolving Blind Inverse Problems under Dynamic Range Compression via Structured Forward Operator Modeling", "abstract": "Recovering radiometric fidelity from unknown dynamic range compression (UDRC), such as low-light enhancement and HDR reconstruction, is a challenging blind inverse problem, due to the unknown forward model and irreversible information loss introduced by compression. To address this challenge, we first identify monotonicity as the fundamental physical invariant shared across UDRC tasks. Leveraging this insight, we introduce the \\textbf{cascaded monotonic Bernstein} (CaMB) operator to parameterize the unknown forward model. CaMB enforces monotonicity as a hard architectural inductive bias, constraining optimization to physically consistent mappings and enabling robust and stable operator estimation. We further integrate CaMB with a plug-and-play diffusion framework, proposing \\textbf{CaMB-Diff}. Within this framework, the diffusion model serves as a powerful geometric prior for structural and semantic recovery, while CaMB explicitly models and corrects radiometric distortions through a physically grounded forward operator. Extensive experiments on a variety of zero-shot UDRC tasks, including low-light enhancement, low-field MRI enhancement, and HDR reconstruction, demonstrate that CaMB-Diff significantly outperforms state-of-the-art zero-shot baselines in terms of both signal fidelity and physical consistency. Moreover, we empirically validate the effectiveness of the proposed CaMB parameterization in accurately modeling the unknown forward operator.", "date": "2026-03-02", "categories": []}
{"id": "2603.01913", "title": "Zero-shot Low-Field MRI Enhancement via Diffusion-Based Adaptive Contrast Transport", "abstract": "Low-field (LF) magnetic resonance imaging (MRI) democratizes access to diagnostic imaging but is fundamentally limited by low signal-to-noise ratio and significant tissue contrast distortion due to field-dependent relaxation dynamics. Reconstructing high-field (HF) quality images from LF data is a blind inverse problem, severely challenged by the scarcity of paired training data and the unknown, non-linear contrast transformation operator. Existing zero-shot methods, which assume simplified linear degradation, often fail to recover authentic tissue contrast. In this paper, we propose DACT(Diffusion-Based Adaptive Contrast Transport), a novel zero-shot framework that restores HF-quality images without paired supervision. DACT synergizes a pre-trained HF diffusion prior to ensure anatomical fidelity with a physically-informed adaptive forward model. Specifically, we introduce a differentiable Sinkhorn optimal transport module that explicitly models and corrects the intensity distribution shift between LF and HF domains during the reverse diffusion process. This allows the framework to dynamically learn the intractable contrast mapping while preserving topological consistency. Extensive experiments on simulated and real clinical LF datasets demonstrate that DACT achieves state-of-the-art performance, yielding reconstructions with superior structural detail and correct tissue contrast.", "date": "2026-03-02", "categories": []}
{"id": "2406.01003", "title": "Uni-ISP: Toward Unifying the Learning of ISPs from Multiple Mobile Cameras", "abstract": "Modern end-to-end image signal processors (ISPs) can learn complex mappings from RAW/XYZ data to sRGB (and vice versa), opening new possibilities in image processing. However, the growing diversity of camera models, particularly in mobile devices, renders the development of individual ISPs unsustainable due to their limited versatility and adaptability across varied camera systems. In this paper, we introduce Uni-ISP, a novel pipeline that unifies ISP learning for diverse mobile cameras, delivering a highly accurate and adaptable processor. The core of Uni-ISP is leveraging device-aware embeddings through learning forward/inverse ISPs and its special training scheme. By doing so, Uni-ISP not only improves the performance of forward and inverse ISPs but also unlocks new applications previously inaccessible to conventional learned ISPs. To support this work, we construct a real-world 4K dataset, FiveCam, comprising more than 2,400 pairs of sRGB-RAW images captured synchronously by five smartphone cameras. Extensive experiments validate Uni-ISP's accuracy in learning forward and inverse ISPs (with improvements of +2.4dB/1.5dB PSNR), versatility in enabling new applications, and adaptability to new camera models.", "date": "2024-06-03", "categories": []}
{"id": "2601.23037", "title": "Scale Equivariance Regularization and Feature Lifting in High Dynamic Range Modulo Imaging", "abstract": "Modulo imaging enables high dynamic range (HDR) acquisition by cyclically wrapping saturated intensities, but accurate reconstruction remains challenging due to ambiguities between natural image edges and artificial wrap discontinuities. This work proposes a learning-based HDR restoration framework that incorporates two key strategies: (i) a scale-equivariant regularization that enforces consistency under exposure variations, and (ii) a feature lifting input design combining the raw modulo image, wrapped finite differences, and a closed-form initialization. Together, these components enhance the network's ability to distinguish true structure from wrapping artifacts, yielding state-of-the-art performance across perceptual and linear HDR quality metrics.", "date": "2026-01-30", "categories": []}
{"id": "2602.23375", "title": "Analytical Expression for Spherically Symmetric Photoacoustic Sources: A Unified General Solution (Theoretical Analysis and Derivation)", "abstract": "Here we present a comprehensive derivation of the analytical expression for the spatiotemporal acoustic pressure generated by photoacoustic sources with spherically symmetric initial pressure distributions. Starting from the fundamental photoacoustic wave equation, we derive a unified analytical solution applicable to arbitrary spherically symmetric initial distributions. Specific expressions are provided for several common distributions including uniform spherical sources, Gaussian distributions, exponential distributions, and power-law distributions. Far-field approximations are also discussed. The derived expressions provide valuable tools for photoacoustic imaging system design and signal analysis. We provide codes for ultrafast forward simulation using the general analytical spherically symmetric model, the implementation is available in the GitHub repository: \\href{https://github.com/JaegerCQ/SlingBAG_Ultra}.", "date": "2026-01-12", "categories": []}
{"id": "2602.23847", "title": "Polarization Uncertainty-Guided Diffusion Model for Color Polarization Image Demosaicking", "abstract": "Color polarization demosaicking (CPDM) aims to reconstruct full-resolution polarization images of four directions from the color-polarization filter array (CPFA) raw image. Due to the challenge of predicting numerous missing pixels and the scarcity of high-quality training data, existing network-based methods, despite effectively recovering scene intensity information, still exhibit significant errors in reconstructing polarization characteristics (degree of polarization, DOP, and angle of polarization, AOP). To address this problem, we introduce the image diffusion prior from text-to-image (T2I) models to overcome the performance bottleneck of network-based methods, with the additional diffusion prior compensating for limited representational capacity caused by restricted data distribution. To effectively leverage the diffusion prior, we explicitly model the polarization uncertainty during reconstruction and use uncertainty to guide the diffusion model in recovering high error regions. Extensive experiments demonstrate that the proposed method accurately recovers scene polarization characteristics with both high fidelity and strong visual perception.", "date": "2026-02-27", "categories": []}
{"id": "2602.23890", "title": "DACESR: Degradation-Aware Conditional Embedding for Real-World Image Super-Resolution", "abstract": "Multimodal large models have shown excellent ability in addressing image super-resolution in real-world scenarios by leveraging language class as condition information, yet their abilities in degraded images remain limited. In this paper, we first revisit the capabilities of the Recognize Anything Model (RAM) for degraded images by calculating text similarity. We find that directly using contrastive learning to fine-tune RAM in the degraded space is difficult to achieve acceptable results. To address this issue, we employ a degradation selection strategy to propose a Real Embedding Extractor (REE), which achieves significant recognition performance gain on degraded image content through contrastive learning. Furthermore, we use a Conditional Feature Modulator (CFM) to incorporate the high-level information of REE for a powerful Mamba-based network, which can leverage effective pixel information to restore image textures and produce visually pleasing results. Extensive experiments demonstrate that the REE can effectively help image super-resolution networks balance fidelity and perceptual quality, highlighting the great potential of Mamba in real-world applications. The source code of this work will be made publicly available at: https://github.com/nathan66666/DACESR.git", "date": "2026-02-27", "categories": []}
{"id": "2602.18509", "title": "Depth from Defocus via Direct Optimization", "abstract": "Though there exists a reasonable forward model for blur based on optical physics, recovering depth from a collection of defocused images remains a computationally challenging optimization problem. In this paper, we show that with contemporary optimization methods and reasonable computing resources, a global optimization approach to depth from defocus is feasible. Our approach rests on alternating minimization. When holding the depth map fixed, the forward model is linear with respect to the all-in-focus image. When holding the all-in-focus image fixed, the depth at each pixel can be computed independently, enabling embarrassingly parallel computation. We show that alternating between convex optimization and parallel grid search can effectively solve the depth-from-defocus problem at higher resolutions than current deep learning methods. We demonstrate our approach on benchmark datasets with synthetic and real defocus blur and show promising results compared to prior approaches. Our code is available at github.com/hollyjackson/dfd.", "date": "2026-02-18", "categories": []}
{"id": "2602.22620", "title": "Coded-E2LF: Coded Aperture Light Field Imaging from Events", "abstract": "We propose Coded-E2LF (coded event to light field), a computational imaging method for acquiring a 4-D light field using a coded aperture and a stationary event-only camera. In a previous work, an imaging system similar to ours was adopted, but both events and intensity images were captured and used for light field reconstruction. In contrast, our method is purely event-based, which relaxes restrictions for hardware implementation. We also introduce several advancements from the previous work that enable us to theoretically support and practically improve light field reconstruction from events alone. In particular, we clarify the key role of a black pattern in aperture coding patterns. We finally implemented our method on real imaging hardware to demonstrate its effectiveness in capturing real 3-D scenes. To the best of our knowledge, we are the first to demonstrate that a 4-D light field with pixel-level accuracy can be reconstructed from events alone. Our software and supplementary video are available from our project website.", "date": "2026-02-26", "categories": []}
{"id": "2602.22800", "title": "GSTurb: Gaussian Splatting for Atmospheric Turbulence Mitigation", "abstract": "Atmospheric turbulence causes significant image degradation due to pixel displacement (tilt) and blur, particularly in long-range imaging applications. In this paper, we propose a novel framework for atmospheric turbulence mitigation, GSTurb, which integrates optical flow-guided tilt correction and Gaussian splatting for modeling non-isoplanatic blur. The framework employs Gaussian parameters to represent tilt and blur, and optimizes them across multiple frames to enhance restoration. Experimental results on the ATSyn-static dataset demonstrate the effectiveness of our method, achieving a peak PSNR of 27.67 dB and SSIM of 0.8735. Compared to the state-of-the-art method, GSTurb improves PSNR by 1.3 dB (a 4.5% increase) and SSIM by 0.048 (a 5.8% increase). Additionally, on real datasets, including the TSRWGAN Real-World and CLEAR datasets, GSTurb outperforms existing methods, showing significant improvements in both qualitative and quantitative performance. These results highlight that combining optical flow-guided tilt correction with Gaussian splatting effectively enhances image restoration under both synthetic and real-world turbulence conditions. The code for this method will be available at https://github.com/DuhlLiamz/3DGS_turbulence/tree/main.", "date": "2026-02-26", "categories": []}
{"id": "2507.06593", "title": "Capturing Stable HDR Videos Using a Dual-Camera System", "abstract": "High Dynamic Range (HDR) video acquisition using the alternating exposure (AE) paradigm has garnered significant attention due to its cost-effectiveness with a single consumer camera. However, despite progress driven by deep neural networks, these methods remain prone to temporal flicker in real-world applications due to inter-frame exposure inconsistencies. To address this challenge while maintaining the cost-effectiveness of the AE paradigm, we propose a novel learning-based HDR video generation solution. Specifically, we propose a dual-stream HDR video generation paradigm that decouples temporal luminance anchoring from exposure-variant detail reconstruction, overcoming the inherent limitations of the AE paradigm. To support this, we design an asynchronous dual-camera system (DCS), which enables independent exposure control across two cameras, eliminating the need for synchronization typically required in traditional multi-camera setups. Furthermore, an exposure-adaptive fusion network (EAFNet) is formulated for the DCS system. EAFNet integrates a pre-alignment subnetwork that aligns features across varying exposures, ensuring robust feature extraction for subsequent fusion, an asymmetric cross-feature fusion subnetwork that emphasizes reference-based attention to effectively merge these features across exposures, and a reconstruction subnetwork to mitigate ghosting artifacts and preserve fine details. Extensive experimental evaluations demonstrate that the proposed method achieves state-of-the-art performance across various datasets, showing the remarkable potential of our solution in HDR video reconstruction. The codes and data captured by DCS will be available at https://zqqqyu.github.io/DCS-HDR/.", "date": "2025-07-09", "categories": []}
{"id": "2602.22140", "title": "Lumosaic: Hyperspectral Video via Active Illumination and Coded-Exposure Pixels", "abstract": "We present Lumosaic, a compact active hyperspectral video system designed for real-time capture of dynamic scenes. Our approach combines a narrowband LED array with a coded-exposure-pixel (CEP) camera capable of high-speed, per-pixel exposure control, enabling joint encoding of scene information across space, time, and wavelength within each video frame. Unlike passive snapshot systems that divide light across multiple spectral channels simultaneously and assume no motion during a frame's exposure, Lumosaic actively synchronizes illumination and pixel-wise exposure, improving photon utilization and preserving spectral fidelity under motion. A learning-based reconstruction pipeline then recovers 31-channel hyperspectral (400-700 nm) video at 30 fps and VGA resolution, producing temporally coherent and spectrally accurate reconstructions. Experiments on synthetic and real data demonstrate that Lumosaic significantly improves reconstruction fidelity and temporal stability over existing snapshot hyperspectral imaging systems, enabling robust hyperspectral video across diverse materials and motion conditions.", "date": "2026-02-25", "categories": []}
{"id": "2508.08114", "title": "Learned Regularization for Microwave Tomography", "abstract": "Microwave Tomography (MWT) aims to reconstruct the dielectric properties of tissues from measured scattered electromagnetic fields. This inverse problem is highly nonlinear and ill-posed, posing significant challenges for conventional optimization-based methods, which, despite being grounded in physical models, often fail to recover fine structural details. Recent deep learning strategies, including end-to-end and post-processing networks, have improved reconstruction quality but typically require large paired training datasets and may struggle to generalize. To overcome these limitations, we propose a physics-informed hybrid framework that integrates diffusion models as learned regularization within a data-consistency-driven variational scheme. Specifically, we introduce Single-Step Diffusion Regularization (SSD-Reg), a novel approach that embeds diffusion priors into the iterative reconstruction process, enabling the recovery of complex anatomical structures without the need for paired data. SSD-Reg maintains fidelity to both the governing physics and learned structural distributions, improving accuracy, stability, and robustness. Extensive experiments demonstrate that SSD-Reg, implemented as a Plug-and-Play (PnP) module, provides a flexible and effective solution for tackling the ill-posedness inherent in functional image reconstruction.", "date": "2025-08-11", "categories": []}
{"id": "2601.12149", "title": "Principal Component Analysis-Based Terahertz Self-Supervised Denoising and Deblurring Deep Neural Networks", "abstract": "Terahertz (THz) systems inherently introduce frequency-dependent degradation effects, resulting in low-frequency blurring and high-frequency noise in amplitude images. Conventional image processing techniques cannot simultaneously address both issues, and manual intervention is often required due to the unknown boundary between denoising and deblurring. To tackle this challenge, we propose a principal component analysis (PCA)-based THz self-supervised denoising and deblurring network (THz-SSDD). The network employs a Recorrupted-to-Recorrupted self-supervised learning strategy to capture the intrinsic features of noise by exploiting invariance under repeated corruption. PCA decomposition and reconstruction are then applied to restore images across both low and high frequencies. The performance of the THz-SSDD network was evaluated on four types of samples. Training requires only a small set of unlabeled noisy images, and testing across samples with different material properties and measurement modes demonstrates effective denoising and deblurring. Quantitative analysis further validates the network feasibility, showing improvements in image quality while preserving the physical characteristics of the original signals.", "date": "2026-01-17", "categories": []}
{"id": "2602.20417", "title": "gQIR: Generative Quanta Image Reconstruction", "abstract": "Capturing high-quality images from only a few detected photons is a fundamental challenge in computational imaging. Single-photon avalanche diode (SPAD) sensors promise high-quality imaging in regimes where conventional cameras fail, but raw \\emph{quanta frames} contain only sparse, noisy, binary photon detections. Recovering a coherent image from a burst of such frames requires handling alignment, denoising, and demosaicing (for color) under noise statistics far outside those assumed by standard restoration pipelines or modern generative models. We present an approach that adapts large text-to-image latent diffusion models to the photon-limited domain of quanta burst imaging. Our method leverages the structural and semantic priors of internet-scale diffusion models while introducing mechanisms to handle Bernoulli photon statistics. By integrating latent-space restoration with burst-level spatio-temporal reasoning, our approach produces reconstructions that are both photometrically faithful and perceptually pleasing, even under high-speed motion. We evaluate the method on synthetic benchmarks and new real-world datasets, including the first color SPAD burst dataset and a challenging \\textit{Deforming (XD)} video benchmark. Across all settings, the approach substantially improves perceptual quality over classical and modern learning-based baselines, demonstrating the promise of adapting large generative priors to extreme photon-limited sensing. Code at \\href{https://github.com/Aryan-Garg/gQIR}{https://github.com/Aryan-Garg/gQIR}.", "date": "2026-02-23", "categories": []}
{"id": "2602.20550", "title": "The Finite Primitive Basis Theorem for Computational Imaging: Formal Foundations of the OperatorGraph Representation", "abstract": "Computational imaging forward models, from coded aperture spectral cameras to MRI scanners, are traditionally implemented as monolithic, modality-specific codes. We prove that every forward model in a broad, precisely defined operator class Cimg (encompassing clinical, scientific, and industrial imaging modalities, both linear and nonlinear) admits an epsilon-approximate representation as a typed directed acyclic graph (DAG) whose nodes are drawn from a library of exactly 11 canonical primitives: Propagate, Modulate, Project, Encode, Convolve, Accumulate, Detect, Sample, Disperse, Scatter, and Transform. We call this the Finite Primitive Basis Theorem. The proof is constructive: we provide an algorithm that, given any H in Cimg, produces a DAG G with relative operator error at most epsilon and graph complexity within prescribed bounds. We further prove that the library is minimal: removing any single primitive causes at least one modality to lose its epsilon-approximate representation. A systematic analysis of nonlinearities in imaging physics shows they fall into two structural categories: pointwise scalar functions (handled by Transform) and self-consistent iterations (unrolled into existing linear primitives). Empirical validation on 31 linear modalities confirms eimg below 0.01 with at most 5 nodes and depth 5, and we provide constructive DAG decompositions for 9 additional nonlinear modalities. These results establish mathematical foundations for the Physics World Model (PWM) framework.", "date": "2026-02-24", "categories": []}
{"id": "2602.18697", "title": "Deep LoRA-Unfolding Networks for Image Restoration", "abstract": "Deep unfolding networks (DUNs), combining conventional iterative optimization algorithms and deep neural networks into a multi-stage framework, have achieved remarkable accomplishments in Image Restoration (IR), such as spectral imaging reconstruction, compressive sensing and super-resolution.It unfolds the iterative optimization steps into a stack of sequentially linked blocks.Each block consists of a Gradient Descent Module (GDM) and a Proximal Mapping Module (PMM) which is equivalent to a denoiser from a Bayesian perspective, operating on Gaussian noise with a known level.However, existing DUNs suffer from two critical limitations: (i) their PMMs share identical architectures and denoising objectives across stages, ignoring the need for stage-specific adaptation to varying noise levels; and (ii) their chain of structurally repetitive blocks results in severe parameter redundancy and high memory consumption, hindering deployment in large-scale or resource-constrained scenarios.To address these challenges, we introduce generalized Deep Low-rank Adaptation (LoRA) Unfolding Networks for image restoration, named LoRun, harmonizing denoising objectives and adapting different denoising levels between stages with compressed memory usage for more efficient DUN.LoRun introduces a novel paradigm where a single pretrained base denoiser is shared across all stages, while lightweight, stage-specific LoRA adapters are injected into the PMMs to dynamically modulate denoising behavior according to the noise level at each unfolding step.This design decouples the core restoration capability from task-specific adaptation, enabling precise control over denoising intensity without duplicating full network parameters and achieving up to $N$ times parameter reduction for an $N$-stage DUN with on-par or better performance.Extensive experiments conducted on three IR tasks validate the efficiency of our method.", "date": "2026-02-21", "categories": []}
{"id": "2406.04388", "title": "Single Exposure Quantitative Phase Imaging with a Conventional Microscope using Diffusion Models", "abstract": "Phase imaging is gaining importance due to its applications in fields like biomedical imaging and material characterization. In biomedical applications, it can provide quantitative information missing in label-free microscopy modalities. One of the most prominent methods in phase quantification is the Transport-of-Intensity Equation (TIE). TIE often requires multiple acquisitions at different defocus distances, which is not always feasible in a clinical setting. To address this issue, we propose to use chromatic aberrations to induce the required through-focus images with a single exposure, effectively generating a through-focus stack. Since the defocus distance induced by the aberrations is small, conventional TIE solvers are insufficient to address the resulting artifacts. We propose Zero-Mean Diffusion, a modified version of diffusion models designed for quantitative image prediction, and train it with synthetic data to ensure robust phase retrieval. Our contributions offer an alternative TIE approach that leverages chromatic aberrations, achieving accurate single-exposure phase measurement with white light and thus improving the efficiency of phase imaging. Moreover, we present a new class of diffusion models that are well-suited for quantitative data and have a sound theoretical basis. To validate our approach, we employ a widespread brightfield microscope equipped with a commercially available color camera. We apply our model to clinical microscopy of patients' urine, obtaining accurate phase measurements.", "date": "2024-06-06", "categories": []}
{"id": "2602.16950", "title": "HS-3D-NeRF: 3D Surface and Hyperspectral Reconstruction From Stationary Hyperspectral Images Using Multi-Channel NeRFs", "abstract": "Advances in hyperspectral imaging (HSI) and 3D reconstruction have enabled accurate, high-throughput characterization of agricultural produce quality and plant phenotypes, both essential for advancing agricultural sustainability and breeding programs. HSI captures detailed biochemical features of produce, while 3D geometric data substantially improves morphological analysis. However, integrating these two modalities at scale remains challenging, as conventional approaches involve complex hardware setups incompatible with automated phenotyping systems. Recent advances in neural radiance fields (NeRF) offer computationally efficient 3D reconstruction but typically require moving-camera setups, limiting throughput and reproducibility in standard indoor agricultural environments. To address these challenges, we introduce HSI-SC-NeRF, a stationary-camera multi-channel NeRF framework for high-throughput hyperspectral 3D reconstruction targeting postharvest inspection of agricultural produce. Multi-view hyperspectral data is captured using a stationary camera while the object rotates within a custom-built Teflon imaging chamber providing diffuse, uniform illumination. Object poses are estimated via ArUco calibration markers and transformed to the camera frame of reference through simulated pose transformations, enabling standard NeRF training on stationary-camera data. A multi-channel NeRF formulation optimizes reconstruction across all hyperspectral bands jointly using a composite spectral loss, supported by a two-stage training protocol that decouples geometric initialization from radiometric refinement. Experiments on three agricultural produce samples demonstrate high spatial reconstruction accuracy and strong spectral fidelity across the visible and near-infrared spectrum, confirming the suitability of HSI-SC-NeRF for integration into automated agricultural workflows.", "date": "2026-02-18", "categories": []}
{"id": "2602.17353", "title": "Application and Evaluation of the Common Circles Method", "abstract": "We investigate the application of the common circle method for estimating sample motion in optical diffraction tomography (ODT) of sub-millimeter sized biological tissue. When samples are confined via contact-free acoustical force fields, their motion must be estimated from the captured images. The common circle method identifies intersections of Ewald spheres in Fourier space to determine rotational motion. This paper presents a practical implementation, incorporating temporal consistency constraints to achieve stable reconstructions. Our results on both simulated and real-world data demonstrate that the common circle method provides a computationally efficient alternative to full optimization methods for motion detection.", "date": "2026-02-19", "categories": []}
{"id": "2504.19223", "title": "CARL: Camera-Agnostic Representation Learning for Spectral Image Analysis", "abstract": "Spectral imaging offers promising applications across diverse domains, including medicine and urban scene understanding, and is already established as a critical modality in remote sensing. However, variability in channel dimensionality and captured wavelengths among spectral cameras impede the development of AI-driven methodologies, leading to camera-specific models with limited generalizability and inadequate cross-camera applicability. To address this bottleneck, we introduce CARL, a model for Camera-Agnostic Representation Learning across RGB, multispectral, and hyperspectral imaging modalities. To enable the conversion of a spectral image with any channel dimensionality to a camera-agnostic representation, we introduce a novel spectral encoder, featuring a self-attention-cross-attention mechanism, to distill salient spectral information into learned spectral representations. Spatio-spectral pre-training is achieved with a novel feature-based self-supervision strategy tailored to CARL. Large-scale experiments across the domains of medical imaging, autonomous driving, and satellite imaging demonstrate our model's unique robustness to spectral heterogeneity, outperforming on datasets with simulated and real-world cross-camera spectral variations. The scalability and versatility of the proposed approach position our model as a backbone for future spectral foundation models. Code and model weights are publicly available at https://github.com/IMSY-DKFZ/CARL.", "date": "2025-04-27", "categories": []}
{"id": "2511.14147", "title": "Imaging with super-resolution in changing random media", "abstract": "We develop an imaging algorithm that exploits strong scattering to achieve super-resolution in changing random media. The method processes large and diverse array datasets using sparse dictionary learning, clustering, and multidimensional scaling. Starting from random initializations, the algorithm reliably extracts the unknown medium properties necessary for accurate imaging using back-propagation, $\\ell_2$ or $\\ell_1$ methods. Remarkably, scattering enhances resolution beyond homogeneous medium limits. When abundant data are available, the algorithm allows the realization of super-resolution in imaging.", "date": "2025-11-18", "categories": []}
{"id": "2602.16357", "title": "Optical Inversion and Spectral Unmixing of Spectroscopic Photoacoustic Images with Physics-Informed Neural Networks", "abstract": "Accurate estimation of the relative concentrations of chromophores in a spectroscopic photoacoustic (sPA) image can reveal immense structural, functional, and molecular information about physiological processes. However, due to nonlinearities and ill-posedness inherent to sPA imaging, concentration estimation is intractable. The Spectroscopic Photoacoustic Optical Inversion Autoencoder (SPOI-AE) aims to address the sPA optical inversion and spectral unmixing problems without assuming linearity. Herein, SPOI-AE was trained and tested on \\textit{in vivo} mouse lymph node sPA images with unknown ground truth chromophore concentrations. SPOI-AE better reconstructs input sPA pixels than conventional algorithms while providing biologically coherent estimates for optical parameters, chromophore concentrations, and the percent oxygen saturation of tissue. SPOI-AE's unmixing accuracy was validated using a simulated mouse lymph node phantom ground truth.", "date": "2026-02-18", "categories": []}
{"id": "2306.17652", "title": "Accurate 2D Reconstruction for PET Scanners based on the Analytical White Image Model", "abstract": "In this paper, we provide a precise mathematical model of crystal-to-crystal response which is used to generate the white image - a necessary compensation model needed to overcome the physical limitations of the PET scanner. We present a closed-form solution, as well as several accurate approximations, due to the complexity of the exact mathematical expressions. We prove, experimentally and analytically, that the difference between the best approximations and real crystal-to-crystal response is insignificant. The obtained responses are used to generate the white image compensation model. It can be written as a single closed-form expression making it easy to implement in known reconstruction methods. The maximum likelihood expectation maximization (MLEM) algorithm is modified and our white image model is integrated into it. The modified MLEM algorithm is not based on the system matrix, rather it is based on ray-driven projections and back-projections. The compensation model provides all necessary information about the system. Finally, we check our approach on synthetic and real data. For the real-world acquisition, we use the Raytest ClearPET camera for small animals and the NEMA NU 4-2008 phantom. The proposed approach overperforms competitive, non-compensated reconstruction methods.", "date": "2023-06-30", "categories": []}
{"id": "2507.11035", "title": "Efficient Dual-domain Image Dehazing with Haze Prior Perception", "abstract": "Transformers offer strong global modeling for single-image dehazing but come with high computational costs. Most methods rely on spatial features to capture long-range dependencies, making them less effective under complex haze conditions. Although some integrate frequency-domain cues, weak coupling between spatial and frequency branches limits their performance. To address these issues, we propose the Dark Channel Guided Frequency-aware Dehazing Network (DGFDNet), a dual-domain framework that explicitly aligns degradation across spatial and frequency domains. At its core, the DGFDBlock consists of two key modules: 1) Haze-Aware Frequency Modulator (HAFM), which uses dark channel priors to generate a haze confidence map for adaptive frequency modulation, achieving global degradation-aware spectral filtering. 2) Multi-level Gating Aggregation Module (MGAM), which fuses multi-scale features via multi-scale convolutions and a hybrid gating mechanism to recover fine-grained structures. Additionally, the Prior Correction Guidance Branch (PCGB) incorporates feedback for iterative refinement of the prior, improving haze localization accuracy, particularly in outdoor scenes. Extensive experiments on four benchmark datasets demonstrate that DGFDNet achieves state-of-the-art performance with improved robustness and real-time efficiency. Code is available at: https://github.com/Dilizlr/DGFDNet.", "date": "2025-07-15", "categories": []}
{"id": "2512.08161", "title": "Fourier-RWKV: A Multi-State Perception Network for Efficient Image Dehazing", "abstract": "Image dehazing is crucial for reliable visual perception, yet it remains highly challenging under real-world non-uniform haze conditions. Although Transformer-based methods excel at capturing global context, their quadratic computational complexity hinders real-time deployment. To address this, we propose Fourier Receptance Weighted Key Value (Fourier-RWKV), a novel dehazing framework based on a Multi-State Perception paradigm. The model achieves comprehensive haze degradation modeling with linear complexity by synergistically integrating three distinct perceptual states: (1) Spatial-form Perception, realized through the Deformable Quad-directional Token Shift (DQ-Shift) operation, which dynamically adjusts receptive fields to accommodate local haze variations; (2) Frequency-domain Perception, implemented within the Fourier Mix block, which extends the core WKV attention mechanism of RWKV from the spatial domain to the Fourier domain, preserving the long-range dependencies essential for global haze estimation while mitigating spatial attenuation; (3) Semantic-relation Perception, facilitated by the Semantic Bridge Module (SBM), which utilizes Dynamic Semantic Kernel Fusion (DSK-Fusion) to precisely align encoder-decoder features and suppress artifacts. Extensive experiments on multiple benchmarks demonstrate that Fourier-RWKV delivers state-of-the-art performance across diverse haze scenarios while significantly reducing computational overhead, establishing a favorable trade-off between restoration quality and practical efficiency. Code is available at: https://github.com/Dilizlr/Fourier-RWKV.", "date": "2025-12-09", "categories": []}
{"id": "2506.05325", "title": "Quasiparticle Interference Kernel Extraction with Variational Autoencoders via Latent Alignment", "abstract": "Quasiparticle interference (QPI) imaging is a powerful tool for probing electronic structures in quantum materials, but extracting the single-scatterer QPI pattern (i.e., the kernel) from a multi-scatterer image remains a fundamentally ill-posed inverse problem, because many different kernels can combine to produce almost the same observed image, and noise or overlaps further obscure the true signal. Existing solutions to this extraction problem rely on manually zooming into small local regions with isolated single-scatterers. This is infeasible for real cases where scattering conditions are too complex. In this work, we propose the first AI-based framework for QPI kernel extraction, which models the space of physically valid kernels and uses this knowledge to guide the inverse mapping. We introduce a two-step learning strategy that decouples kernel representation learning from observation-to-kernel inference. In the first step, we train a variational autoencoder to learn a compact latent space of scattering kernels. In the second step, we align the latent representation of QPI observations with those of the pre-learned kernels using a dedicated encoder. This design enables the model to infer kernels robustly under complex, entangled scattering conditions. We construct a diverse and physically realistic QPI dataset comprising 100 unique kernels and evaluate our method against a direct one-step baseline. Experimental results demonstrate that our approach achieves significantly higher extraction accuracy, improved generalization to unseen kernels. To further validate its effectiveness, we also apply the method to real QPI data from Ag and FeSe samples, where it reliably extracts meaningful kernels under complex scattering conditions.", "date": "2025-06-05", "categories": []}
{"id": "2602.12044", "title": "A DMD-Based Adaptive Modulation Method for High Dynamic Range Imaging in High-Glare Environments", "abstract": "Background The accuracy of photomechanics measurements critically relies on image quality,particularly under extreme illumination conditions such as welding arc monitoring and polished metallic surface analysis. High dynamic range (HDR) imaging above 120 dB is essential in these contexts. Conventional CCD/CMOS sensors, with dynamic ranges typically below 70 dB, are highly susceptible to saturation under glare, resulting in irreversible loss of detail and significant errors in digital image correlation (DIC). Methods This paper presents an HDR imaging system that leverages the spatial modulation capability of a digital micromirror device (DMD). The system architecture enables autonomous regional segmentation and adaptive exposure control for high-dynamic-range scenes through an integrated framework comprising two synergistic subsystems: a DMD-based optical modulation unit and an adaptive computational imaging pipeline. Results The system achieves a measurable dynamic range of 127 dB, effectively eliminating satu ration artifacts under high glare. Experimental results demonstrate a 78% reduction in strain error and improved DIC positioning accuracy, confirming reliable performance across extreme intensity variations. Conclusion The DMD-based system provides high fidelity adaptive HDR imaging, overcoming key limitations of conventional sensors. It exhibits strong potential for optical metrology and stress analysis in high-glare environments where traditional methods are inadequate.", "date": "2026-02-12", "categories": []}
{"id": "2602.11625", "title": "PLOT-CT: Pre-log Voronoi Decomposition Assisted Generation for Low-dose CT Reconstruction", "abstract": "Low-dose computed tomography (LDCT) reconstruction is fundamentally challenged by severe noise and compromised data fidelity under reduced radiation exposure. Most existing methods operate either in the image or post-log projection domain, which fails to fully exploit the rich structural information in pre-log measurements while being highly susceptible to noise. The requisite logarithmic transformation critically amplifies noise within these data, imposing exceptional demands on reconstruction precision. To overcome these challenges, we propose PLOT-CT, a novel framework for Pre-Log vOronoi decomposiTion-assisted CT generation. Our method begins by applying Voronoi decomposition to pre-log sinograms, disentangling the data into distinct underlying components, which are embedded in separate latent spaces. This explicit decomposition significantly enhances the model's capacity to learn discriminative features, directly improving reconstruction accuracy by mitigating noise and preserving information inherent in the pre-log domain. Extensive experiments demonstrate that PLOT-CT achieves state-of-the-art performance, attaining a 2.36dB PSNR improvement over traditional methods at the 1e4 incident photon level in the pre-log domain.", "date": "2026-02-12", "categories": []}
{"id": "2602.11704", "title": "U-DAVI: Uncertainty-Aware Diffusion-Prior-Based Amortized Variational Inference for Image Reconstruction", "abstract": "Ill-posed imaging inverse problems remain challenging due to the ambiguity in mapping degraded observations to clean images. Diffusion-based generative priors have recently shown promise, but typically rely on computationally intensive iterative sampling or per-instance optimization. Amortized variational inference frameworks address this inefficiency by learning a direct mapping from measurements to posteriors, enabling fast posterior sampling without requiring the optimization of a new posterior for every new set of measurements. However, they still struggle to reconstruct fine details and complex textures. To address this, we extend the amortized framework by injecting spatially adaptive perturbations to measurements during training, guided by uncertainty estimates, to emphasize learning in the most uncertain regions. Experiments on deblurring and super-resolution demonstrate that our method achieves superior or competitive performance to previous diffusion-based approaches, delivering more realistic reconstructions without the computational cost of iterative refinement.", "date": "2026-02-12", "categories": []}
{"id": "2602.11814", "title": "A Comparative Study of MAP and LMMSE Estimators for Blind Inverse Problems", "abstract": "Maximum-a-posteriori (MAP) approaches are an effective framework for inverse problems with known forward operators, particularly when combined with expressive priors and careful parameter selection. In blind settings, however, their use becomes significantly less stable due to the inherent non-convexity of the problem and the potential non-identifiability of the solutions. (Linear) minimum mean square error (MMSE) estimators provide a compelling alternative that can circumvent these limitations. In this work, we study synthetic two-dimensional blind deconvolution problems under fully controlled conditions, with complete prior knowledge of both the signal and kernel distributions. We compare tailored MAP algorithms with simple LMMSE estimators whose functional form is closely related to that of an optimal Tikhonov estimator. Our results show that, even in these highly controlled settings, MAP methods remain unstable and require extensive parameter tuning, whereas the LMMSE estimator yields a robust and reliable baseline. Moreover, we demonstrate empirically that the LMMSE solution can serve as an effective initialization for MAP approaches, improving their performance and reducing sensitivity to regularization parameters, thereby opening the door to future theoretical and practical developments.", "date": "2026-02-12", "categories": []}
{"id": "2602.11880", "title": "SynthRAR: Ring Artifacts Reduction in CT with Unrolled Network and Synthetic Data Training", "abstract": "Defective and inconsistent responses in CT detectors can cause ring and streak artifacts in the reconstructed images, making them unusable for clinical purposes. In recent years, several ring artifact reduction solutions have been proposed in the image domain or in the sinogram domain using supervised deep learning methods. However, these methods require dedicated datasets for training, leading to a high data collection cost. Furthermore, existing approaches focus exclusively on either image-space or sinogram-space correction, neglecting the intrinsic correlations from the forward operation of the CT geometry. Based on the theoretical analysis of non-ideal CT detector responses, the RAR problem is reformulated as an inverse problem by using an unrolled network, which considers non-ideal response together with linear forward-projection with CT geometry. Additionally, the intrinsic correlations of ring artifacts between the sinogram and image domains are leveraged through synthetic data derived from natural images, enabling the trained model to correct artifacts without requiring real-world clinical data. Extensive evaluations on diverse scanning geometries and anatomical regions demonstrate that the model trained on synthetic data consistently outperforms existing state-of-the-art methods.", "date": "2026-02-12", "categories": []}
{"id": "2503.19925", "title": "Accurate, provable and fast polychromatic tomographic reconstruction: A variational inequality approach", "abstract": "We consider the problem of signal reconstruction for computed tomography (CT) under a nonlinear forward model that accounts for exponential signal attenuation, a polychromatic X-ray source, general measurement noise (e.g., Poisson shot noise), and observations acquired over multiple wavelength windows. We develop a simple iterative algorithm for single-material reconstruction, which we call EXACT (EXtragradient Algorithm for Computed Tomography), based on formulating our estimate as the fixed point of a monotone variational inequality. We prove guarantees on the statistical and computational performance of EXACT under realistic assumptions on the measurement process. We also consider a recently introduced variant of this model with Gaussian measurements and present sample and iteration complexity bounds for EXACT that improve upon those of existing algorithms. We apply our EXACT algorithm to a CT phantom image recovery task and show that it often requires fewer X-ray views, lower source intensity, and less computation time to achieve reconstruction quality similar to existing methods. Code is available at https://github.com/voilalab/exact.", "date": "2025-03-13", "categories": []}
{"id": "2503.20184", "title": "Spectrum from Defocus: Fast Spectral Imaging with Chromatic Focal Stack", "abstract": "Hyperspectral cameras face harsh trade-offs between spatial, spectral, and temporal resolution in inherently low-photon conditions. Computational imaging systems break through these trade-offs with compressive sensing, but have required complex optics and/or extensive compute. We present Spectrum from Defocus (SfD), a chromatic focal sweep method that achieves state-of-the-art hyperspectral imaging with only two off-the-shelf lenses, a grayscale sensor, and less than one second of reconstruction time. By capturing a chromatically-aberrated focal stack that preserves nearly all incident light, and reconstructing it with a fast physics-based iterative algorithm, SfD delivers sharp, accurate hyperspectral images. The combination of photon efficiency, optical simplicity, and physical interpretability makes SfD a promising solution for fast, compact, interpretable hyperspectral imaging.", "date": "2025-03-26", "categories": []}
{"id": "2510.15198", "title": "HyperAIRI: a plug-and-play algorithm for precise hyperspectral image reconstruction in radio interferometry", "abstract": "The next-generation radio-interferometric (RI) telescopes require imaging algorithms capable of forming high-resolution high-dynamic-range images from large data volumes spanning wide frequency bands. Recently, AIRI, a plug-and-play (PnP) approach taking the forward-backward algorithmic structure (FB), has demonstrated state-of-the-art performance in monochromatic RI imaging by alternating a data-fidelity step with a regularization step via learned denoisers. In this work, we introduce HyperAIRI, its hyperspectral extension, underpinned by learned hyperspectral denoisers enforcing a power-law spectral model. For each spectral channel, the HyperAIRI denoiser takes as input its current image estimate, alongside estimates of its two immediate neighboring channels and the spectral index map, and provides as output its associated denoised image. To ensure convergence of HyperAIRI, the denoisers are trained with a Jacobian regularization enforcing non-expansiveness. To accommodate varying dynamic ranges, we assemble a shelf of pre-trained denoisers, each tailored to a specific dynamic range. At each HyperAIRI iteration, the spectral channels of the target image cube are updated in parallel using dynamic-range-matched denoisers from the pre-trained shelf. The denoisers are also endowed with a spatial image faceting functionality, enabling scalability to varied image sizes. Additionally, we formally introduce Hyper-uSARA, a variant of the optimization-based algorithm HyperSARA, promoting joint sparsity across spectral channels via the $\\ell_{2,1}$-norm, also adopting FB. We evaluate HyperAIRI's performance on simulated and real observations. We showcase its superior performance compared to its optimization-based counterpart Hyper-uSARA, CLEAN's hyperspectral variant in WSClean, and the monochromatic imaging algorithms AIRI and uSARA.", "date": "2025-10-16", "categories": []}
{"id": "2601.18245", "title": "Tractable Gaussian Phase Retrieval with Heavy Tails and Adversarial Corruption with Near-Linear Sample Complexity", "abstract": "Phase retrieval is the classical problem of recovering a signal $x^* \\in \\mathbb{R}^n$ from its noisy phaseless measurements $y_i = \\langle a_i, x^* \\rangle^2 + \\zeta_i$ (where $\\zeta_i$ denotes noise, and $a_i$ is the sensing vector) for $i \\in [m]$. The problem of phase retrieval has a rich history, with a variety of applications such as optics, crystallography, heteroscedastic regression, astrophysics, etc. A major consideration in algorithms for phase retrieval is robustness against measurement errors. In recent breakthroughs in algorithmic robust statistics, efficient algorithms have been developed for several parameter estimation tasks such as mean estimation, covariance estimation, robust principal component analysis (PCA), etc. in the presence of heavy-tailed noise and adversarial corruptions. In this paper, we study efficient algorithms for robust phase retrieval with heavy-tailed noise when a constant fraction of both the measurements $y_i$ and the sensing vectors $a_i$ may be arbitrarily adversarially corrupted. For this problem, Buna and Rebeschini (AISTATS 2025) very recently gave an exponential time algorithm with sample complexity $O(n \\log n)$. Their algorithm needs a robust spectral initialization, specifically, a robust estimate of the top eigenvector of a covariance matrix, which they deemed to be beyond known efficient algorithmic techniques (similar spectral initializations are a key ingredient of a large family of phase retrieval algorithms). In this work, we make a connection between robust spectral initialization and recent algorithmic advances in robust PCA, yielding the first polynomial-time algorithms for robust phase retrieval with both heavy-tailed noise and adversarial corruptions, in fact with near-linear (in $n$) sample complexity.", "date": "2026-01-26", "categories": []}
{"id": "2602.10344", "title": "Monte Carlo Maximum Likelihood Reconstruction for Digital Holography with Speckle", "abstract": "In coherent imaging, speckle is statistically modeled as multiplicative noise, posing a fundamental challenge for image reconstruction. While maximum likelihood estimation (MLE) provides a principled framework for speckle mitigation, its application to coherent imaging system such as digital holography with finite apertures is hindered by the prohibitive cost of high-dimensional matrix inversion, especially at high resolutions. This computational burden has prevented the use of MLE-based reconstruction with physically accurate aperture modeling. In this work, we propose a randomized linear algebra approach that enables scalable MLE optimization without explicit matrix inversions in gradient computation. By exploiting the structural properties of sensing matrix and using conjugate gradient for likelihood gradient evaluation, the proposed algorithm supports accurate aperture modeling without the simplifying assumptions commonly imposed for tractability. We term the resulting method projected gradient descent with Monte Carlo estimation (PGD-MC). The proposed PGD-MC framework (i) demonstrates robustness to diverse and physically accurate aperture models, (ii) achieves substantial improvements in reconstruction quality and computational efficiency, and (iii) scales effectively to high-resolution digital holography. Extensive experiments incorporating three representative denoisers as regularization show that PGD-MC provides a flexible and effective MLE-based reconstruction framework for digital holography with finite apertures, consistently outperforming prior Plug-and-Play model-based iterative reconstruction methods in both accuracy and speed. Our code is available at: https://github.com/Computational-Imaging-RU/MC_Maximum_Likelihood_Digital_Holography_Speckle.", "date": "2026-02-10", "categories": []}
{"id": "2602.02948", "title": "Variational Sparse Paired Autoencoders (vsPAIR) for Inverse Problems and Uncertainty Quantification", "abstract": "Inverse problems are fundamental to many scientific and engineering disciplines; they arise when one seeks to reconstruct hidden, underlying quantities from noisy measurements. Many applications demand not just point estimates but interpretable uncertainty. Providing fast inference alongside uncertainty estimates remains challenging yet desirable in numerous applications.\n  We propose the Variational Sparse Paired Autoencoder (vsPAIR) to address this challenge. The architecture pairs a standard VAE encoding observations with a sparse VAE encoding quantities of interest, connected through a learned latent mapping. The variational structure enables uncertainty estimation, the paired architecture encourages interpretability by anchoring QoI representations to clean data, and sparse encodings provide structure by concentrating information into identifiable factors rather than diffusing across all dimensions. To validate the effectiveness of our proposed architecture, we conduct experiments on blind inpainting and computed tomography, demonstrating that vsPAIR is a capable inverse problem solver that can provide interpretable and structured uncertainty estimates.", "date": "2026-02-03", "categories": []}
{"id": "2602.09510", "title": "Robust Depth Super-Resolution via Adaptive Diffusion Sampling", "abstract": "We propose AdaDS, a generalizable framework for depth super-resolution that robustly recovers high-resolution depth maps from arbitrarily degraded low-resolution inputs. Unlike conventional approaches that directly regress depth values and often exhibit artifacts under severe or unknown degradation, AdaDS capitalizes on the contraction property of Gaussian smoothing: as noise accumulates in the forward process, distributional discrepancies between degraded inputs and their pristine high-quality counterparts diminish, ultimately converging to isotropic Gaussian prior. Leveraging this, AdaDS adaptively selects a starting timestep in the reverse diffusion trajectory based on estimated refinement uncertainty, and subsequently injects tailored noise to position the intermediate sample within the high-probability region of the target posterior distribution. This strategy ensures inherent robustness, enabling generative prior of a pre-trained diffusion model to dominate recovery even when upstream estimations are imperfect. Extensive experiments on real-world and synthetic benchmarks demonstrate AdaDS's superior zero-shot generalization and resilience to diverse degradation patterns compared to state-of-the-art methods.", "date": "2026-02-10", "categories": []}
{"id": "2409.06714", "title": "FCDM: A Physics-Guided Bidirectional Frequency Aware Convolution and Diffusion-Based Model for Sinogram Inpainting", "abstract": "Computed tomography (CT) is widely used in scientific imaging systems such as synchrotron and laboratory-based nano-CT, but acquiring full-view sinograms requires high radiation dose and long scan times. Sparse-view CT reduces this burden but produces incomplete sinograms with structured signal loss, degrading reconstruction quality. Unlike RGB images, sinograms encode globally coupled projections and exhibit directional spectral patterns, making conventional RGB-oriented inpainting methods, including diffusion models, ineffective because they ignore angular dependencies and physical constraints inherent to tomographic data. We propose FCDM, a diffusion-based framework for sinogram restoration that incorporates bidirectional frequency reasoning, angular-aware masking, and physics-guided regularization to preserve global structure and physical plausibility. Experiments on real-world datasets show that FCDM consistently outperforms existing baselines, achieving over 0.93 SSIM and 31 dB PSNR across diverse sparse-view settings.", "date": "2024-08-26", "categories": []}
{"id": "2602.07056", "title": "MTS-CSNet: Multiscale Tensor Factorization for Deep Compressive Sensing on RGB Images", "abstract": "Deep learning based compressive sensing (CS) methods typically learn sampling operators using convolutional or block wise fully connected layers, which limit receptive fields and scale poorly for high dimensional data. We propose MTSCSNet, a CS framework based on Multiscale Tensor Summation (MTS) factorization, a structured operator for efficient multidimensional signal processing. MTS performs mode-wise linear transformations with multiscale summation, enabling large receptive fields and effective modeling of cross-dimensional correlations. In MTSCSNet, MTS is first used as a learnable CS operator that performs linear dimensionality reduction in tensor space, with its adjoint defining the initial back-projection, and is then applied in the reconstruction stage to directly refine this estimate. This results in a simple feed-forward architecture without iterative or proximal optimization, while remaining parameter and computation efficient. Experiments on standard CS benchmarks show that MTSCSNet achieves state-of-the-art reconstruction performance on RGB images, with notable PSNR gains and faster inference, even compared to recent diffusion-based CS methods, while using a significantly more compact feed-forward architecture.", "date": "2026-02-04", "categories": []}
{"id": "2602.07060", "title": "U-Net Based Image Enhancement for Short-time Muon Scattering Tomography", "abstract": "Muon Scattering Tomography (MST) is a promising non-invasive inspection technique, yet the practical application of short-time MST is hindered by poor image quality due to limited muon flux. To address this limitation, we propose a U-Net-based framework trained on Point of Closest Approach (PoCA) images reconstructed with simulation MST data to enhance image quality. When applied to experimental MST data, the framework significantly improves image quality, increasing the Structural Similarity Index Measure (SSIM) from 0.7232 to 0.9699 and decreasing the Learned Perceptual Image Patch Similarity (LPIPS) from 0.3604 to 0.0270. These results demonstrate that our method can effectively enhance low-statistics MST images, thereby paving the way for the practical deployment of short-time MST.", "date": "2026-02-05", "categories": []}
{"id": "2602.07094", "title": "Exploring Polarimetric Properties Preservation during Reconstruction of PolSAR images using Complex-valued Convolutional Neural Networks", "abstract": "The inherently complex-valued nature of Polarimetric SAR data necessitates using specialized algorithms capable of directly processing complex-valued representations. However, this aspect remains underexplored in the deep learning community, with many studies opting to convert complex signals into the real domain before applying conventional real-valued models. In this work, we leverage complex-valued neural networks and investigate the performance of complex-valued Convolutional AutoEncoders. We show that these networks can effectively compress and reconstruct fully polarimetric SAR data while preserving essential physical characteristics, as demonstrated through Pauli, Krogager, and Cameron coherent decompositions, as well as the non-coherent $H-\\alpha$ decomposition. Finally, we highlight the advantages of complex-valued neural networks over their real-valued counterparts. These insights pave the way for developing robust, physics-informed, complex-valued generative models for SAR data processing.", "date": "2026-02-06", "categories": []}
{"id": "2602.07102", "title": "Fast and Robust Likelihood-Guided Diffusion Posterior Sampling with Amortized Variational Inference", "abstract": "Zero-shot diffusion posterior sampling offers a flexible framework for inverse problems by accommodating arbitrary degradation operators at test time, but incurs high computational cost due to repeated likelihood-guided updates. In contrast, previous amortized diffusion approaches enable fast inference by replacing likelihood-based sampling with implicit inference models, but at the expense of robustness to unseen degradations. We introduce an amortization strategy for diffusion posterior sampling that preserves explicit likelihood guidance by amortizing the inner optimization problems arising in variational diffusion posterior sampling. This accelerates inference for in-distribution degradations while maintaining robustness to previously unseen operators, thereby improving the trade-off between efficiency and flexibility in diffusion-based inverse problems.", "date": "2026-02-06", "categories": []}
{"id": "2602.07820", "title": "Back to Physics: Operator-Guided Generative Paths for SMS MRI Reconstruction", "abstract": "Simultaneous multi-slice (SMS) imaging with in-plane undersampling enables highly accelerated MRI but yields a strongly coupled inverse problem with deterministic inter-slice interference and missing k-space data. Most diffusion-based reconstructions are formulated around Gaussian-noise corruption and rely on additional consistency steps to incorporate SMS physics, which can be mismatched to the operator-governed degradations in SMS acquisition. We propose an operator-guided framework that models the degradation trajectory using known acquisition operators and inverts this process via deterministic updates. Within this framework, we introduce an operator-conditional dual-stream interaction network (OCDI-Net) that explicitly disentangles target-slice content from inter-slice interference and predicts structured degradations for operator-aligned inversion, and we instantiate reconstruction as a two-stage chained inference procedure that performs SMS slice separation followed by in-plane completion. Experiments on fastMRI brain data and prospectively acquired in vivo diffusion MRI data demonstrate improved fidelity and reduced slice leakage over conventional and learning-based SMS reconstructions.", "date": "2026-02-08", "categories": []}
{"id": "2602.07980", "title": "Continuity-driven Synergistic Diffusion with Neural Priors for Ultra-Sparse-View CBCT Reconstruction", "abstract": "The clinical application of cone-beam computed tomography (CBCT) is constrained by the inherent trade-off between radiation exposure and image quality. Ultra-sparse angular sampling, employed to reduce dose, introduces severe undersampling artifacts and inter-slice inconsistencies, compromising diagnostic reliability. Existing reconstruction methods often struggle to balance angular continuity with spatial detail fidelity. To address these challenges, we propose a Continuity-driven Synergistic Diffusion with Neural priors (CSDN) for ultra-sparse-view CBCT reconstruction. Neural priors are introduced as a structural foundation to encode a continuous threedimensional attenuation representation, enabling the synthesis of physically consistent dense projections from ultra-sparse measurements. Building upon this neural-prior-based initialization, a synergistic diffusion strategy is developed, consisting of two collaborative refinement paths: a Sinogram Refinement Diffusion (Sino-RD) process that restores angular continuity and a Digital Radiography Refinement Diffusion (DR-RD) process that enforces inter-slice consistency from the projection image perspective. The outputs of the two diffusion paths are adaptively fused by the Dual-Projection Reconstruction Fusion (DPRF) module to achieve coherent volumetric reconstruction. Extensive experiments demonstrate that the proposed CSDN effectively suppresses artifacts and recovers fine textures under ultra-sparse-view conditions, outperforming existing state-of-the-art techniques.", "date": "2026-02-08", "categories": []}
{"id": "2511.14469", "title": "CompEvent: Complex-valued Event-RGB Fusion for Low-light Video Enhancement and Deblurring", "abstract": "Low-light video deblurring poses significant challenges in applications like nighttime surveillance and autonomous driving due to dim lighting and long exposures. While event cameras offer potential solutions with superior low-light sensitivity and high temporal resolution, existing fusion methods typically employ staged strategies, limiting their effectiveness against combined low-light and motion blur degradations. To overcome this, we propose CompEvent, a complex neural network framework enabling holistic full-process fusion of event data and RGB frames for enhanced joint restoration. CompEvent features two core components: 1) Complex Temporal Alignment GRU, which utilizes complex-valued convolutions and processes video and event streams iteratively via GRU to achieve temporal alignment and continuous fusion; and 2) Complex Space-Frequency Learning module, which performs unified complex-valued signal processing in both spatial and frequency domains, facilitating deep fusion through spatial structures and system-level characteristics. By leveraging the holistic representation capability of complex-valued neural networks, CompEvent achieves full-process spatiotemporal fusion, maximizes complementary learning between modalities, and significantly strengthens low-light video deblurring capability. Extensive experiments demonstrate that CompEvent outperforms SOTA methods in addressing this challenging task.", "date": "2025-11-18", "categories": []}
{"id": "2512.10524", "title": "Inverse problems with diffusion models: MAP estimation via mode-seeking loss", "abstract": "A pre-trained unconditional diffusion model, combined with posterior sampling or maximum a posteriori (MAP) estimation techniques, can solve arbitrary inverse problems without task-specific training or fine-tuning. However, existing posterior sampling and MAP estimation methods often rely on modeling approximations and can also be computationally demanding. In this work, we propose a new MAP estimation strategy for solving inverse problems with a pre-trained unconditional diffusion model. Specifically, we introduce the variational mode-seeking loss (VML) and show that its minimization at each reverse diffusion step guides the generated sample towards the MAP estimate (modes in practice). VML arises from a novel perspective of minimizing the Kullback-Leibler (KL) divergence between the diffusion posterior $p(\\mathbf{x}_0|\\mathbf{x}_t)$ and the measurement posterior $p(\\mathbf{x}_0|\\mathbf{y})$, where $\\mathbf{y}$ denotes the measurement. Importantly, for linear inverse problems, VML can be analytically derived without any modeling approximations. Based on further theoretical insights, we propose VML-MAP, an empirically effective algorithm for solving inverse problems via VML minimization, and validate its efficacy in both performance and computational time through extensive experiments on diverse image-restoration tasks across multiple datasets.", "date": "2025-12-11", "categories": []}
{"id": "2504.17732", "title": "DPMambaIR: All-in-One Image Restoration via Degradation-Aware Prompt State Space Model", "abstract": "All-in-One image restoration aims to address multiple image degradation problems using a single model, offering a more practical and versatile solution compared to designing dedicated models for each degradation type. Existing approaches typically rely on Degradation-specific models or coarse-grained degradation prompts to guide image restoration. However, they lack fine-grained modeling of degradation information and face limitations in balancing multi-task conflicts. To overcome these limitations, we propose DPMambaIR, a novel All-in-One image restoration framework that introduces a fine-grained degradation extractor and a Degradation-Aware Prompt State Space Model (DP-SSM). The DP-SSM leverages the fine-grained degradation features captured by the extractor as dynamic prompts, which are then incorporated into the state space modeling process. This enhances the model's adaptability to diverse degradation types, while a complementary High-Frequency Enhancement Block (HEB) recovers local high-frequency details. Extensive experiments on a mixed dataset containing seven degradation types show that DPMambaIR achieves the best performance, with 27.69dB and 0.893 in PSNR and SSIM, respectively. These results highlight the potential and superiority of DPMambaIR as a unified solution for All-in-One image restoration.", "date": "2025-04-24", "categories": []}
{"id": "2601.12666", "title": "Near-Light Color Photometric Stereo for Mono-Chromatic Non-Lambertian Surfaces", "abstract": "Color photometric stereo enables single-shot surface reconstruction, extending conventional photometric stereo that requires multiple images of a static scene under varying illumination to dynamic scenarios. However, most existing approaches assume ideal distant lighting and Lambertian reflectance, leaving more practical near-light conditions and non-Lambertian surfaces underexplored. To overcome this limitation, we propose a framework that leverages neural implicit representations for depth and BRDF modeling under the assumption of mono-chromaticity (uniform chromaticity and homogeneous material), which alleviates the inherent ill-posedness of color photometric stereo and allows for detailed surface recovery from just one image. Furthermore, we design a compact optical tactile sensor to validate our approach. Experiments on both synthetic and real-world datasets demonstrate that our method achieves accurate and robust surface reconstruction.", "date": "2026-01-19", "categories": []}
{"id": "2602.03893", "title": "GPAIR: Gaussian-Kernel-Based Ultrafast 3D Photoacoustic Iterative Reconstruction", "abstract": "Although the iterative reconstruction (IR) algorithm can substantially correct reconstruction artifacts in photoacoustic (PA) computed tomography (PACT), it suffers from long reconstruction times, especially for large-scale three-dimensional (3D) imaging in which IR takes hundreds of seconds to hours. The computing burden severely limits the practical applicability of IR algorithms. In this work, we proposed an ultrafast IR method for 3D PACT, called Gaussian-kernel-based Ultrafast 3D Photoacoustic Iterative Reconstruction (GPAIR), which achieves orders-of-magnitude acceleration in computing. GPAIR transforms traditional spatial grids with continuous isotropic Gaussian kernels. By deriving analytical closed-form expression for pressure waves and implementing powerful GPU-accelerated differentiable Triton operators, GPAIR demonstrates extraordinary ultrafast sub-second reconstruction speed for 3D targets containing 8.4 million voxels in animal experiments. This revolutionary ultrafast image reconstruction enables near-real-time large-scale 3D PA reconstruction, significantly advancing 3D PACT toward clinical applications.", "date": "2026-02-03", "categories": []}
{"id": "2602.02713", "title": "Perfusion Imaging and Single Material Reconstruction in Polychromatic Photon Counting CT", "abstract": "Background: Perfusion computed tomography (CT) images the dynamics of a contrast agent through the body over time, and is one of the highest X-ray dose scans in medical imaging. Recently, a theoretically justified reconstruction algorithm based on a monotone variational inequality (VI) was proposed for single material polychromatic photon-counting CT, and showed promising early results at low-dose imaging.\n  Purpose: We adapt this reconstruction algorithm for perfusion CT, to reconstruct the concentration map of the contrast agent while the static background tissue is assumed known; we call our method VI-PRISM (VI-based PeRfusion Imaging and Single Material reconstruction). We evaluate its potential for dose-reduced perfusion CT, using a digital phantom with water and iodine of varying concentration.\n  Methods: Simulated iodine concentrations range from 0.05 to 2.5 mg/ml. The simulated X-ray source emits photons up to 100 keV, with average intensity ranging from $10^5$ down to $10^2$ photons per detector element. The number of tomographic projections was varied from 984 down to 8 to characterize the tradeoff in photon allocation between views and intensity.\n  Results: We compare VI-PRISM against filtered back-projection (FBP), and find that VI-PRISM recovers iodine concentration with error below 0.4 mg/ml at all source intensity levels tested. Even with a dose reduction between 10x and 100x compared to FBP, VI-PRISM exhibits reconstruction quality on par with FBP.\n  Conclusion: Across all photon budgets and angular sampling densities tested, VI-PRISM achieved consistently lower RMSE, reduced noise, and higher SNR compared to filtered back-projection. Even in extremely photon-limited and sparsely sampled regimes, VI-PRISM recovered iodine concentrations with errors below 0.4 mg/ml, showing that VI-PRISM can support accurate and dose-efficient perfusion imaging in photon-counting CT.", "date": "2026-02-02", "categories": []}
{"id": "2602.02721", "title": "End-to-end reconstruction of OCT optical properties and speckle-reduced structural intensity via physics-based learning", "abstract": "Inverse scattering in optical coherence tomography (OCT) seeks to recover both structural images and intrinsic tissue optical properties, including refractive index, scattering coefficient, and anisotropy. This inverse problem is challenging due to attenuation, speckle noise, and strong coupling among parameters. We propose a regularized end-to-end deep learning framework that jointly reconstructs optical parameter maps and speckle-reduced OCT structural intensity for layer visualization. Trained with Monte Carlo-simulated ground truth, our network incorporates a physics-based OCT forward model that generates predicted signals from the estimated parameters, providing physics-consistent supervision for parameter recovery and artifact suppression. Experiments on the synthetic corneal OCT dataset demonstrate robust optical map recovery under noise, improved resolution, and enhanced structural fidelity. This approach enables quantitative multi-parameter tissue characterization and highlights the benefit of combining physics-informed modeling with deep learning for computational OCT.", "date": "2026-02-02", "categories": []}
{"id": "2602.03317", "title": "Multiparameter Uncertainty Mapping in Quantitative Molecular MRI using a Physics-Structured Variational Autoencoder (PS-VAE)", "abstract": "Quantitative imaging methods, such as magnetic resonance fingerprinting (MRF), aim to extract interpretable pathology biomarkers by estimating biophysical tissue parameters from signal evolutions. However, the pattern-matching algorithms or neural networks used in such inverse problems often lack principled uncertainty quantification, which limits the trustworthiness and transparency, required for clinical acceptance. Here, we describe a physics-structured variational autoencoder (PS-VAE) designed for rapid extraction of voxelwise multi-parameter posterior distributions. Our approach integrates a differentiable spin physics simulator with self-supervised learning, and provides a full covariance that captures the inter-parameter correlations of the latent biophysical space. The method was validated in a multi-proton pool chemical exchange saturation transfer (CEST) and semisolid magnetization transfer (MT) molecular MRF study, across in-vitro phantoms, tumor-bearing mice, healthy human volunteers, and a subject with glioblastoma. The resulting multi-parametric posteriors are in good agreement with those calculated using a brute-force Bayesian analysis, while providing an orders-of-magnitude acceleration in whole brain quantification. In addition, we demonstrate how monitoring the multi-parameter posterior dynamics across progressively acquired signals provides practical insights for protocol optimization and may facilitate real-time adaptive acquisition.", "date": "2026-02-03", "categories": []}
{"id": "2602.03449", "title": "Score-based diffusion models for diffuse optical tomography with uncertainty quantification", "abstract": "Score-based diffusion models are a recently developed framework for posterior sampling in Bayesian inverse problems with a state-of-the-art performance for severely ill-posed problems by leveraging a powerful prior distribution learned from empirical data. Despite generating significant interest especially in the machine-learning community, a thorough study of realistic inverse problems in the presence of modelling error and utilization of physical measurement data is still outstanding. In this work, the framework of unconditional representation for the conditional score function (UCoS) is evaluated for linearized difference imaging in diffuse optical tomography (DOT). DOT uses boundary measurements of near-infrared light to estimate the spatial distribution of absorption and scattering parameters in biological tissues. The problem is highly ill-posed and thus sensitive to noise and modelling errors. We introduce a novel regularization approach that prevents overfitting of the score function by constructing a mixed score composed of a learned and a model-based component. Validation of this approach is done using both simulated and experimental measurement data. The experiments demonstrate that a data-driven prior distribution results in posterior samples with low variance, compared to classical model-based estimation, and centred around the ground truth, even in the context of a highly ill-posed problem and in the presence of modelling errors.", "date": "2026-02-03", "categories": []}
{"id": "2411.02843", "title": "Advances in Photoacoustic Imaging Reconstruction and Quantitative Analysis for Biomedical Applications", "abstract": "Photoacoustic imaging (PAI) represents an innovative biomedical imaging modality that harnesses the advantages of optical resolution and acoustic penetration depth while ensuring enhanced safety. Despite its promising potential across a diverse array of preclinical and clinical applications, the clinical implementation of PAI faces significant challenges, including the trade-off between penetration depth and spatial resolution, as well as the demand for faster imaging speeds. This paper explores the fundamental principles underlying PAI, with a particular emphasis on three primary implementations: photoacoustic computed tomography (PACT), photoacoustic microscopy (PAM), and photoacoustic endoscopy (PAE). We undertake a critical assessment of their respective strengths and practical limitations. Furthermore, recent developments in utilizing conventional or deep learning (DL) methodologies for image reconstruction and artefact mitigation across PACT, PAM, and PAE are outlined, demonstrating considerable potential to enhance image quality and accelerate imaging processes. Furthermore, this paper examines the recent developments in quantitative analysis within PAI, including the quantification of haemoglobin concentration, oxygen saturation, and other physiological parameters within tissues. Finally, our discussion encompasses current trends and future directions in PAI research while emphasizing the transformative impact of deep learning on advancing PAI.", "date": "2024-11-05", "categories": []}
{"id": "2512.03020", "title": "Unrolled Networks are Conditional Probability Flows in MRI Reconstruction", "abstract": "Unrolled networks have been widely used for Magnetic Resonance Imaging (MRI) reconstruction due to their efficiency. However, they typically exhibit unstable output quality across cascades, resulting in sub-optimal final reconstruction results. In this work, we address this inherent limitation of unrolled networks, drawing inspiration from recent Flow Matching paradigm. We first theoretically prove that unrolled networks are discretizations of conditional probability flows. This connection shows that unrolled networks and Flow Matching are analogous in MRI reconstruction. Building upon this insight, we propose FLow-Aligned Training (FLAT), which (1) derives important cascade parameters from the Flow Matching discretization; and (2) aligns intermediate reconstructions with the ideal Flow Matching trajectory to improve cascade iteration stability and convergence. Experiments on three MRI datasets show that FLAT results in a stable trajectory across sub-networks, improving the quality of the final reconstruction.", "date": "2025-12-02", "categories": []}
{"id": "2602.00215", "title": "A Renderer-Enabled Framework for Computing Parameter Estimation Lower Bounds in Plenoptic Imaging Systems", "abstract": "This work focuses on assessing the information-theoretic limits of scene parameter estimation in plenoptic imaging systems. A general framework to compute lower bounds on the parameter estimation error from noisy plenoptic observations is presented, with a particular focus on passive indirect imaging problems, where the observations do not contain line-of-sight information about the parameter(s) of interest. Using computer graphics rendering software to synthesize the often-complicated dependence among parameter(s) of interest and observations, i.e. the forward model, the proposed framework evaluates the Hammersley-Chapman-Robbins bound to establish lower bounds on the variance of any unbiased estimator of the unknown parameters. The effects of inexact rendering of the true forward model on the computed lower bounds are also analyzed, both theoretically and via simulations. Experimental evaluations compare the computed lower bounds with the performance of the Maximum Likelihood Estimator on a canonical object localization problem, showing that the lower bounds computed via the framework proposed here are indicative of the true underlying fundamental limits in several nominally representative scenarios.", "date": "2026-01-30", "categories": []}
{"id": "2602.00348", "title": "MASC: Metal-Aware Sampling and Correction via Reinforcement Learning for Accelerated MRI", "abstract": "Metal implants in MRI cause severe artifacts that degrade image quality and hinder clinical diagnosis. Traditional approaches address metal artifact reduction (MAR) and accelerated MRI acquisition as separate problems. We propose MASC, a unified reinforcement learning framework that jointly optimizes metal-aware k-space sampling and artifact correction for accelerated MRI. To enable supervised training, we construct a paired MRI dataset using physics-based simulation, generating k-space data and reconstructions for phantoms with and without metal implants. This paired dataset provides simulated 3D MRI scans with and without metal implants, where each metal-corrupted sample has an exactly matched clean reference, enabling direct supervision for both artifact reduction and acquisition policy learning. We formulate active MRI acquisition as a sequential decision-making problem, where an artifact-aware Proximal Policy Optimization (PPO) agent learns to select k-space phase-encoding lines under a limited acquisition budget. The agent operates on undersampled reconstructions processed through a U-Net-based MAR network, learning patterns that maximize reconstruction quality. We further propose an end-to-end training scheme where the acquisition policy learns to select k-space lines that best support artifact removal while the MAR network simultaneously adapts to the resulting undersampling patterns. Experiments demonstrate that MASC's learned policies outperform conventional sampling strategies, and end-to-end training improves performance compared to using a frozen pre-trained MAR network, validating the benefit of joint optimization. Cross-dataset experiments on FastMRI with physics-based artifact simulation further confirm generalization to realistic clinical MRI data. The code and models of MASC have been made publicly available: https://github.com/hrlblab/masc", "date": "2026-01-30", "categories": []}
{"id": "2602.00490", "title": "HSSDCT: Factorized Spatial-Spectral Correlation for Hyperspectral Image Fusion", "abstract": "Hyperspectral image (HSI) fusion aims to reconstruct a high-resolution HSI (HR-HSI) by combining the rich spectral information of a low-resolution HSI (LR-HSI) with the fine spatial details of a high-resolution multispectral image (HR-MSI). Although recent deep learning methods have achieved notable progress, they still suffer from limited receptive fields, redundant spectral bands, and the quadratic complexity of self-attention, which restrict both efficiency and robustness. To overcome these challenges, we propose the Hierarchical Spatial-Spectral Dense Correlation Network (HSSDCT). The framework introduces two key modules: (i) a Hierarchical Dense-Residue Transformer Block (HDRTB) that progressively enlarges windows and employs dense-residue connections for multi-scale feature aggregation, and (ii) a Spatial-Spectral Correlation Layer (SSCL) that explicitly factorizes spatial and spectral dependencies, reducing self-attention to linear complexity while mitigating spectral redundancy. Extensive experiments on benchmark datasets demonstrate that HSSDCT delivers superior reconstruction quality with significantly lower computational costs, achieving new state-of-the-art performance in HSI fusion. Our code is available at https://github.com/jemmyleee/HSSDCT.", "date": "2026-01-31", "categories": []}
{"id": "2602.01057", "title": "Radioactive 3D Gaussian Ray Tracing for Tomographic Reconstruction", "abstract": "3D Gaussian Splatting (3DGS) has recently emerged in computer vision as a promising rendering technique. By adapting the principles of Elliptical Weighted Average (EWA) splatting to a modern differentiable pipeline, 3DGS enables real-time, high-quality novel view synthesis. Building upon this, R2-Gaussian extended the 3DGS paradigm to tomographic reconstruction by rectifying integration bias, achieving state-of-the-art performance in computed tomography (CT). To enable differentiability, R2-Gaussian adopts a local affine approximation: each 3D Gaussian is locally mapped to a 2D Gaussian on the detector and composed via alpha blending to form projections. However, the affine approximation can degrade reconstruction quantitative accuracy and complicate the incorporation of nonlinear geometric corrections. To address these limitations, we propose a tomographic reconstruction framework based on 3D Gaussian ray tracing. Our approach provides two key advantages over splatting-based models: (i) it computes the line integral through 3D Gaussian primitives analytically, avoiding the local affine collapse and thus yielding a more physically consistent forward projection model; and (ii) the ray-tracing formulation gives explicit control over ray origins and directions, which facilitates the precise application of nonlinear geometric corrections, e.g., arc-correction used in positron emission tomography (PET). These properties extend the applicability of Gaussian-based reconstruction to a wider range of realistic tomography systems while improving projection accuracy.", "date": "2026-02-01", "categories": []}
{"id": "2602.01268", "title": "OASIS-DC: Generalizable Depth Completion via Output-level Alignment of Sparse-Integrated Monocular Pseudo Depth", "abstract": "Recent monocular foundation models excel at zero-shot depth estimation, yet their outputs are inherently relative rather than metric, limiting direct use in robotics and autonomous driving. We leverage the fact that relative depth preserves global layout and boundaries: by calibrating it with sparse range measurements, we transform it into a pseudo metric depth prior. Building on this prior, we design a refinement network that follows the prior where reliable and deviates where necessary, enabling accurate metric predictions from very few labeled samples. The resulting system is particularly effective when curated validation data are unavailable, sustaining stable scale and sharp edges across few-shot regimes. These findings suggest that coupling foundation priors with sparse anchors is a practical route to robust, deployment-ready depth completion under real-world label scarcity.", "date": "2026-02-01", "categories": []}
{"id": "2602.01681", "title": "Hyperspectral Image Fusion with Spectral-Band and Fusion-Scale Agnosticism", "abstract": "Current deep learning models for Multispectral and Hyperspectral Image Fusion (MS/HS fusion) are typically designed for fixed spectral bands and spatial scales, which limits their transferability across diverse sensors. To address this, we propose SSA, a universal framework for MS/HS fusion with spectral-band and fusion-scale agnosticism. Specifically, we introduce Matryoshka Kernel (MK), a novel operator that enables a single model to adapt to arbitrary numbers of spectral channels. Meanwhile, we build SSA upon an Implicit Neural Representation (INR) backbone that models the HS signal as a continuous function, enabling reconstruction at arbitrary spatial resolutions. Together, these two forms of agnosticism enable a single MS/HS fusion model that generalizes effectively to unseen sensors and spatial scales. Extensive experiments demonstrate that our single model achieves state-of-the-art performance while generalizing well to unseen sensors and scales, paving the way toward future HS foundation models.", "date": "2026-02-02", "categories": []}
{"id": "2601.12526", "title": "Deep Lightweight Unrolled Network for High Dynamic Range Modulo Imaging", "abstract": "Modulo-Imaging (MI) offers a promising alternative for expanding the dynamic range of images by resetting the signal intensity when it reaches the saturation level. Subsequently, high-dynamic range (HDR) modulo imaging requires a recovery process to obtain the HDR image. MI is a non-convex and ill-posed problem where recent recovery networks suffer in high-noise scenarios. In this work, we formulate the HDR reconstruction task as an optimization problem that incorporates a deep prior and subsequently unrolls it into an optimization-inspired deep neural network. The network employs a lightweight convolutional denoiser for fast inference with minimal computational overhead, effectively recovering intensity values while mitigating noise. Moreover, we introduce the Scaling Equivariance term that facilitates self-supervised fine-tuning, thereby enabling the model to adapt to new modulo images that fall outside the original training distribution. Extensive evaluations demonstrate the superiority of our method compared to state-of-the-art recovery algorithms in terms of performance and quality.", "date": "2026-01-18", "categories": []}
{"id": "2601.22878", "title": "Development of Domain-Invariant Visual Enhancement and Restoration (DIVER) Approach for Underwater Images", "abstract": "Underwater images suffer severe degradation due to wavelength-dependent attenuation, scattering, and illumination non-uniformity that vary across water types and depths. We propose an unsupervised Domain-Invariant Visual Enhancement and Restoration (DIVER) framework that integrates empirical correction with physics-guided modeling for robust underwater image enhancement. DIVER first applies either IlluminateNet for adaptive luminance enhancement or a Spectral Equalization Filter for spectral normalization. An Adaptive Optical Correction Module then refines hue and contrast using channel-adaptive filtering, while Hydro-OpticNet employs physics-constrained learning to compensate for backscatter and wavelength-dependent attenuation. The parameters of IlluminateNet and Hydro-OpticNet are optimized via unsupervised learning using a composite loss function. DIVER is evaluated on eight diverse datasets covering shallow, deep, and highly turbid environments, including both naturally low-light and artificially illuminated scenes, using reference and non-reference metrics. While state-of-the-art methods such as WaterNet, UDNet, and Phaseformer perform reasonably in shallow water, their performance degrades in deep, unevenly illuminated, or artificially lit conditions. In contrast, DIVER consistently achieves best or near-best performance across all datasets, demonstrating strong domain-invariant capability. DIVER yields at least a 9% improvement over SOTA methods in UCIQE. On the low-light SeaThru dataset, where color-palette references enable direct evaluation of color restoration, DIVER achieves at least a 4.9% reduction in GPMAE compared to existing methods. Beyond visual quality, DIVER also improves robotic perception by enhancing ORB-based keypoint repeatability and matching performance, confirming its robustness across diverse underwater environments.", "date": "2026-01-30", "categories": []}
{"id": "2601.23148", "title": "Compressed BC-LISTA via Low-Rank Convolutional Decomposition", "abstract": "We study Sparse Signal Recovery (SSR) methods for multichannel imaging with compressed {forward and backward} operators that preserve reconstruction accuracy. We propose a Compressed Block-Convolutional (C-BC) measurement model based on a low-rank Convolutional Neural Network (CNN) decomposition that is analytically initialized from a low-rank factorization of physics-derived forward/backward operators in time delay-based measurements. We use Orthogonal Matching Pursuit (OMP) to select a compact set of basis filters from the analytic model and compute linear mixing coefficients to approximate the full model. We consider the Learned Iterative Shrinkage-Thresholding Algorithm (LISTA) network as a representative example for which the C-BC-LISTA extension is presented. In simulated multichannel ultrasound imaging across multiple Signal-to-Noise Ratios (SNRs), C-BC-LISTA requires substantially fewer parameters and smaller model size than other state-of-the-art (SOTA) methods while improving reconstruction accuracy. In ablations over OMP, Singular Value Decomposition (SVD)-based, and random initializations, OMP-initialized structured compression performs best, yielding the most efficient training and the best performance.", "date": "2026-01-30", "categories": []}
{"id": "2409.05809", "title": "OmniLens: Towards Universal Lens Aberration Correction via LensLib-to-Specific Domain Adaptation", "abstract": "Emerging universal Computational Aberration Correction (CAC) paradigms provide an inspiring solution to light-weight and high-quality imaging with a universal model trained on a lens library (LensLib) to address arbitrary lens optical aberrations blindly. However, the limited coverage of existing LensLibs leads to poor generalization of the trained models to unseen lenses, whose fine-tuning pipeline is also confined to the lens-descriptions-known case. In this work, we introduce OmniLens, a flexible solution to universal CAC via (i) establishing a convincing LensLib with comprehensive coverage for pre-training a robust base model, and (ii) adapting the model to any specific lens designs with unknown lens descriptions via fast LensLib-to-specific domain adaptation. To achieve these, an Evolution-based Automatic Optical Design (EAOD) pipeline is proposed to generate a rich variety of lens samples with realistic aberration behaviors. Then, we design an unsupervised regularization term for efficient domain adaptation on a few easily accessible real-captured images based on the statistical observation of dark channel priors in degradation induced by lens aberrations. Extensive experiments demonstrate that the LensLib generated by EAOD effectively develops a universal CAC model with strong generalization capabilities, which can also improve the non-blind lens-specific methods by 0.35~1.81dB in PSNR. Additionally, the proposed domain adaptation method significantly improves the base model, especially in severe aberration cases (at most 2.59dB in PSNR). The code and data will be available at https://github.com/zju-jiangqi/OmniLens.", "date": "2024-09-09", "categories": []}
{"id": "2601.20995", "title": "Low performing pixel correction in computed tomography with unrolled network and synthetic data training", "abstract": "Low performance pixels (LPP) in Computed Tomography (CT) detectors would lead to ring and streak artifacts in the reconstructed images, making them clinically unusable. In recent years, several solutions have been proposed to correct LPP artifacts, either in the image domain or in the sinogram domain using supervised deep learning methods. However, these methods require dedicated datasets for training, which are expensive to collect. Moreover, existing approaches focus solely either on image-space or sinogram-space correction, ignoring the intrinsic correlations from the forward operation of the CT geometry. In this work, we propose an unrolled dual-domain method based on synthetic data to correct LPP artifacts. Specifically, the intrinsic correlations of LPP between the sinogram and image domains are leveraged through synthetic data generated from natural images, enabling the trained model to correct artifacts without requiring any real-world clinical data. In experiments simulating 1-2% detectors defect near the isocenter, the proposed method outperformed the state-of-the-art approaches by a large margin. The results indicate that our solution can correct LPP artifacts without the cost of data collection for model training, and it is adaptable to different scanner settings for software-based applications.", "date": "2026-01-28", "categories": []}
{"id": "2601.21248", "title": "NFCDS: A Plug-and-Play Noise Frequency-Controlled Diffusion Sampling Strategy for Image Restoration", "abstract": "Diffusion sampling-based Plug-and-Play (PnP) methods produce images with high perceptual quality but often suffer from reduced data fidelity, primarily due to the noise introduced during reverse diffusion. To address this trade-off, we propose Noise Frequency-Controlled Diffusion Sampling (NFCDS), a spectral modulation mechanism for reverse diffusion noise. We show that the fidelity-perception conflict can be fundamentally understood through noise frequency: low-frequency components induce blur and degrade fidelity, while high-frequency components drive detail generation. Based on this insight, we design a Fourier-domain filter that progressively suppresses low-frequency noise and preserves high-frequency content. This controlled refinement injects a data-consistency prior directly into sampling, enabling fast convergence to results that are both high-fidelity and perceptually convincing--without additional training. As a PnP module, NFCDS seamlessly integrates into existing diffusion-based restoration frameworks and improves the fidelity-perception balance across diverse zero-shot tasks.", "date": "2026-01-29", "categories": []}
{"id": "2601.21338", "title": "SR$^{2}$-Net: A General Plug-and-Play Model for Spectral Refinement in Hyperspectral Image Super-Resolution", "abstract": "HSI-SR aims to enhance spatial resolution while preserving spectrally faithful and physically plausible characteristics. Recent methods have achieved great progress by leveraging spatial correlations to enhance spatial resolution. However, these methods often neglect spectral consistency across bands, leading to spurious oscillations and physically implausible artifacts. While spectral consistency can be addressed by designing the network architecture, it results in a loss of generality and flexibility. To address this issue, we propose a lightweight plug-and-play rectifier, physically priors Spectral Rectification Super-Resolution Network (SR$^{2}$-Net), which can be attached to a wide range of HSI-SR models without modifying their architectures. SR$^{2}$-Net follows an enhance-then-rectify pipeline consisting of (i) Hierarchical Spectral-Spatial Synergy Attention (H-S$^{3}$A) to reinforce cross-band interactions and (ii) Manifold Consistency Rectification (MCR) to constrain the reconstructed spectra to a compact, physically plausible spectral manifold. In addition, we introduce a degradation-consistency loss to enforce data fidelity by encouraging the degraded SR output to match the observed low resolution input. Extensive experiments on multiple benchmarks and diverse backbones demonstrate consistent improvements in spectral fidelity and overall reconstruction quality with negligible computational overhead. Our code will be released upon publication.", "date": "2026-01-29", "categories": []}
{"id": "2501.09718", "title": "FLOL: Fast Baselines for Real-World Low-Light Enhancement", "abstract": "Low-Light Image Enhancement (LLIE) is a key task in computational photography and imaging. The problem of enhancing images captured during night or in dark environments has been well-studied in the computer vision literature. However, current deep learning-based solutions struggle with efficiency and robustness for real-world scenarios (e.g., scenes with noise, saturated pixels). We propose a lightweight neural network that combines image processing in the frequency and spatial domains. Our baseline method, FLOL, is one of the fastest models for this task, achieving results comparable to the state-of-the-art on popular real-world benchmarks such as LOLv2, LSRW, MIT-5K and UHD-LL. Moreover, we are able to process 1080p images in real-time under 12ms. Code and models at https://github.com/cidautai/FLOL", "date": "2025-01-16", "categories": []}
{"id": "2601.07599", "title": "Diffusion in SPAD Signals", "abstract": "We derive the likelihood of a raw signal in a single photon avalanche diode (SPAD), given a fixed photon flux. The raw signal comprises timing of detection events, which are nonlinearly related to the flux. Moreover, they are naturally stochastic. We then derive a score function of the signal. This is a key for solving inverse problems based on SPAD signals. We focus on deriving solutions involving a diffusion model, to express image priors. We demonstrate the effect of low or high photon counts, and the consequence of exploiting timing of detection events.", "date": "2026-01-12", "categories": []}
{"id": "2510.26601", "title": "ResMatching: Noise-Resilient Computational Super-Resolution via Guided Conditional Flow Matching", "abstract": "Computational Super-Resolution (CSR) in fluorescence microscopy has, despite being an ill-posed problem, a long history. At its very core, CSR is about finding a prior that can be used to extrapolate frequencies in a micrograph that have never been imaged by the image-generating microscope. It stands to reason that, with the advent of better data-driven machine learning techniques, stronger prior can be learned and hence CSR can lead to better results. Here, we present ResMatching, a novel CSR method that uses guided conditional flow matching to learn such improved data-priors. We evaluate ResMatching on 4 diverse biological structures from the BioSR dataset and compare its results against 7 baselines. ResMatching consistently achieves competitive results, demonstrating in all cases the best trade-off between data fidelity and perceptual realism. We observe that CSR using ResMatching is particularly effective in cases where a strong prior is hard to learn, e.g. when the given low-resolution images contain a lot of noise. Additionally, we show that ResMatching can be used to sample from an implicitly learned posterior distribution and that this distribution is calibrated for all tested use-cases, enabling our method to deliver a pixel-wise data-uncertainty term that can guide future users to reject uncertain predictions.", "date": "2025-10-30", "categories": []}
{"id": "2601.19157", "title": "GTFMN: Guided Texture and Feature Modulation Network for Low-Light Image Enhancement and Super-Resolution", "abstract": "Low-light image super-resolution (LLSR) is a challenging task due to the coupled degradation of low resolution and poor illumination. To address this, we propose the Guided Texture and Feature Modulation Network (GTFMN), a novel framework that decouples the LLSR task into two sub-problems: illumination estimation and texture restoration. First, our network employs a dedicated Illumination Stream whose purpose is to predict a spatially varying illumination map that accurately captures lighting distribution. Further, this map is utilized as an explicit guide within our novel Illumination Guided Modulation Block (IGM Block) to dynamically modulate features in the Texture Stream. This mechanism achieves spatially adaptive restoration, enabling the network to intensify enhancement in poorly lit regions while preserving details in well-exposed areas. Extensive experiments demonstrate that GTFMN achieves the best performance among competing methods on the OmniNormal5 and OmniNormal15 datasets, outperforming them in both quantitative metrics and visual quality.", "date": "2026-01-27", "categories": []}
{"id": "2601.19243", "title": "Contrast-Source-Based Physics-Driven Neural Network for Inverse Scattering Problems", "abstract": "Deep neural networks (DNNs) have recently been applied to inverse scattering problems (ISPs) due to their strong nonlinear mapping capabilities. However, supervised DNN solvers require large-scale datasets, which limits their generalization in practical applications. Untrained neural networks (UNNs) address this issue by updating weights from measured electric fields and prior physical knowledge, but existing UNN solvers suffer from long inference time. To overcome these limitations, this paper proposes a contrast-source-based physics-driven neural network (CSPDNN), which predicts the induced current distribution to improve efficiency and incorporates an adaptive total variation loss for robust reconstruction under varying contrast and noise conditions. The improved imaging performance is validated through comprehensive numerical simulations and experimental data.", "date": "2026-01-27", "categories": []}
{"id": "2601.19403", "title": "Learned split-spectrum metalens for obstruction-free broadband imaging in the visible", "abstract": "Obstructions such as raindrops, fences, or dust degrade captured images, especially when mechanical cleaning is infeasible. Conventional solutions to obstructions rely on a bulky compound optics array or computational inpainting, which compromise compactness or fidelity. Metalenses composed of subwavelength meta-atoms promise compact imaging, but simultaneous achievement of broadband and obstruction-free imaging remains a challenge, since a metalens that images distant scenes across a broadband spectrum cannot properly defocus near-depth occlusions. Here, we introduce a learned split-spectrum metalens that enables broadband obstruction-free imaging. Our approach divides the spectrum of each RGB channel into pass and stop bands with multi-band spectral filtering and learns the metalens to focus light from far objects through pass bands, while filtering focused near-depth light through stop bands. This optical signal is further enhanced using a neural network. Our learned split-spectrum metalens achieves broadband and obstruction-free imaging with relative PSNR gains of 32.29% and improves object detection and semantic segmentation accuracies with absolute gains of +13.54% mAP, +48.45% IoU, and +20.35% mIoU over a conventional hyperbolic design. This promises robust obstruction-free sensing and vision for space-constrained systems, such as mobile robots, drones, and endoscopes.", "date": "2026-01-27", "categories": []}
{"id": "2502.03502", "title": "DC-VSR: Spatially and Temporally Consistent Video Super-Resolution with Video Diffusion Prior", "abstract": "Video super-resolution (VSR) aims to reconstruct a high-resolution (HR) video from a low-resolution (LR) counterpart. Achieving successful VSR requires producing realistic HR details and ensuring both spatial and temporal consistency. To restore realistic details, diffusion-based VSR approaches have recently been proposed. However, the inherent randomness of diffusion, combined with their tile-based approach, often leads to spatio-temporal inconsistencies. In this paper, we propose DC-VSR, a novel VSR approach to produce spatially and temporally consistent VSR results with realistic textures. To achieve spatial and temporal consistency, DC-VSR adopts a novel Spatial Attention Propagation (SAP) scheme and a Temporal Attention Propagation (TAP) scheme that propagate information across spatio-temporal tiles based on the self-attention mechanism. To enhance high-frequency details, we also introduce Detail-Suppression Self-Attention Guidance (DSSAG), a novel diffusion guidance scheme. Comprehensive experiments demonstrate that DC-VSR achieves spatially and temporally consistent, high-quality VSR results, outperforming previous approaches.", "date": "2025-02-05", "categories": []}
{"id": "2601.17047", "title": "A Contrastive Pre-trained Foundation Model for Deciphering Imaging Noisomics across Modalities", "abstract": "Characterizing imaging noise is notoriously data-intensive and device-dependent, as modern sensors entangle physical signals with complex algorithmic artifacts. Current paradigms struggle to disentangle these factors without massive supervised datasets, often reducing noise to mere interference rather than an information resource. Here, we introduce \"Noisomics\", a framework shifting the focus from suppression to systematic noise decoding via the Contrastive Pre-trained (CoP) Foundation Model. By leveraging the manifold hypothesis and synthetic noise genome, CoP employs contrastive learning to disentangle semantic signals from stochastic perturbations. Crucially, CoP breaks traditional deep learning scaling laws, achieving superior performance with only 100 training samples, outperforming supervised baselines trained on 100,000 samples, thereby reducing data and computational dependency by three orders of magnitude. Extensive benchmarking across 12 diverse out-of-domain datasets confirms its robust zero-shot generalization, demonstrating a 63.8% reduction in estimation error and an 85.1% improvement in the coefficient of determination compared to the conventional training strategy. We demonstrate CoP's utility across scales: from deciphering non-linear hardware-noise interplay in consumer photography to optimizing photon-efficient protocols for deep-tissue microscopy. By decoding noise as a multi-parametric footprint, our work redefines stochastic degradation as a vital information resource, empowering precise imaging diagnostics without prior device calibration.", "date": "2026-01-21", "categories": []}
{"id": "2601.17165", "title": "Benchmarking Deep Learning-Based Reconstruction Methods for Photoacoustic Computed Tomography with Clinically Relevant Synthetic Datasets", "abstract": "Deep learning (DL)-based image reconstruction methods for photoacoustic computed tomography (PACT) have developed rapidly in recent years. However, most existing methods have not employed standardized datasets, and their evaluations rely on traditional image quality (IQ) metrics that may lack clinical relevance. The absence of a standardized framework for clinically meaningful IQ assessment hinders fair comparison and raises concerns about the reproducibility and reliability of reported advancements in PACT. A benchmarking framework is proposed that provides open-source, anatomically plausible synthetic datasets and evaluation strategies for DL-based acoustic inversion methods in PACT. The datasets each include over 11,000 two-dimensional (2D) stochastic breast objects with clinically relevant lesions and paired measurements at varying modeling complexity. The evaluation strategies incorporate both traditional and task-based IQ measures to assess fidelity and clinical utility. A preliminary benchmarking study is conducted to demonstrate the framework's utility by comparing DL-based and physics-based reconstruction methods. The benchmarking study demonstrated that the proposed framework enabled comprehensive, quantitative comparisons of reconstruction performance and revealed important limitations in certain DL-based methods. Although they performed well according to traditional IQ measures, they often failed to accurately recover lesions. This highlights the inadequacy of traditional metrics and motivates the need for task-based assessments. The proposed benchmarking framework enables systematic comparisons of DL-based acoustic inversion methods for 2D PACT. By integrating clinically relevant synthetic datasets with rigorous evaluation protocols, it enables reproducible, objective assessments and facilitates method development and system optimization in PACT.", "date": "2026-01-23", "categories": []}
{"id": "2601.18021", "title": "MlPET: A Localized Neural Network Approach for Probabilistic Post-Reconstruction PET Image Analysis Using Informed Priors", "abstract": "We develop and evaluate MlPET, a fast localized machine learning approach for probabilistic PET image analysis addressing the noise-resolution trade-off in conventional reconstructions. MlPET replaces computationally demanding Markov chain Monte Carlo sampling with a localized neural network trained to estimate posterior mean voxel activity from small image neighborhoods. The method incorporates scanner-specific point spread functions, spatially correlated noise modeling, and flexible priors. Performance was evaluated on NEMA IEC phantom data from three PET systems (GE Discovery MI, Siemens Biograph Vision 600, and Quadra) under varying reconstruction settings and acquisition times. On phantom data, MlPET achieved contrast recovery coefficients consistently higher than standard PET and close to 1.0 (including 10 mm spheres), while reducing background noise and improving spatial definition. Effective pointspread function full width at half maximum decreased from approximately 2 mm in standard PET to below 1 mm with MlPET, a 2.5 fold reduction in blur. Comparable image quality was obtained at 40-80 s acquisition time with MlPET versus 900 s with conventional PET. MlPET provides an efficient approach for quantitative probabilistic post-reconstruction PET analysis. By combining informed priors with neural network speed, it achieves noise suppression and resolution enhancement without altering reconstruction algorithms. The method shows promise for improved small-lesion detectability and quantitative reliability in clinical PET imaging. Future studies will evaluate performance on patient data.", "date": "2026-01-25", "categories": []}
{"id": "2601.18099", "title": "Computational Framework for Estimating Relative Gaussian Blur Kernels between Image Pairs", "abstract": "Following the earlier verification for Gaussian model in \\cite{ASaa2026}, this paper introduces a zero training forward computational framework for the model to realize it in real time applications. The framework is based on discrete calculation of the analytic expression of the defocused image from the sharper one for the application range of the standard deviation of the Gaussian kernels and selecting the best matches. The analytic expression yields multiple solutions at certain image points, but is filtered down to a single solution using similarity measures over neighboring points.The framework is structured to handle cases where two given images are partial blurred versions of each other. Experimental evaluations on real images demonstrate that the proposed framework achieves a mean absolute error (MAE) below $1.7\\%$ in estimating synthetic blur values. Furthermore, the discrepancy between actual blurred image intensities and their corresponding estimates remains under $2\\%$, obtained by applying the extracted defocus filters to less blurred images.", "date": "2026-01-26", "categories": []}
{"id": "2601.18314", "title": "A Master Class on Reproducibility: A Student Hackathon on Advanced MRI Reconstruction Methods", "abstract": "We report the design, protocol, and outcomes of a student reproducibility hackathon focused on replicating the results of three influential MRI reconstruction papers: (a) MoDL, an unrolled model-based network with learned denoising; (b) HUMUS-Net, a hybrid unrolled multiscale CNN+Transformer architecture; and (c) an untrained, physics-regularized dynamic MRI method that uses a quantitative MR model for early stopping. We describe the setup of the hackathon and present reproduction outcomes alongside additional experiments, and we detail fundamental practices for building reproducible codebases.", "date": "2026-01-26", "categories": []}
{"id": "2511.06853", "title": "Computational TIRF enables optical sectioning beyond the evanescent field for widefield fluorescence microscopy", "abstract": "The resolving ability of widefield fluorescence microscopy is fundamentally limited by out-of-focus background owing to its low axial resolution, particularly for densely labeled biological samples. Although total internal reflection fluorescence (TIRF) microscopy provides strong near-surface sectioning, they are intrinsically restricted to shallow imaging depths. Here we present computational TIRF (cTIRF), a deep learning-based imaging modality that generates TIRF-like sectioned images directly from conventional widefield epifluorescence measurements without any optical modification. By integrating a physics-informed forward model into network training, cTIRF achieves effective background suppression and axial resolution enhancement while maintaining consistency with the measured widefield data. We demonstrate that cTIRF recovers near-surface structures with performance comparable to experimental TIRF, and further enables both single-frame and volumetric sectioned reconstruction in densely labeled samples where conventional TIRF fails. This work establishes cTIRF as a practical and deployable alternative to hardware-based optical sectioning in fluorescence microscopy, enabled by rapid adaptation to new imaging systems with minimal calibration data.", "date": "2025-11-10", "categories": []}
{"id": "2601.16733", "title": "Using Shadows in Circular Synthetic Aperture Sonar Imaging for Target Analysis", "abstract": "Circular Synthetic Aperture Sonar (CSAS) provides a 360{\\deg} azimuth view of the seabed, surpassing the limited aperture and mono-view image of conventional side-scan SAS. This makes CSAS a valuable tool for target recognition in mine warfare where the diversity of point of view is essential for reducing false alarms. CSAS processing typically produces a very high-resolution two-dimensional image. However, the parallax introduced by the circular displacement of the illuminator fill-in the shadow regions, and the shadow cast by an object on the seafloor is lost in favor of azimuth coverage and resolution. Yet the shadows provide complementary information on target shape useful for target recognition. In this paper, we explore a way to retrieve shadow information from CSAS data to improve target analysis and carry 3D reconstruction. Sub-aperture filtering is used to get a collection of images at various points of view along the circular trajectory and fixed focus shadow enhancement (FFSE) is applied to obtain sharp shadows. An interactive interface is also proposed to allow human operators to visualize these shadows along the circular trajectory. A space-carving reconstruction method is applied to infer the 3D shape of the object from the segmented shadows. The results demonstrate the potential of shadows in circular SAS for improving target analysis and 3D reconstruction.", "date": "2026-01-23", "categories": []}
{"id": "2505.22797", "title": "Fast Trajectory-Independent Model-Based Reconstruction Algorithm for Multi-Dimensional Magnetic Particle Imaging", "abstract": "Magnetic Particle Imaging (MPI) is a promising tomographic technique for visualizing the spatio-temporal distribution of superparamagnetic nanoparticles, with applications ranging from cancer detection to real-time cardiovascular monitoring. Traditional MPI reconstruction relies on either time-consuming calibration (measured system matrix) or model-based simulation of the forward operator. Recent developments have shown the applicability of Chebyshev polynomials to multi-dimensional Lissajous Field-Free Point (FFP) scans. This method is bound to the particular choice of sinusoidal scanning trajectories. In this paper, we present the first reconstruction on real 2D MPI data with a trajectory-independent model-based MPI reconstruction algorithm. We further develop the zero-shot Plug-and-Play (PnP) algorithm of the authors -- with automatic noise level estimation -- to address the present deconvolution problem, leveraging a state-of-the-art denoiser trained on natural images without retraining on MPI-specific data. We evaluate our method on the publicly available 2D FFP MPI dataset ``MPIdata: Equilibrium Model with Anisotropy\", featuring scans of six phantoms acquired using a Bruker preclinical scanner. Moreover, we show reconstruction performed on custom data on a 2D scanner with additional high-frequency excitation field and partial data. Our results demonstrate strong reconstruction capabilities across different scanning scenarios -- setting a precedent for general-purpose, flexible model-based MPI reconstruction.", "date": "2025-05-28", "categories": []}
{"id": "2206.04186", "title": "Reinforced Inverse Scattering", "abstract": "Inverse wave scattering aims at determining the properties of an object using data on how the object scatters incoming waves. In order to collect information, sensors are put in different locations to send and receive waves from each other. The choice of sensor positions and incident wave frequencies determines the reconstruction quality of scatterer properties. This paper introduces reinforcement learning to develop precision imaging that decides sensor positions and wave frequencies adaptive to different scatterers in an intelligent way, thus obtaining a significant improvement in reconstruction quality with limited imaging resources. Extensive numerical results will be provided to demonstrate the superiority of the proposed method over existing methods.", "date": "2022-06-08", "categories": []}
{"id": "2509.23885", "title": "Extendable Generalization Self-Supervised Diffusion for Low-Dose CT Reconstruction", "abstract": "Current methods based on deep learning for self-supervised low-dose CT (LDCT) reconstruction, while reducing the dependence on paired data, face the problem of significantly decreased generalization when training with single-dose data and extending to other doses. To enable dose-extensive generalization using only single-dose projection data for training, this work proposes a novel method of Extendable GENeraLization self-supervised Diffusion (EGenDiff) for low-dose CT reconstruction. Specifically, a contextual subdata self-enhancing similarity strategy is designed to provide an initial prior for the subsequent progress. During training, the initial prior is used to combine knowledge distillation with a deep combination of latent diffusion models for optimizing image details. On the stage of inference, the pixel-wise self-correcting fusion technique is proposed for data fidelity enhancement, resulting in extensive generalization of higher and lower doses or even unseen doses. EGenDiff requires only LDCT projection data for training and testing. Comprehensive evaluation on benchmark datasets, clinical data, photon counting CT data, and across all three anatomical planes (transverse, coronal, and sagittal) demonstrates that EGenDiff enables extendable generalization multi-dose, yielding reconstructions that consistently outperform leading existing methods.", "date": "2025-09-28", "categories": []}
{"id": "2601.14530", "title": "PAS-Mamba: Phase-Amplitude-Spatial State Space Model for MRI Reconstruction", "abstract": "Joint feature modeling in both the spatial and frequency domains has become a mainstream approach in MRI reconstruction. However, existing methods generally treat the frequency domain as a whole, neglecting the differences in the information carried by its internal components. According to Fourier transform theory, phase and amplitude represent different types of information in the image. Our spectrum swapping experiments show that magnitude mainly reflects pixel-level intensity, while phase predominantly governs image structure. To prevent interference between phase and magnitude feature learning caused by unified frequency-domain modeling, we propose the Phase-Amplitude-Spatial State Space Model (PAS-Mamba) for MRI Reconstruction, a framework that decouples phase and magnitude modeling in the frequency domain and combines it with image-domain features for better reconstruction. In the image domain, LocalMamba preserves spatial locality to sharpen fine anatomical details. In frequency domain, we disentangle amplitude and phase into two specialized branches to avoid representational coupling. To respect the concentric geometry of frequency information, we propose Circular Frequency Domain Scanning (CFDS) to serialize features from low to high frequencies. Finally, a Dual-Domain Complementary Fusion Module (DDCFM) adaptively fuses amplitude phase representations and enables bidirectional exchange between frequency and image domains, delivering superior reconstruction. Extensive experiments on the IXI and fastMRI knee datasets show that PAS-Mamba consistently outperforms state of the art reconstruction methods.", "date": "2026-01-20", "categories": []}
{"id": "2507.09609", "title": "I2I-PR: Deep Iterative Refinement for Phase Retrieval using Image-to-Image Diffusion Models", "abstract": "Phase retrieval aims to recover a signal from intensity-only measurements, a fundamental problem in many fields such as imaging, holography, optical computing, crystallography, and microscopy. Although there are several well-known phase retrieval algorithms, including classical alternating projection-based solvers, the reconstruction performance often remains sensitive to initialization and measurement noise. Recently, diffusion models have gained traction in various image reconstruction tasks, yielding significant theoretical insights and practical advances. In this work, we introduce a deep iterative refinement framework that redefines the role of diffusion models in phase retrieval. Instead of generating images from random noise, our method starts with multiple physically consistent initial estimates and iteratively refines them through a learned image-to-image diffusion process. This enables data-driven phase retrieval that is both interpretable and robust, leveraging the strengths of classical solvers while mitigating their weaknesses. Furthermore, we propose an enhanced initialization strategy that integrates classical algorithms with a novel acceleration mechanism to obtain reliable initial estimates. During inference, we adopt a geometric self-ensemble strategy based on input flipping, together with output aggregation to further improve the final reconstruction quality. Comprehensive experiments demonstrate that our approach achieves substantial gains in both training efficiency and reconstruction quality, consistently outperforming classical and recent state-of-the-art methods. These results highlight the potential of diffusion-driven refinement as an effective and general framework for robust phase retrieval across diverse applications. The source code and trained models are available at https://github.com/METU-SPACE-Lab/I2I-PR-for-Phase-Retrieval", "date": "2025-07-13", "categories": []}
{"id": "2601.11694", "title": "Anisotropic Tensor Deconvolution of Hyperspectral Images", "abstract": "Hyperspectral image (HSI) deconvolution is a challenging ill-posed inverse problem, made difficult by the data's high dimensionality.We propose a parameter-parsimonious framework based on a low-rank Canonical Polyadic Decomposition (CPD) of the entire latent HSI $\\mathbf{\\mathcal{X}} \\in \\mathbb{R}^{P\\times Q \\times N}$.This approach recasts the problem from recovering a large-scale image with $PQN$ variables to estimating the CPD factors with $(P+Q+N)R$ variables.This model also enables a structure-aware, anisotropic Total Variation (TV) regularization applied only to the spatial factors, preserving the smooth spectral signatures.An efficient algorithm based on the Proximal Alternating Linearized Minimization (PALM) framework is developed to solve the resulting non-convex optimization problem.Experiments confirm the model's efficiency, showing a numerous parameter reduction of over two orders of magnitude and a compelling trade-off between model compactness and reconstruction accuracy.", "date": "2026-01-16", "categories": []}
{"id": "2601.12257", "title": "Soft Shadow Diffusion (SSD): Physics-inspired Learning for 3D Computational Periscopy", "abstract": "Conventional imaging requires a line of sight to create accurate visual representations of a scene. In certain circumstances, however, obtaining a suitable line of sight may be impractical, dangerous, or even impossible. Non-line-of-sight (NLOS) imaging addresses this challenge by reconstructing the scene from indirect measurements. Recently, passive NLOS methods that use an ordinary photograph of the subtle shadow cast onto a visible wall by the hidden scene have gained interest. These methods are currently limited to 1D or low-resolution 2D color imaging or to localizing a hidden object whose shape is approximately known. Here, we generalize this class of methods and demonstrate a 3D reconstruction of a hidden scene from an ordinary NLOS photograph. To achieve this, we propose a novel reformulation of the light transport model that conveniently decomposes the hidden scene into \\textit{light-occluding} and \\textit{non-light-occluding} components to yield a separable non-linear least squares (SNLLS) inverse problem. We develop two solutions: A gradient-based optimization method and a physics-inspired neural network approach, which we call Soft Shadow diffusion (SSD). Despite the challenging ill-conditioned inverse problem encountered here, our approaches are effective on numerous 3D scenes in real experimental scenarios. Moreover, SSD is trained in simulation but generalizes well to unseen classes in simulation and real-world NLOS scenes. SSD also shows surprising robustness to noise and ambient illumination.", "date": "2026-01-18", "categories": []}
{"id": "2601.12831", "title": "Data-Consistent Learning of Inverse Problems", "abstract": "Inverse problems are inherently ill-posed, suffering from non-uniqueness and instability. Classical regularization methods provide mathematically well-founded solutions, ensuring stability and convergence, but often at the cost of reduced flexibility or visual quality. Learned reconstruction methods, such as convolutional neural networks, can produce visually compelling results, yet they typically lack rigorous theoretical guarantees. DC (DC) networks address this gap by enforcing the measurement model within the network architecture. In particular, null-space networks combined with a classical regularization method as an initial reconstruction define a convergent regularization method. This approach preserves the theoretical reliability of classical schemes while leveraging the expressive power of data-driven learning, yielding reconstructions that are both accurate and visually appealing.", "date": "2026-01-19", "categories": []}
{"id": "2601.12850", "title": "Accurate Simulation Pipeline for Passive Single-Photon Imaging", "abstract": "Single-Photon Avalanche Diodes (SPADs) are new and promising imaging sensors. These sensors are sensitive enough to detect individual photons hitting each pixel, with extreme temporal resolution and without readout noise. Thus, SPADs stand out as an optimal choice for low-light imaging. Due to the high price and limited availability of SPAD sensors, the demand for an accurate data simulation pipeline is substantial. Indeed, the scarcity of SPAD datasets hinders the development of SPAD-specific processing algorithms and impedes the training of learning-based solutions.\n  In this paper, we present a comprehensive SPAD simulation pipeline and validate it with multiple experiments using two recent commercial SPAD sensors. Our simulator is used to generate the SPAD-MNIST, a single-photon version of the seminal MNIST dataset, to investigate the effectiveness of convolutional neural network (CNN) classifiers on reconstructed fluxes, even at extremely low light conditions, e.g., 5 mlux. We also assess the performance of classifiers exclusively trained on simulated data on real images acquired from SPAD sensors at different light conditions. The synthetic dataset encompasses different SPAD imaging modalities and is made available for download. Project page: https://boracchi.faculty.polimi.it/Projects/SPAD-MNIST.html.", "date": "2026-01-19", "categories": []}
{"id": "2601.13498", "title": "Optical Linear Systems Framework for Event Sensing and Computational Neuromorphic Imaging", "abstract": "Event vision sensors (neuromorphic cameras) output sparse, asynchronous ON/OFF events triggered by log-intensity threshold crossings, enabling microsecond-scale sensing with high dynamic range and low data bandwidth. As a nonlinear system, this event representation does not readily integrate with the linear forward models that underpin most computational imaging and optical system design. We present a physics-grounded processing pipeline that maps event streams to estimates of per-pixel log-intensity and intensity derivatives, and embeds these measurements in a dynamic linear systems model with a time-varying point spread function. This enables inverse filtering directly from event data, using frequency-domain Wiener deconvolution with a known (or parameterised) dynamic transfer function. We validate the approach in simulation for single and overlapping point sources under modulated defocus, and on real event data from a tunable-focus telescope imaging a star field, demonstrating source localisation and separability. The proposed framework provides a practical bridge between event sensing and model-based computational imaging for dynamic optical systems.", "date": "2026-01-20", "categories": []}
{"id": "2601.13986", "title": "Equivariant Learning for Unsupervised Image Dehazing", "abstract": "Image Dehazing (ID) aims to produce a clear image from an observation contaminated by haze. Current ID methods typically rely on carefully crafted priors or extensive haze-free ground truth, both of which are expensive or impractical to acquire, particularly in the context of scientific imaging. We propose a new unsupervised learning framework called Equivariant Image Dehazing (EID) that exploits the symmetry of image signals to restore clarity to hazy observations. By enforcing haze consistency and systematic equivariance, EID can recover clear patterns directly from raw, hazy images. Additionally, we propose an adversarial learning strategy to model unknown haze physics and facilitate EID learning. Experiments on two scientific image dehazing benchmarks (including cell microscopy and medical endoscopy) and on natural image dehazing have demonstrated that EID significantly outperforms state-of-the-art approaches. By unifying equivariant learning with modelling haze physics, we hope that EID will enable more versatile and effective haze removal in scientific imaging. Code and datasets will be published.", "date": "2026-01-20", "categories": []}
{"id": "2601.14165", "title": "ASBA: A-line State Space Model and B-line Attention for Sparse Optical Doppler Tomography Reconstruction", "abstract": "Optical Doppler Tomography (ODT) is an emerging blood flow analysis technique. A 2D ODT image (B-scan) is generated by sequentially acquiring 1D depth-resolved raw A-scans (A-line) along the lateral axis (B-line), followed by Doppler phase-subtraction analysis. To ensure high-fidelity B-scan images, current practices rely on dense sampling, which prolongs scanning time, increases storage demands, and limits the capture of rapid blood flow dynamics. Recent studies have explored sparse sampling of raw A-scans to alleviate these limitations, but their effectiveness is hindered by the conservative sampling rates and the uniform modeling of flow and background signals. In this study, we introduce a novel blood flow-aware network, named ASBA (A-line ROI State space model and B-line phase Attention), to reconstruct ODT images from highly sparsely sampled raw A-scans. Specifically, we propose an A-line ROI state space model to extract sparsely distributed flow features along the A-line, and a B-line phase attention to capture long-range flow signals along each B-line based on phase difference. Moreover, we introduce a flow-aware weighted loss function that encourages the network to prioritize the accurate reconstruction of flow signals. Extensive experiments on real animal data demonstrate that the proposed approach clearly outperforms existing state-of-the-art reconstruction methods.", "date": "2026-01-20", "categories": []}
{"id": "2503.19190", "title": "Universal Architectures for the Learning of Polyhedral Norms and Convex Regularizers", "abstract": "This paper addresses the task of learning convex regularizers to guide the reconstruction of images from limited data. By imposing that the reconstruction be amplitude-equivariant, we narrow down the class of admissible functionals to those that can be expressed as a power of a seminorm. We then show that such functionals can be approximated to arbitrary precision with the help of polyhedral norms. In particular, we identify two dual parameterizations of such systems: (i) a synthesis form with an $\\ell_1$-penalty that involves some learnable dictionary; and (ii) an analysis form with an $\\ell_\\infty$-penalty that involves a trainable regularization operator. After having provided geometric insights and proved that the two forms are universal, we propose an implementation that relies on a specific architecture (tight frame with a weighted $\\ell_1$ penalty) that is easy to train. We illustrate its use for denoising and the reconstruction of biomedical images. We find that the proposed framework outperforms the sparsity-based methods of compressed sensing, while it offers essentially the same convergence and robustness guarantees.", "date": "2025-03-24", "categories": []}
{"id": "2601.08749", "title": "A Single-Parameter Factor-Graph Image Prior", "abstract": "We propose a novel piecewise smooth image model with piecewise constant local parameters that are automatically adapted to each image. Technically, the model is formulated in terms of factor graphs with NUP (normal with unknown parameters) priors, and the pertinent computations amount to iterations of conjugate-gradient steps and Gaussian message passing. The proposed model and algorithms are demonstrated with applications to denoising and contrast enhancement.", "date": "2026-01-13", "categories": []}
{"id": "2503.08546", "title": "End-to-End PET Image Reconstruction via a Posterior-Mean Diffusion Model", "abstract": "Positron Emission Tomography (PET) is a functional imaging modality that enables the visualization of biochemical and physiological processes across various tissues. Recently, deep learning (DL)-based methods have demonstrated significant progress in directly mapping sinograms to PET images. However, regression-based DL models often yield overly smoothed reconstructions lacking of details (i.e., low distortion, low perceptual quality), whereas GAN-based and likelihood-based posterior sampling models tend to introduce undesirable artifacts in predictions (i.e., high distortion, high perceptual quality), limiting their clinical applicability. To achieve a robust perception-distortion tradeoff, we propose Posterior-Mean Denoising Diffusion Model (PMDM-PET), a novel approach that builds upon a recently established mathematical theory to explore the closed-form expression of perception-distortion function in diffusion model space for PET image reconstruction from sinograms. Specifically, PMDM-PET first obtained posterior-mean PET predictions under minimum mean square error (MSE), then optimally transports the distribution of them to the ground-truth PET images distribution. Experimental results demonstrate that PMDM-PET not only generates realistic PET images with possible minimum distortion and optimal perceptual quality but also outperforms five recent state-of-the-art (SOTA) DL baselines in both qualitative visual inspection and quantitative pixel-wise metrics PSNR (dB)/SSIM/NRMSE.", "date": "2025-03-11", "categories": []}
{"id": "2507.09608", "title": "prNet: Data-Driven Phase Retrieval via Stochastic Refinement", "abstract": "Phase retrieval is an ill-posed inverse problem in which classical and deep learning-based methods struggle to jointly achieve measurement fidelity and perceptual realism. We propose a novel framework for phase retrieval that leverages Langevin dynamics to enable efficient posterior sampling, yielding reconstructions that explicitly balance distortion and perceptual quality. Unlike conventional approaches that prioritize pixel-wise accuracy, our methods navigate the perception-distortion tradeoff through a principled combination of stochastic sampling, learned denoising, and model-based updates. The framework comprises three variants of increasing complexity, integrating theoretically grounded Langevin inference, adaptive noise schedule learning, parallel reconstruction sampling, and warm-start initialization from classical solvers. Extensive experiments demonstrate that our methods achieve state-of-the-art performance across multiple benchmarks, both in terms of fidelity and perceptual quality. The source code and trained models are available at https://github.com/METU-SPACE-Lab/prNet-for-Phase-Retrieval", "date": "2025-07-13", "categories": []}
{"id": "2510.01755", "title": "Learning Regularization Functionals for Inverse Problems: A Comparative Study", "abstract": "In recent years, a variety of learned regularization frameworks for solving inverse problems in imaging have emerged. These offer flexible modeling together with mathematical insights. The proposed methods differ in their architectural design and training strategies, making direct comparison challenging due to non-modular implementations. We address this gap by collecting and unifying the available code into a common framework. This unified view allows us to systematically compare the approaches and highlight their strengths and limitations, providing valuable insights into their future potential. We also provide concise descriptions of each method, complemented by practical guidelines.", "date": "2025-10-02", "categories": []}
{"id": "2411.07426", "title": "Evaluating Detection Thresholds: The Impact of False Positives and Negatives on Super-Resolution Ultrasound Localization Microscopy", "abstract": "Super-resolution ultrasound imaging with ultrasound localization microscopy (ULM) offers a high-resolution view of microvascular structures. Yet, ULM image quality heavily relies on precise microbubble (MB) detection. Despite the crucial role of localization algorithms, there has been limited focus on the practical pitfalls in MB detection tasks such as setting the detection threshold. This study examines how False Positives (FPs) and False Negatives (FNs) affect ULM image quality by systematically adding controlled detection errors to simulated data. Results indicate that while both FP and FN rates impact Peak Signal-to-Noise Ratio (PSNR) similarly, increasing FP rates from 0\\% to 20\\% decreases Structural Similarity Index (SSIM) by 7\\%, whereas same FN rates cause a greater drop of around 45\\%. Moreover, dense MB regions are more resilient to detection errors, while sparse regions show high sensitivity, showcasing the need for robust MB detection frameworks to enhance super-resolution imaging.", "date": "2024-11-11", "categories": []}
{"id": "2512.11354", "title": "A Multi-Mode Structured Light 3D Imaging System with Multi-Source Information Fusion for Underwater Pipeline Detection", "abstract": "Underwater pipelines are highly susceptible to corrosion, which not only shorten their service life but also pose significant safety risks. Compared with manual inspection, the intelligent real-time imaging system for underwater pipeline detection has become a more reliable and practical solution. Among various underwater imaging techniques, structured light 3D imaging can restore the sufficient spatial detail for precise defect characterization. Therefore, this paper develops a multi-mode underwater structured light 3D imaging system for pipeline detection (UW-SLD system) based on multi-source information fusion. First, a rapid distortion correction (FDC) method is employed for efficient underwater image rectification. To overcome the challenges of extrinsic calibration among underwater sensors, a factor graph-based parameter optimization method is proposed to estimate the transformation matrix between the structured light and acoustic sensors. Furthermore, a multi-mode 3D imaging strategy is introduced to adapt to the geometric variability of underwater pipelines. Given the presence of numerous disturbances in underwater environments, a multi-source information fusion strategy and an adaptive extended Kalman filter (AEKF) are designed to ensure stable pose estimation and high-accuracy measurements. In particular, an edge detection-based ICP (ED-ICP) algorithm is proposed. This algorithm integrates pipeline edge detection network with enhanced point cloud registration to achieve robust and high-fidelity reconstruction of defect structures even under variable motion conditions. Extensive experiments are conducted under different operation modes, velocities, and depths. The results demonstrate that the developed system achieves superior accuracy, adaptability and robustness, providing a solid foundation for autonomous underwater pipeline detection.", "date": "2025-12-12", "categories": []}
{"id": "2601.09439", "title": "DeepLight: A Sobolev-trained Image-to-Image Surrogate Model for Light Transport in Tissue", "abstract": "In optoacoustic imaging, recovering the absorption coefficients of tissue by inverting the light transport remains a challenging problem. Improvements in solving this problem can greatly benefit the clinical value of optoacoustic imaging. Existing variational inversion methods require an accurate and differentiable model of this light transport. As neural surrogate models allow fast and differentiable simulations of complex physical processes, they are considered promising candidates to be used in solving such inverse problems. However, there are in general no guarantees that the derivatives of these surrogate models accurately match those of the underlying physical operator. As accurate derivatives are central to solving inverse problems, errors in the model derivative can considerably hinder high fidelity reconstructions. To overcome this limitation, we present a surrogate model for light transport in tissue that uses Sobolev training to improve the accuracy of the model derivatives. Additionally, the form of Sobolev training we used is suitable for high-dimensional models in general. Our results demonstrate that Sobolev training for a light transport surrogate model not only improves derivative accuracy but also reduces generalization error for in-distribution and out-of-distribution samples. These improvements promise to considerably enhance the utility of the surrogate model in downstream tasks, especially in solving inverse problems.", "date": "2026-01-14", "categories": []}
{"id": "2512.06400", "title": "Perceptual Region-Driven Infrared-Visible Co-Fusion for Extreme Scene Enhancement", "abstract": "In photogrammetry, accurately fusing infrared (IR) and visible (VIS) spectra while preserving the geometric fidelity of visible features and incorporating thermal radiation is a significant challenge, particularly under extreme conditions. Existing methods often compromise visible imagery quality, impacting measurement accuracy. To solve this, we propose a region perception-based fusion framework that combines multi-exposure and multi-modal imaging using a spatially varying exposure (SVE) camera. This framework co-fuses multi-modal and multi-exposure data, overcoming single-exposure method limitations in extreme environments. The framework begins with region perception-based feature fusion to ensure precise multi-modal registration, followed by adaptive fusion with contrast enhancement. A structural similarity compensation mechanism, guided by regional saliency maps, optimizes IR-VIS spectral integration. Moreover, the framework adapts to single-exposure scenarios for robust fusion across different conditions. Experiments conducted on both synthetic and real-world data demonstrate superior image clarity and improved performance compared to state-of-the-art methods, as evidenced by both quantitative and visual evaluations.", "date": "2025-12-06", "categories": []}
{"id": "2601.08162", "title": "A Hardware-Algorithm Co-Designed Framework for HDR Imaging and Dehazing in Extreme Rocket Launch Environments", "abstract": "Quantitative optical measurement of critical mechanical parameters -- such as plume flow fields, shock wave structures, and nozzle oscillations -- during rocket launch faces severe challenges due to extreme imaging conditions. Intense combustion creates dense particulate haze and luminance variations exceeding 120 dB, degrading image data and undermining subsequent photogrammetric and velocimetric analyses. To address these issues, we propose a hardware-algorithm co-design framework that combines a custom Spatially Varying Exposure (SVE) sensor with a physics-aware dehazing algorithm. The SVE sensor acquires multi-exposure data in a single shot, enabling robust haze assessment without relying on idealized atmospheric models. Our approach dynamically estimates haze density, performs region-adaptive illumination optimization, and applies multi-scale entropy-constrained fusion to effectively separate haze from scene radiance. Validated on real launch imagery and controlled experiments, the framework demonstrates superior performance in recovering physically accurate visual information of the plume and engine region. This offers a reliable image basis for extracting key mechanical parameters, including particle velocity, flow instability frequency, and structural vibration, thereby supporting precise quantitative analysis in extreme aerospace environments.", "date": "2026-01-13", "categories": []}
{"id": "2601.08293", "title": "M3SR: Multi-Scale Multi-Perceptual Mamba for Efficient Spectral Reconstruction", "abstract": "The Mamba architecture has been widely applied to various low-level vision tasks due to its exceptional adaptability and strong performance. Although the Mamba architecture has been adopted for spectral reconstruction, it still faces the following two challenges: (1) Single spatial perception limits the ability to fully understand and analyze hyperspectral images; (2) Single-scale feature extraction struggles to capture the complex structures and fine details present in hyperspectral images. To address these issues, we propose a multi-scale, multi-perceptual Mamba architecture for the spectral reconstruction task, called M3SR. Specifically, we design a multi-perceptual fusion block to enhance the ability of the model to comprehensively understand and analyze the input features. By integrating the multi-perceptual fusion block into a U-Net structure, M3SR can effectively extract and fuse global, intermediate, and local features, thereby enabling accurate reconstruction of hyperspectral images at multiple scales. Extensive quantitative and qualitative experiments demonstrate that the proposed M3SR outperforms existing state-of-the-art methods while incurring a lower computational cost.", "date": "2026-01-13", "categories": []}
{"id": "2601.08666", "title": "Blind Deconvolution in Astronomy: How Does a Standalone U-Net Perform?", "abstract": "Aims: This study investigates whether a U-Net architecture can perform standalone end-to-end blind deconvolution of astronomical images without any prior knowledge of the Point Spread Function (PSF) or noise characteristics. Our goal is to evaluate its performance against the number of training images, classical Tikhonov deconvolution and to assess its generalization capability under varying seeing conditions and noise levels.\n  Methods: Realistic astronomical observations are simulated using the GalSim toolkit, incorporating random transformations, PSF convolution (accounting for both optical and atmospheric effects), and Gaussian white noise. A U-Net model is trained using a Mean Square Error (MSE) loss function on datasets of varying sizes, up to 40,000 images of size 48x48 from the COSMOS Real Galaxy Dataset. Performance is evaluated using PSNR, SSIM, and cosine similarity metrics, with the latter employed in a two-model framework to assess solution stability.\n  Results: The U-Net model demonstrates effectiveness in blind deconvolution, with performance improving consistently as the training dataset size increases, saturating beyond 5,000 images. Cosine similarity analysis reveals convergence between independently trained models, indicating stable solutions. Remarkably, the U-Net outperforms the oracle-like Tikhonov method in challenging conditions (low PSNR/medium SSIM). The model also generalizes well to unseen seeing and noise conditions, although optimal performance is achieved when training parameters include validation conditions. Experiments on synthetic $C^\\alpha$ images further support the hypothesis that the U-Net learns a geometry-adaptive harmonic basis, akin to sparse representations observed in denoising tasks. These results align with recent mathematical insights into its adaptive learning capabilities.", "date": "2026-01-13", "categories": []}
{"id": "2305.11321", "title": "JoIN: Joint GANs Inversion for Intrinsic Image Decomposition", "abstract": "Intrinsic Image Decomposition (IID) is a challenging inverse problem that seeks to decompose a natural image into its underlying intrinsic components such as albedo and shading. While recent image decomposition methods rely on learning-based priors on these components, they often suffer from component cross-contamination owing to joint training of priors; or from Sim-to-Real gap since the priors trained on synthetic data are kept frozen during the inference on real images. In this work, we propose to solve the intrinsic image decomposition problem using a bank of Generative Adversarial Networks (GANs) as priors where each GAN is independently trained only on a single intrinsic component, providing stronger and more disentangled priors. At the core of our approach is the idea that the latent space of a GAN is a well-suited optimization domain to solve inverse problems. Given an input image, we propose to jointly invert the latent codes of a set of GANs and combine their outputs to reproduce the input. Contrary to all existing GAN inversion methods that are limited to inverting only a single GAN, our proposed approach, JoIN, is able to jointly invert multiple GANs using only a single image as supervision while still maintaining distribution priors of each intrinsic component. We show that our approach is modular, allowing various forward imaging models, and that it can successfully decompose both synthetic and real images. Further, taking inspiration from existing GAN inversion approaches, we allow for careful fine-tuning of the generator priors during the inference on real images. This way, our method is able to achieve excellent generalization on real images even though it uses only synthetic data to train the GAN priors. We demonstrate the success of our approach through exhaustive qualitative and quantitative evaluations and ablation studies on various datasets.", "date": "2023-05-18", "categories": []}
{"id": "2511.12471", "title": "Diffusion Model Based Signal Recovery Under 1-Bit Quantization", "abstract": "Diffusion models (DMs) have demonstrated to be powerful priors for signal recovery, but their application to 1-bit quantization tasks, such as 1-bit compressed sensing and logistic regression, remains a challenge. This difficulty stems from the inherent non-linear link function in these tasks, which is either non-differentiable or lacks an explicit characterization. To tackle this issue, we introduce Diff-OneBit, which is a fast and effective DM-based approach for signal recovery under 1-bit quantization. Diff-OneBit addresses the challenge posed by non-differentiable or implicit links functions via leveraging a differentiable surrogate likelihood function to model 1-bit quantization, thereby enabling gradient based iterations. This function is integrated into a flexible plug-and-play framework that decouples the data-fidelity term from the diffusion prior, allowing any pretrained DM to act as a denoiser within the iterative reconstruction process. Extensive experiments on the FFHQ, CelebA and ImageNet datasets demonstrate that Diff-OneBit gives high-fidelity reconstructed images, outperforming state-of-the-art methods in both reconstruction quality and computational efficiency across 1-bit compressed sensing and logistic regression tasks. Our code is available at https://github.com/Chenyouming123/DiffOneBit.", "date": "2025-11-16", "categories": []}
{"id": "2512.00796", "title": "CircleFlow: Flow-Guided Camera Blur Estimation using a Circle Grid Target", "abstract": "The point spread function (PSF) serves as a fundamental descriptor linking the real-world scene to the captured signal, manifesting as camera blur. Accurate PSF estimation is crucial for both optical characterization and computational vision, yet remains challenging due to the inherent ambiguity and the ill-posed nature of intensity-based deconvolution. We introduce CircleFlow, a high-fidelity PSF estimation framework that employs flow-guided edge localization for precise blur characterization. CircleFlow begins with a structured capture that encodes locally anisotropic and spatially varying PSFs by imaging a circle grid target, while leveraging the target's binary luminance prior to decouple image and kernel estimation. The latent sharp image is then reconstructed through subpixel alignment of an initialized binary structure guided by optical flow, whereas the PSF is modeled as an energy-constrained implicit neural representation. Both components are jointly optimized within a demosaicing-aware differentiable framework, ensuring physically consistent and robust PSF estimation enabled by accurate edge localization. Extensive experiments on simulated and real-world data demonstrate that CircleFlow achieves state-of-the-art accuracy and reliability, validating its effectiveness for practical PSF calibration.", "date": "2025-11-30", "categories": []}
{"id": "2601.06285", "title": "NAS-GS: Noise-Aware Sonar Gaussian Splatting", "abstract": "Underwater sonar imaging plays a crucial role in various applications, including autonomous navigation in murky water, marine archaeology, and environmental monitoring. However, the unique characteristics of sonar images, such as complex noise patterns and the lack of elevation information, pose significant challenges for 3D reconstruction and novel view synthesis. In this paper, we present NAS-GS, a novel Noise-Aware Sonar Gaussian Splatting framework specifically designed to address these challenges. Our approach introduces a Two-Ways Splatting technique that accurately models the dual directions for intensity accumulation and transmittance calculation inherent in sonar imaging, significantly improving rendering speed without sacrificing quality. Moreover, we propose a Gaussian Mixture Model (GMM) based noise model that captures complex sonar noise patterns, including side-lobes, speckle, and multi-path noise. This model enhances the realism of synthesized images while preventing 3D Gaussian overfitting to noise, thereby improving reconstruction accuracy. We demonstrate state-of-the-art performance on both simulated and real-world large-scale offshore sonar scenarios, achieving superior results in novel view synthesis and 3D reconstruction.", "date": "2026-01-09", "categories": []}
{"id": "2601.06475", "title": "VVTRec: Radio Interferometric Reconstruction through Visual and Textual Modality Enrichment", "abstract": "Radio astronomy is an indispensable discipline for observing distant celestial objects. Measurements of wave signals from radio telescopes, called visibility, need to be transformed into images for astronomical observations. These dirty images blend information from real sources and artifacts. Therefore, astronomers usually perform reconstruction before imaging to obtain cleaner images. Existing methods consider only a single modality of sparse visibility data, resulting in images with remaining artifacts and insufficient modeling of correlation. To enhance the extraction of visibility information and emphasize output quality in the image domain, we propose VVTRec, a multimodal radio interferometric data reconstruction method with visibility-guided visual and textual modality enrichment. In our VVTRec, sparse visibility is transformed into image-form and text-form features to obtain enhancements in terms of spatial and semantic information, improving the structural integrity and accuracy of images. Also, we leverage Vision-Language Models (VLMs) to achieve additional training-free performance improvements. VVTRec enables sparse visibility, as a foreign modality unseen by VLMs, to accurately extract pre-trained knowledge as a supplement. Our experiments demonstrate that VVTRec effectively enhances imaging results by exploiting multimodal information without introducing excessive computational overhead.", "date": "2026-01-10", "categories": []}
{"id": "2601.06525", "title": "Toward Generalizable Deblurring: Leveraging Massive Blur Priors with Linear Attention for Real-World Scenarios", "abstract": "Image deblurring has advanced rapidly with deep learning, yet most methods exhibit poor generalization beyond their training datasets, with performance dropping significantly in real-world scenarios. Our analysis shows this limitation stems from two factors: datasets face an inherent trade-off between realism and coverage of diverse blur patterns, and algorithmic designs remain restrictive, as pixel-wise losses drive models toward local detail recovery while overlooking structural and semantic consistency, whereas diffusion-based approaches, though perceptually strong, still fail to generalize when trained on narrow datasets with simplistic strategies. Through systematic investigation, we identify blur pattern diversity as the decisive factor for robust generalization and propose Blur Pattern Pretraining (BPP), which acquires blur priors from simulation datasets and transfers them through joint fine-tuning on real data. We further introduce Motion and Semantic Guidance (MoSeG) to strengthen blur priors under severe degradation, and integrate it into GLOWDeblur, a Generalizable reaL-wOrld lightWeight Deblur model that combines convolution-based pre-reconstruction & domain alignment module with a lightweight diffusion backbone. Extensive experiments on six widely-used benchmarks and two real-world datasets validate our approach, confirming the importance of blur priors for robust generalization and demonstrating that the lightweight design of GLOWDeblur ensures practicality in real-world applications. The project page is available at https://vegdog007.github.io/GLOWDeblur_Website/.", "date": "2026-01-10", "categories": []}
{"id": "2601.07356", "title": "Efficient Convolutional Forward Model for Passive Acoustic Mapping and Temporal Monitoring", "abstract": "Passive acoustic mapping (PAM) is a key imaging technique for characterizing cavitation activity in therapeutic ultrasound applications. Recent model-based beamforming algorithms offer high reconstruction quality and strong physical interpretability. However, their computational burden and limited temporal resolution restrict their use in applications with time-evolving cavitation. To address these challenges, we introduce a PAM beamforming framework based on a novel convolutional formulation in the time domain, which enables efficient computation. In this framework, PAM is formulated as an inverse problem in which the forward operator maps spatiotemporal cavitation activity to recorded radio-frequency signals accounting for time-of-flight delays defined by the acquisition geometry. We then formulate a regularized inversion algorithm that incorporates prior knowledge on cavitation activity. Experimental results demonstrate that our framework outperforms classical beamforming methods, providing higher temporal resolution than frequency-domain techniques while substantially reducing computational burden compared with iterative time-domain formulations.", "date": "2026-01-12", "categories": []}
{"id": "2412.00065", "title": "DYRECT Computed Tomography: DYnamic Reconstruction of Events on a Continuous Timescale", "abstract": "Time-resolved high-resolution X-ray Computed Tomography (4D $\\mu$CT) is an imaging technique that offers insight into the evolution of dynamic processes inside materials that are opaque to visible light. Conventional tomographic reconstruction techniques are based on recording a sequence of 3D images that represent the sample state at different moments in time. This frame-based approach limits the temporal resolution compared to dynamic radiography experiments due to the time needed to make CT scans. Moreover, it leads to an inflation of the amount of data and thus to costly post-processing computations to quantify the dynamic behaviour from the sequence of time frames, hereby often ignoring the temporal correlations of the sample structure. Our proposed 4D $\\mu$CT reconstruction technique, named DYRECT, estimates individual attenuation evolution profiles for each position in the sample. This leads to a novel memory-efficient event-based representation of the sample, using as little as three image volumes: its initial attenuation, its final attenuation and the transition times. This third volume represents local events on a continuous timescale instead of the discrete global time frames. We propose a method to iteratively reconstruct the transition times and the attenuation volumes. The dynamic reconstruction technique was validated on synthetic ground truth data and experimental data, and was found to effectively pinpoint the transition times in the synthetic dataset with a time resolution corresponding to less than a tenth of the amount of projections required to reconstruct traditional $\\mu$CT time frames.", "date": "2024-11-15", "categories": []}
{"id": "2503.16222", "title": "Efficient Bayesian Computation Using Plug-and-Play Priors for Poisson Inverse Problems", "abstract": "This paper studies plug-and-play (PnP) Langevin sampling strategies for Bayesian inference in low-photon Poisson imaging problems, a challenging class of problems with significant applications in astronomy, medicine, and biology. PnP Langevin sampling offers a powerful framework for Bayesian image restoration, enabling accurate point estimation as well as advanced inference tasks, including uncertainty quantification and visualization analyses, and empirical Bayesian inference for automatic model parameter tuning. Herein, we leverage and adapt recent developments in this framework to tackle challenging imaging problems involving weakly informative Poisson data. Existing PnP Langevin algorithms are not well-suited for low-photon Poisson imaging due to high solution uncertainty and poor regularity properties, such as exploding gradients and non-negativity constraints. To address these challenges, we explore two strategies for extending Langevin PnP sampling to Poisson imaging models: (i) an accelerated PnP Langevin method that incorporates boundary reflections and a Poisson likelihood approximation and (ii) a mirror sampling algorithm that leverages a Riemannian geometry to handle the constraints and the poor regularity of the likelihood without approximations. The effectiveness of these approaches is evaluated and contrasted through extensive numerical experiments and comparisons with state-of-the-art methods. The source code accompanying this paper is available at https://github.com/freyyia/pnp-langevin-poisson.", "date": "2025-03-20", "categories": []}
{"id": "2509.25104", "title": "Towards generalizable deep ptychography neural networks", "abstract": "X-ray ptychography is a data-intensive imaging technique expected to become ubiquitous at next-generation light sources delivering many-fold increases in coherent flux. The need for real-time feedback under accelerated acquisition rates motivates surrogate reconstruction models like deep neural networks, which offer orders-of-magnitude speedup over conventional methods. However, existing deep learning approaches lack robustness across diverse experimental conditions. We propose an unsupervised training workflow emphasizing probe learning by combining experimentally-measured probes with synthetic, procedurally generated objects. This probe-centric approach enables a single physics-informed neural network to reconstruct unseen experiments across multiple beamlines; among the first demonstrations of multi-probe generalization. We find probe learning is equally important as in-distribution learning; models trained using this synthetic workflow achieve reconstruction fidelity comparable to those trained exclusively on experimental data, even when changing the type of synthetic training object. The proposed approach enables training of experiment-steering models that provide real-time feedback under dynamic experimental conditions.", "date": "2025-09-29", "categories": []}
{"id": "2601.04779", "title": "Defocus Aberration Theory Confirms Gaussian Model in Most Imaging Devices", "abstract": "Over the past three decades, defocus has consistently provided groundbreaking depth information in scene images. However, accurately estimating depth from 2D images continues to be a persistent and fundamental challenge in the field of 3D recovery. Heuristic approaches involve with the ill-posed problem for inferring the spatial variant defocusing blur, as the desired blur cannot be distinguished from the inherent blur. Given a prior knowledge of the defocus model, the problem become well-posed with an analytic solution for the relative blur between two images, taken at the same viewpoint with different camera settings for the focus. The Gaussian model stands out as an optimal choice for real-time applications, due to its mathematical simplicity and computational efficiency. And theoretically, it is the only model can be applied at the same time to both the absolute blur caused by depth in a single image and the relative blur resulting from depth differences between two images. This paper introduces the settings, for conventional imaging devices, to ensure that the defocusing operator adheres to the Gaussian model. Defocus analysis begins within the framework of geometric optics and is conducted by defocus aberration theory in diffraction-limited optics to obtain the accuracy of fitting the actual model to its Gaussian approximation. The results for a typical set of focused depths between $1$ and $100$ meters, with a maximum depth variation of $10\\%$ at the focused depth, confirm the Gaussian model's applicability for defocus operators in most imaging devices. The findings demonstrate a maximum Mean Absolute Error $(\\!M\\!A\\!E)$ of less than $1\\%$, underscoring the model's accuracy and reliability.", "date": "2026-01-08", "categories": []}
{"id": "2601.03526", "title": "Physics-Constrained Cross-Resolution Enhancement Network for Optics-Guided Thermal UAV Image Super-Resolution", "abstract": "Optics-guided thermal UAV image super-resolution has attracted significant research interest due to its potential in all-weather monitoring applications. However, existing methods typically compress optical features to match thermal feature dimensions for cross-modal alignment and fusion, which not only causes the loss of high-frequency information that is beneficial for thermal super-resolution, but also introduces physically inconsistent artifacts such as texture distortions and edge blurring by overlooking differences in the imaging physics between modalities. To address these challenges, we propose PCNet to achieve cross-resolution mutual enhancement between optical and thermal modalities, while physically constraining the optical guidance process via thermal conduction to enable robust thermal UAV image super-resolution. In particular, we design a Cross-Resolution Mutual Enhancement Module (CRME) to jointly optimize thermal image super-resolution and optical-to-thermal modality conversion, facilitating effective bidirectional feature interaction across resolutions while preserving high-frequency optical priors. Moreover, we propose a Physics-Driven Thermal Conduction Module (PDTM) that incorporates two-dimensional heat conduction into optical guidance, modeling spatially-varying heat conduction properties to prevent inconsistent artifacts. In addition, we introduce a temperature consistency loss that enforces regional distribution consistency and boundary gradient smoothness to ensure generated thermal images align with real-world thermal radiation principles. Extensive experiments on VGTSR2.0 and DroneVehicle datasets demonstrate that PCNet significantly outperforms state-of-the-art methods on both reconstruction quality and downstream tasks including semantic segmentation and object detection.", "date": "2026-01-07", "categories": []}
{"id": "2504.13177", "title": "Spatial Polarization Multiplexing: Single-Shot Invisible Shape and Reflectance Recovery", "abstract": "We propose spatial polarization multiplexing (SPM) for joint sensing of shape and reflectance of a static or dynamic deformable object, which is also invisible to the naked eye. Past structured-light methods are limited to shape acquisition and cannot recover reflectance as they alter scene appearance. Our key idea is to spatially multiplex a polarization pattern to encode the incident ray and also densely sample the reflected light. We derive a quantized polarized light pattern that can be robustly and uniquely decoded from the reflected Angle of Linear Polarization (AoLP) values. It also enables single-shot disentanglement of polarimetric diffuse and specular reflections for accurate BRDF estimation. We achieve this spatial polarization multiplexing (SPM) with a constrained de Bruijn sequence. We validate this novel invisible single-shot shape and reflectance method with real static and dynamic objects. The results demonstrate the effectiveness of SPM for accurate shape and BRDF measurement which opens new avenues of application for 3D sensing thanks to its invisibility and ability to jointly recover the radiometric properties.", "date": "2025-04-17", "categories": []}
{"id": "2601.00551", "title": "SlingBAG Pro: Accelerating point cloud-based iterative reconstruction for 3D photoacoustic imaging with arbitrary array geometries", "abstract": "High-quality three-dimensional (3D) photoacoustic imaging (PAI) is gaining increasing attention in clinical applications. To address the challenges of limited space and high costs, irregular geometric transducer arrays that conform to specific imaging regions are promising for achieving high-quality 3D PAI with fewer transducers. However, traditional iterative reconstruction algorithms struggle with irregular array configurations, suffering from high computational complexity, substantial memory requirements, and lengthy reconstruction times. In this work, we introduce SlingBAG Pro, an advanced reconstruction algorithm based on the point cloud iteration concept of the Sliding ball adaptive growth (SlingBAG) method, while extending its compatibility to arbitrary array geometries. SlingBAG Pro maintains high reconstruction quality, reduces the number of required transducers, and employs a hierarchical optimization strategy that combines zero-gradient filtering with progressively increased temporal sampling rates during iteration. This strategy rapidly removes redundant spatial point clouds, accelerates convergence, and significantly shortens overall reconstruction time. Compared to the original SlingBAG algorithm, SlingBAG Pro achieves up to a 2.2-fold speed improvement in point cloud-based 3D PA reconstruction under irregular array geometries. The proposed method is validated through both simulation and in vivo mouse experiments, and the source code is publicly available at https://github.com/JaegerCQ/SlingBAG_Pro.", "date": "2026-01-02", "categories": []}
{"id": "2601.02594", "title": "Annealed Langevin Posterior Sampling (ALPS): A Rapid Algorithm for Image Restoration with Multiscale Energy Models", "abstract": "Solving inverse problems in imaging requires models that support efficient inference, uncertainty quantification, and principled probabilistic reasoning. Energy-Based Models (EBMs), with their interpretable energy landscapes and compositional structure, are well-suited for this task but have historically suffered from high computational costs and training instability. To overcome the historical shortcomings of EBMs, we introduce a fast distillation strategy to transfer the strengths of pre-trained diffusion models into multi-scale EBMs. These distilled EBMs enable efficient sampling and preserve the interpretability and compositionality inherent to potential-based frameworks. Leveraging EBM compositionality, we propose Annealed Langevin Posterior Sampling (ALPS) algorithm for Maximum-A-Posteriori (MAP), Minimum Mean Square Error (MMSE), and uncertainty estimates for inverse problems in imaging. Unlike diffusion models that use complex guidance strategies for latent variables, we perform annealing on static posterior distributions that are well-defined and composable. Experiments on image inpainting and MRI reconstruction demonstrate that our method matches or surpasses diffusion-based baselines in both accuracy and efficiency, while also supporting MAP recovery. Overall, our framework offers a scalable and principled solution for inverse problems in imaging, with potential for practical deployment in scientific and clinical settings. ALPS code is available at the GitHub repository \\href{https://github.com/JyoChand/ALPS}{ALPS}.", "date": "2026-01-05", "categories": []}
{"id": "2601.01228", "title": "HyDRA: Hybrid Denoising Regularization for Measurement-Only DEQ Training", "abstract": "Solving image reconstruction problems of the form \\(\\mathbf{A} \\mathbf{x} = \\mathbf{y}\\) remains challenging due to ill-posedness and the lack of large-scale supervised datasets. Deep Equilibrium (DEQ) models have been used successfully but typically require supervised pairs \\((\\mathbf{x},\\mathbf{y})\\). In many practical settings, only measurements \\(\\mathbf{y}\\) are available. We introduce HyDRA (Hybrid Denoising Regularization Adaptation), a measurement-only framework for DEQ training that combines measurement consistency with an adaptive denoising regularization term, together with a data-driven early stopping criterion. Experiments on sparse-view CT demonstrate competitive reconstruction quality and fast inference.", "date": "2026-01-03", "categories": []}
{"id": "2601.01541", "title": "Sim2Real SAR Image Restoration: Metadata-Driven Models for Joint Despeckling and Sidelobes Reduction", "abstract": "Synthetic aperture radar (SAR) provides valuable information about the Earth's surface under all weather and illumination conditions. However, the inherent phenomenon of speckle and the presence of sidelobes around bright targets pose challenges for accurate interpretation of SAR imagery. Most existing SAR image restoration methods address despeckling and sidelobes reduction as separate tasks. In this paper, we propose a unified framework that jointly performs both tasks using neural networks (NNs) trained on a realistic SAR simulated dataset generated with MOCEM. Inference can then be performed on real SAR images, demonstrating effective simulation to real (Sim2Real) transferability. Additionally, we incorporate acquisition metadata as auxiliary input to the NNs, demonstrating improved restoration performance.", "date": "2026-01-04", "categories": []}
{"id": "2601.01998", "title": "Nighttime Hazy Image Enhancement via Progressively and Mutually Reinforcing Night-Haze Priors", "abstract": "Enhancing the visibility of nighttime hazy images is challenging due to the complex degradation distributions. Existing methods mainly address a single type of degradation (e.g., haze or low-light) at a time, ignoring the interplay of different degradation types and resulting in limited visibility improvement. We observe that the domain knowledge shared between low-light and haze priors can be reinforced mutually for better visibility. Based on this key insight, in this paper, we propose a novel framework that enhances visibility in nighttime hazy images by reinforcing the intrinsic consistency between haze and low-light priors mutually and progressively. In particular, our model utilizes image-, patch-, and pixel-level experts that operate across visual and frequency domains to recover global scene structure, regional patterns, and fine-grained details progressively. A frequency-aware router is further introduced to adaptively guide the contribution of each expert, ensuring robust image restoration. Extensive experiments demonstrate the superior performance of our model on nighttime dehazing benchmarks both quantitatively and qualitatively. Moreover, we showcase the generalizability of our model in daytime dehazing and low-light enhancement tasks.", "date": "2026-01-05", "categories": []}
{"id": "2601.02141", "title": "Efficient Unrolled Networks for Large-Scale 3D Inverse Problems", "abstract": "Deep learning-based methods have revolutionized the field of imaging inverse problems, yielding state-of-the-art performance across various imaging domains. The best performing networks incorporate the imaging operator within the network architecture, typically in the form of deep unrolling. However, in large-scale problems, such as 3D imaging, most existing methods fail to incorporate the operator in the architecture due to the prohibitive amount of memory required by global forward operators, which hinder typical patching strategies. In this work, we present a domain partitioning strategy and normal operator approximations that enable the training of end-to-end reconstruction models incorporating forward operators of arbitrarily large problems into their architecture. The proposed method achieves state-of-the-art performance on 3D X-ray cone-beam tomography and 3D multi-coil accelerated MRI, while requiring only a single GPU for both training and inference.", "date": "2026-01-05", "categories": []}
{"id": "2601.02206", "title": "Seeing the Unseen: Zooming in the Dark with Event Cameras", "abstract": "This paper addresses low-light video super-resolution (LVSR), aiming to restore high-resolution videos from low-light, low-resolution (LR) inputs. Existing LVSR methods often struggle to recover fine details due to limited contrast and insufficient high-frequency information. To overcome these challenges, we present RetinexEVSR, the first event-driven LVSR framework that leverages high-contrast event signals and Retinex-inspired priors to enhance video quality under low-light scenarios. Unlike previous approaches that directly fuse degraded signals, RetinexEVSR introduces a novel bidirectional cross-modal fusion strategy to extract and integrate meaningful cues from noisy event data and degraded RGB frames. Specifically, an illumination-guided event enhancement module is designed to progressively refine event features using illumination maps derived from the Retinex model, thereby suppressing low-light artifacts while preserving high-contrast details. Furthermore, we propose an event-guided reflectance enhancement module that utilizes the enhanced event features to dynamically recover reflectance details via a multi-scale fusion mechanism. Experimental results show that our RetinexEVSR achieves state-of-the-art performance on three datasets. Notably, on the SDSD benchmark, our method can get up to 2.95 dB gain while reducing runtime by 65% compared to prior event-based methods. Code: https://github.com/DachunKai/RetinexEVSR.", "date": "2026-01-05", "categories": []}
{"id": "2601.00322", "title": "Depth-Synergized Mamba Meets Memory Experts for All-Day Image Reflection Separation", "abstract": "Image reflection separation aims to disentangle the transmission layer and the reflection layer from a blended image. Existing methods rely on limited information from a single image, tending to confuse the two layers when their contrasts are similar, a challenge more severe at night. To address this issue, we propose the Depth-Memory Decoupling Network (DMDNet). It employs the Depth-Aware Scanning (DAScan) to guide Mamba toward salient structures, promoting information flow along semantic coherence to construct stable states. Working in synergy with DAScan, the Depth-Synergized State-Space Model (DS-SSM) modulates the sensitivity of state activations by depth, suppressing the spread of ambiguous features that interfere with layer disentanglement. Furthermore, we introduce the Memory Expert Compensation Module (MECM), leveraging cross-image historical knowledge to guide experts in providing layer-specific compensation. To address the lack of datasets for nighttime reflection separation, we construct the Nighttime Image Reflection Separation (NightIRS) dataset. Extensive experiments demonstrate that DMDNet outperforms state-of-the-art methods in both daytime and nighttime.", "date": "2026-01-01", "categories": []}
{"id": "2503.06898", "title": "Illuminating Darkness: Learning to Enhance Low-light Images In-the-Wild", "abstract": "Single-shot low-light image enhancement (SLLIE) remains challenging due to the limited availability of diverse, real-world paired datasets. To bridge this gap, we introduce the Low-Light Smartphone Dataset (LSD), a large-scale, high-resolution (4K+) dataset collected in the wild across a wide range of challenging lighting conditions (0.1 to 200 lux). LSD contains 6,425 precisely aligned low and normal-light image pairs, selected from over 8,000 dynamic indoor and outdoor scenes through multi-frame acquisition and expert evaluation. To evaluate generalization and aesthetic quality, we collect 2,117 unpaired low-light images from previously unseen devices. To fully exploit LSD, we propose TFFormer, a hybrid model that encodes luminance and chrominance (LC) separately to reduce color-structure entanglement. We further propose a cross-attention-driven joint decoder for context-aware fusion of LC representations, along with LC refinement and LC-guided supervision to significantly enhance perceptual fidelity and structural consistency. TFFormer achieves state-of-the-art results on LSD (+2.45 dB PSNR) and substantially improves downstream vision tasks, such as low-light object detection (+6.80 mAP on ExDark).", "date": "2025-03-10", "categories": []}
{"id": "2512.23667", "title": "IDT: A Physically Grounded Transformer for Feed-Forward Multi-View Intrinsic Decomposition", "abstract": "Intrinsic image decomposition is fundamental for visual understanding, as RGB images entangle material properties, illumination, and view-dependent effects. Recent diffusion-based methods have achieved strong results for single-view intrinsic decomposition; however, extending these approaches to multi-view settings remains challenging, often leading to severe view inconsistency. We propose \\textbf{Intrinsic Decomposition Transformer (IDT)}, a feed-forward framework for multi-view intrinsic image decomposition. By leveraging transformer-based attention to jointly reason over multiple input images, IDT produces view-consistent intrinsic factors in a single forward pass, without iterative generative sampling. IDT adopts a physically grounded image formation model that explicitly decomposes images into diffuse reflectance, diffuse shading, and specular shading. This structured factorization separates Lambertian and non-Lambertian light transport, enabling interpretable and controllable decomposition of material and illumination effects across views. Experiments on both synthetic and real-world datasets demonstrate that IDT achieves cleaner diffuse reflectance, more coherent diffuse shading, and better-isolated specular components, while substantially improving multi-view consistency compared to prior intrinsic decomposition methods.", "date": "2025-12-29", "categories": []}
{"id": "2512.24674", "title": "An Adaptive, Disentangled Representation for Multidimensional MRI Reconstruction", "abstract": "We present a new approach for representing and reconstructing multidimensional magnetic resonance imaging (MRI) data. Our method builds on a novel, learned feature-based image representation that disentangles different types of features, such as geometry and contrast, into distinct low-dimensional latent spaces, enabling better exploitation of feature correlations in multidimensional images and incorporation of pre-learned priors specific to different feature types for reconstruction. More specifically, the disentanglement was achieved via an encoderdecoder network and image transfer training using large public data, enhanced by a style-based decoder design. A latent diffusion model was introduced to impose stronger constraints on distinct feature spaces. New reconstruction formulations and algorithms were developed to integrate the learned representation with a zero-shot selfsupervised learning adaptation and subspace modeling. The proposed method has been evaluated on accelerated T1 and T2 parameter mapping, achieving improved performance over state-of-the-art reconstruction methods, without task-specific supervised training or fine-tuning. This work offers a new strategy for learning-based multidimensional image reconstruction where only limited data are available for problem-specific or task-specific training.", "date": "2025-12-31", "categories": []}
{"id": "2401.12004", "title": "NLCG-Net: A Model-Based Zero-Shot Learning Framework for Undersampled Quantitative MRI Reconstruction", "abstract": "Typical quantitative MRI (qMRI) methods estimate parameter maps in a two-step pipeline that first reconstructs images from undersampled k-space data and then performs model fitting, which is prone to biases and error propagation. We propose NLCG-Net, a model-based nonlinear conjugate gradient (NLCG) framework for joint T2/T1 estimation that incorporates a U-Net regularizer trained in a scan-specific, zero-shot fashion. The method directly estimates qMRI maps from undersampled k-space using mono-exponential signal modeling with scan-specific neural network regularization, enabling high-fidelity T1 and T2 mapping. Experimental results on T2 and T1 mapping demonstrate that NLCG-Net improves estimation quality over subspace reconstruction at high acceleration factors.", "date": "2024-01-22", "categories": []}
{"id": "2501.03737", "title": "Re-Visible Dual-Domain Self-Supervised Deep Unfolding Network for MRI Reconstruction", "abstract": "Magnetic Resonance Imaging (MRI) is widely used in clinical practice, but suffered from prolonged acquisition time. Although deep learning methods have been proposed to accelerate acquisition and demonstrate promising performance, they rely on high-quality fully-sampled datasets for training in a supervised manner. However, such datasets are time-consuming and expensive-to-collect, which constrains their broader applications. On the other hand, self-supervised methods offer an alternative by enabling learning from under-sampled data alone, but most existing methods rely on further partitioned under-sampled k-space data as model's input for training, resulting in a loss of valuable information. Additionally, their models have not fully incorporated image priors, leading to degraded reconstruction performance. In this paper, we propose a novel re-visible dual-domain self-supervised deep unfolding network to address these issues when only under-sampled datasets are available. Specifically, by incorporating re-visible dual-domain loss, all under-sampled k-space data are utilized during training to mitigate information loss caused by further partitioning. This design enables the model to implicitly adapt to all under-sampled k-space data as input. Additionally, we design a deep unfolding network based on Chambolle and Pock Proximal Point Algorithm (DUN-CP-PPA) to achieve end-to-end reconstruction, incorporating imaging physics and image priors to guide the reconstruction process. By employing a Spatial-Frequency Feature Extraction (SFFE) block to capture global and local feature representation, we enhance the model's efficiency to learn comprehensive image priors. Experiments conducted on the fastMRI and IXI datasets demonstrate that our method significantly outperforms state-of-the-art approaches in terms of reconstruction performance.", "date": "2025-01-07", "categories": []}
{"id": "2503.09131", "title": "MP-HSIR: A Multi-Prompt Framework for Universal Hyperspectral Image Restoration", "abstract": "Hyperspectral images (HSIs) often suffer from diverse and unknown degradations during imaging, leading to severe spectral and spatial distortions. Existing HSI restoration methods typically rely on specific degradation assumptions, limiting their effectiveness in complex scenarios. In this paper, we propose \\textbf{MP-HSIR}, a novel multi-prompt framework that effectively integrates spectral, textual, and visual prompts to achieve universal HSI restoration across diverse degradation types and intensities. Specifically, we develop a prompt-guided spatial-spectral transformer, which incorporates spatial self-attention and a prompt-guided dual-branch spectral self-attention. Since degradations affect spectral features differently, we introduce spectral prompts in the local spectral branch to provide universal low-rank spectral patterns as prior knowledge for enhancing spectral reconstruction. Furthermore, the text-visual synergistic prompt fuses high-level semantic representations with fine-grained visual features to encode degradation information, thereby guiding the restoration process. Extensive experiments on 9 HSI restoration tasks, including all-in-one scenarios, generalization tests, and real-world cases, demonstrate that MP-HSIR not only consistently outperforms existing all-in-one methods but also surpasses state-of-the-art task-specific approaches across multiple tasks. The code and models are available at https://github.com/ZhehuiWu/MP-HSIR.", "date": "2025-03-12", "categories": []}
{"id": "2512.22202", "title": "Complex Swin Transformer for Accelerating Enhanced SMWI Reconstruction", "abstract": "Susceptibility Map Weighted Imaging (SMWI) is an advanced magnetic resonance imaging technique used to detect nigral hyperintensity in Parkinsons disease. However, full resolution SMWI acquisition is limited by long scan times. Efficient reconstruction methods are therefore required to generate high quality SMWI from reduced k space data while preserving diagnostic relevance. In this work, we propose a complex valued Swin Transformer based network for super resolution reconstruction of multi echo MRI data. The proposed method reconstructs high quality SMWI images from low resolution k space inputs. Experimental results demonstrate that the method achieves a structural similarity index of 0.9116 and a mean squared error of 0.076 when reconstructing SMWI from 256 by 256 k space data, while maintaining critical diagnostic features. This approach enables high quality SMWI reconstruction from reduced k space sampling, leading to shorter scan times without compromising diagnostic detail. The proposed method has the potential to improve the clinical applicability of SMWI for Parkinsons disease and support faster and more efficient neuroimaging workflows.", "date": "2025-12-21", "categories": []}
{"id": "2512.22822", "title": "KANO: Kolmogorov-Arnold Neural Operator for Image Super-Resolution", "abstract": "The highly nonlinear degradation process, complex physical interactions, and various sources of uncertainty render single-image Super-resolution (SR) a particularly challenging task. Existing interpretable SR approaches, whether based on prior learning or deep unfolding optimization frameworks, typically rely on black-box deep networks to model latent variables, which leaves the degradation process largely unknown and uncontrollable. Inspired by the Kolmogorov-Arnold theorem (KAT), we for the first time propose a novel interpretable operator, termed Kolmogorov-Arnold Neural Operator (KANO), with the application to image SR. KANO provides a transparent and structured representation of the latent degradation fitting process. Specifically, we employ an additive structure composed of a finite number of B-spline functions to approximate continuous spectral curves in a piecewise fashion. By learning and optimizing the shape parameters of these spline functions within defined intervals, our KANO accurately captures key spectral characteristics, such as local linear trends and the peak-valley structures at nonlinear inflection points, thereby endowing SR results with physical interpretability. Furthermore, through theoretical modeling and experimental evaluations across natural images, aerial photographs, and satellite remote sensing data, we systematically compare multilayer perceptrons (MLPs) and Kolmogorov-Arnold networks (KANs) in handling complex sequence fitting tasks. This comparative study elucidates the respective advantages and limitations of these models in characterizing intricate degradation mechanisms, offering valuable insights for the development of interpretable SR techniques.", "date": "2025-12-28", "categories": []}
{"id": "2512.23375", "title": "Diffusion priors enhanced velocity model building from time-lag images using a neural operator", "abstract": "Velocity model building serves as a crucial component for achieving high precision subsurface imaging. However, conventional velocity model building methods are often computationally expensive and time consuming. In recent years, with the rapid advancement of deep learning, particularly the success of generative models and neural operators, deep learning based approaches that integrate data and their statistics have attracted increasing attention in addressing the limitations of traditional methods. In this study, we propose a novel framework that combines generative models with neural operators to obtain high resolution velocity models efficiently. Within this workflow, the neural operator functions as a forward mapping operator to rapidly generate time lag reverse time migration (RTM) extended images from the true and migration velocity models. In this framework, the neural operator is acting as a surrogate for modeling followed by migration, which uses the true and migration velocities, respectively. The trained neural operator is then employed, through automatic differentiation, to gradually update the migration velocity placed in the true velocity input channel with high resolution components so that the output of the network matches the time lag images of observed data obtained using the migration velocity. By embedding a generative model, trained on a high-resolution velocity model distribution, which corresponds to the true velocity model distribution used to train the neural operator, as a regularizer, the resulting predictions are cleaner with higher resolution information. Both synthetic and field data experiments demonstrate the effectiveness of the proposed generative neural operator based velocity model building approach.", "date": "2025-12-29", "categories": []}
{"id": "2509.16509", "title": "SlowFast-SCI: Slow-Fast Deep Unfolding Learning for Spectral Compressive Imaging", "abstract": "Humans learn in two complementary ways: a slow, cumulative process that builds broad, general knowledge, and a fast, on-the-fly process that captures specific experiences. Existing deep-unfolding methods for spectral compressive imaging (SCI) mirror only the slow component-relying on heavy pre-training with many unfolding stages-yet they lack the rapid adaptation needed to handle new optical configurations. As a result, they falter on out-of-distribution cameras, especially in bespoke spectral setups unseen during training. This depth also incurs heavy computation and slow inference. To bridge this gap, we introduce SlowFast-SCI, a dual-speed framework seamlessly integrated into any deep unfolding network beyond SCI systems. During slow learning, we pre-train or reuse a priors-based backbone and distill it via imaging guidance into a compact fast-unfolding model. In the fast learning stage, lightweight adaptation modules are embedded within each block and trained self-supervised at test time via a dual-domain loss-without retraining the backbone. To the best of our knowledge, SlowFast-SCI is the first test-time adaptation-driven deep unfolding framework for efficient, self-adaptive spectral reconstruction. Its dual-stage design unites offline robustness with on-the-fly per-sample calibration-yielding over 70% reduction in parameters and FLOPs, up to 5.79 dB PSNR improvement on out-of-distribution data, preserved cross-domain adaptability, and a 4x faster adaptation speed. In addition, its modularity integrates with any deep-unfolding network, paving the way for self-adaptive, field-deployable imaging and expanded computational imaging modalities. The models, datasets, and code are available at https://github.com/XuanLu11/SlowFast-SCI.", "date": "2025-09-20", "categories": []}
{"id": "2509.17792", "title": "Degradation-Aware All-in-One Image Restoration via Latent Prior Encoding", "abstract": "Real-world images often suffer from spatially diverse degradations such as haze, rain, snow, and low-light, significantly impacting visual quality and downstream vision tasks. Existing all-in-one restoration (AIR) approaches either depend on external text prompts or embed hand-crafted architectural priors (e.g., frequency heuristics); both impose discrete, brittle assumptions that weaken generalization to unseen or mixed degradations. To address this limitation, we propose to reframe AIR as learned latent prior inference, where degradation-aware representations are automatically inferred from the input without explicit task cues. Based on latent priors, we formulate AIR as a structured reasoning paradigm: (1) which features to route (adaptive feature selection), (2) where to restore (spatial localization), and (3) what to restore (degradation semantics). We design a lightweight decoding module that efficiently leverages these latent encoded cues for spatially-adaptive restoration. Extensive experiments across six common degradation tasks, five compound settings, and previously unseen degradations demonstrate that our method outperforms state-of-the-art (SOTA) approaches, achieving an average PSNR improvement of 1.68 dB while being three times more efficient.", "date": "2025-09-22", "categories": []}
{"id": "2512.21513", "title": "MuS-Polar3D: A Benchmark Dataset for Computational Polarimetric 3D Imaging under Multi-Scattering Conditions", "abstract": "Polarization-based underwater 3D imaging exploits polarization cues to suppress background scattering, exhibiting distinct advantages in turbid water. Although data-driven polarization-based underwater 3D reconstruction methods show great potential, existing public datasets lack sufficient diversity in scattering and observation conditions, hindering fair comparisons among different approaches, including single-view and multi-view polarization imaging methods.\n  To address this limitation, we construct MuS-Polar3D, a benchmark dataset comprising polarization images of 42 objects captured under seven quantitatively controlled scattering conditions and five viewpoints, together with high-precision 3D models (+/- 0.05 mm accuracy), normal maps, and foreground masks. The dataset supports multiple vision tasks, including normal estimation, object segmentation, descattering, and 3D reconstruction.\n  Inspired by computational imaging, we further decouple underwater 3D reconstruction under scattering into a two-stage pipeline, namely descattering followed by 3D reconstruction, from an imaging-chain perspective. Extensive evaluations using multiple baseline methods under complex scattering conditions demonstrate the effectiveness of the proposed benchmark, achieving a best mean angular error of 15.49 degrees. To the best of our knowledge, MuS-Polar3D is the first publicly available benchmark dataset for quantitative turbidity underwater polarization-based 3D imaging, enabling accurate reconstruction and fair algorithm evaluation under controllable scattering conditions. The dataset and code are publicly available at https://github.com/WangPuyun/MuS-Polar3D.", "date": "2025-12-25", "categories": []}
{"id": "2512.21975", "title": "RT-Focuser: A Real-Time Lightweight Model for Edge-side Image Deblurring", "abstract": "Motion blur caused by camera or object movement severely degrades image quality and poses challenges for real-time applications such as autonomous driving, UAV perception, and medical imaging. In this paper, a lightweight U-shaped network tailored for real-time deblurring is presented and named RT-Focuser. To balance speed and accuracy, we design three key components: Lightweight Deblurring Block (LD) for edge-aware feature extraction, Multi-Level Integrated Aggregation module (MLIA) for encoder integration, and Cross-source Fusion Block (X-Fuse) for progressive decoder refinement. Trained on a single blurred input, RT-Focuser achieves 30.67 dB PSNR with only 5.85M parameters and 15.76 GMACs. It runs 6ms per frame on GPU and mobile, exceeds 140 FPS on both, showing strong potential for deployment on the edge. The official code and usage are available on: https://github.com/ReaganWu/RT-Focuser.", "date": "2025-12-26", "categories": []}
{"id": "2512.21040", "title": "A Large-Depth-Range Layer-Based Hologram Dataset for Machine Learning-Based 3D Computer-Generated Holography", "abstract": "Machine learning-based computer-generated holography (ML-CGH) has advanced rapidly in recent years, yet progress is constrained by the limited availability of high-quality, large-scale hologram datasets. To address this, we present KOREATECH-CGH, a publicly available dataset comprising 6,000 pairs of RGB-D images and complex holograms across resolutions ranging from 256*256 to 2048*2048, with depth ranges extending to the theoretical limits of the angular spectrum method for wide 3D scene coverage. To improve hologram quality at large depth ranges, we introduce amplitude projection, a post-processing technique that replaces amplitude components of hologram wavefields at each depth layer while preserving phase. This approach enhances reconstruction fidelity, achieving 27.01 dB PSNR and 0.87 SSIM, surpassing a recent optimized silhouette-masking layer-based method by 2.03 dB and 0.04 SSIM, respectively. We further validate the utility of KOREATECH-CGH through experiments on hologram generation and super-resolution using state-of-the-art ML models, confirming its applicability for training and evaluating next-generation ML-CGH systems.", "date": "2025-12-24", "categories": []}
{"id": "2512.21180", "title": "Equivariant Multiscale Learned Invertible Reconstruction for Cone Beam CT: From Simulated to Real Data", "abstract": "Cone Beam CT (CBCT) is an important imaging modality nowadays, however lower image quality of CBCT compared to more conventional Computed Tomography (CT) remains a limiting factor in CBCT applications. Deep learning reconstruction methods are a promising alternative to classical analytical and iterative reconstruction methods, but applying such methods to CBCT is often difficult due to the lack of ground truth data, memory limitations and the need for fast inference at clinically-relevant resolutions. In this work we propose LIRE++, an end-to-end rotationally-equivariant multiscale learned invertible primal-dual scheme for fast and memory-efficient CBCT reconstruction. Memory optimizations and multiscale reconstruction allow for fast training and inference, while rotational equivariance improves parameter efficiency. LIRE++ was trained on simulated projection data from a fast quasi-Monte Carlo CBCT projection simulator that we developed as well. Evaluated on synthetic data, LIRE++ gave an average improvement of 1 dB in Peak Signal-to-Noise Ratio over alternative deep learning baselines. On real clinical data, LIRE++ improved the average Mean Absolute Error between the reconstruction and the corresponding planning CT by 10 Hounsfield Units with respect to current proprietary state-of-the-art hybrid deep-learning/iterative method.", "date": "2025-12-24", "categories": []}
{"id": "2512.19817", "title": "Generating the Past, Present and Future from a Motion-Blurred Image", "abstract": "We seek to answer the question: what can a motion-blurred image reveal about a scene's past, present, and future? Although motion blur obscures image details and degrades visual quality, it also encodes information about scene and camera motion during an exposure. Previous techniques leverage this information to estimate a sharp image from an input blurry one, or to predict a sequence of video frames showing what might have occurred at the moment of image capture. However, they rely on handcrafted priors or network architectures to resolve ambiguities in this inverse problem, and do not incorporate image and video priors on large-scale datasets. As such, existing methods struggle to reproduce complex scene dynamics and do not attempt to recover what occurred before or after an image was taken. Here, we introduce a new technique that repurposes a pre-trained video diffusion model trained on internet-scale datasets to recover videos revealing complex scene dynamics during the moment of capture and what might have occurred immediately into the past or future. Our approach is robust and versatile; it outperforms previous methods for this task, generalizes to challenging in-the-wild images, and supports downstream tasks such as recovering camera trajectories, object motion, and dynamic 3D scene structure. Code and data are available at https://blur2vid.github.io", "date": "2025-12-22", "categories": []}
{"id": "2412.09775", "title": "WaveOrder: A differentiable wave-optical framework for scalable biological microscopy with diverse modalities", "abstract": "Correlative computational microscopy can accelerate imaging and modeling of cellular dynamics by relaxing trade-offs inherent to dynamic imaging. Existing computational microscopy frameworks are either specialized or overly generic, limiting use to fixed configurations or domain experts. We introduce WaveOrder, a generalist wave-optical framework for imaging the architectural order of biomolecules. WaveOrder reconstructs diverse specimen properties from multi-channel acquisitions, with or without fluorescence. It provides a unified representation of linear optical properties and differentiable physics-based image formation models spanning widefield, confocal, light-sheet, and oblique label-free geometries. WaveOrder uses physics-informed ML to auto-tune model parameters and solve blind shift-variant restoration problems. This open-source, PyTorch-based framework enables scalable quantitative imaging across scales from organelles to adult zebrafish, and improves restoration of cellular structures in high-throughput experiments. We validate WaveOrder on diverse imaging applications, demonstrating its ability to recover biomolecular structure beyond the limits of existing approaches.", "date": "2024-12-13", "categories": []}
{"id": "2502.00404", "title": "Exploring Linear Attention Alternative for Single Image Super-Resolution", "abstract": "Deep learning-based single-image super-resolution (SISR) technology focuses on enhancing low-resolution (LR) images into high-resolution (HR) ones. Although significant progress has been made, challenges remain in computational complexity and quality, particularly in remote sensing image processing. To address these issues, we propose our Omni-Scale RWKV Super-Resolution (OmniRWKVSR) model which presents a novel approach that combines the Receptance Weighted Key Value (RWKV) architecture with feature extraction techniques such as Visual RWKV Spatial Mixing (VRSM) and Visual RWKV Channel Mixing (VRCM), aiming to overcome the limitations of existing methods and achieve superior SISR performance. This work has proved able to provide effective solutions for high-quality image reconstruction. Under the 4x Super-Resolution tasks, compared to the MambaIR model, we achieved an average improvement of 0.26% in PSNR and 0.16% in SSIM.", "date": "2025-02-01", "categories": []}
{"id": "2504.09377", "title": "Gradient as Conditions: Rethinking HOG for All-in-one Image Restoration", "abstract": "All-in-one image restoration (AIR) aims to address diverse degradations within a unified model by leveraging informative degradation conditions to guide the restoration process. However, existing methods often rely on implicitly learned priors, which may entangle feature representations and hinder performance in complex or unseen scenarios. Histogram of Oriented Gradients (HOG) as a classical gradient representation, we observe that it has strong discriminative capability across diverse degradations, making it a powerful and interpretable prior for AIR. Based on this insight, we propose HOGformer, a Transformer-based model that integrates learnable HOG features for degradation-aware restoration. The core of HOGformer is a Dynamic HOG-aware Self-Attention (DHOGSA) mechanism, which adaptively models long-range spatial dependencies conditioned on degradation-specific cues encoded by HOG descriptors. To further adapt the heterogeneity of degradations in AIR, we propose a Dynamic Interaction Feed-Forward (DIFF) module that facilitates channel-spatial interactions, enabling robust feature transformation under diverse degradations. Besides, we propose a HOG loss to explicitly enhance structural fidelity and edge sharpness. Extensive experiments on a variety of benchmarks, including adverse weather and natural degradations, demonstrate that HOGformer achieves state-of-the-art performance and generalizes well to complex real-world scenarios.Code is available at https://github.com/Fire-friend/HOGformer.", "date": "2025-04-12", "categories": []}
{"id": "2507.09885", "title": "MCGA: Mixture of Codebooks Hyperspectral Reconstruction via Grayscale-Aware Attention", "abstract": "Reconstructing hyperspectral images (HSIs) from RGB inputs provides a cost-effective alternative to hyperspectral cameras, but reconstructing high-dimensional spectra from three channels is inherently ill-posed. Existing methods typically directly regress RGB-to-HSI mappings using large attention networks, which are computationally expensive and handle ill-posedness only implicitly. We propose MCGA, a Mixture-of-Codebooks with Grayscale-aware Attention framework that explicitly addresses these challenges using spectral priors and photometric consistency. MCGA first learns transferable spectral priors via a mixture-of-codebooks (MoC) from heterogeneous HSI datasets, then aligns RGB features with these priors through grayscale-aware photometric attention (GANet). Efficiency and robustness are further improved via top-K attention design and test-time adaptation (TTA). Experiments on multiple real-world benchmarks demonstrate the state-of-the-art accuracy, strong cross-dataset generalization, and 4-5x faster inference. Codes will be available once acceptance at https://github.com/Fibonaccirabbit/MCGA.", "date": "2025-07-14", "categories": []}
{"id": "2508.15553", "title": "Deep Equilibrium Convolutional Sparse Coding for Hyperspectral Image Denoising", "abstract": "Hyperspectral images (HSIs) play a crucial role in remote sensing but are often degraded by complex noise patterns. Ensuring the physical property of the denoised HSIs is vital for robust HSI denoising, giving the rise of deep unfolding-based methods. However, these methods map the optimization of a physical model to a learnable network with a predefined depth, which lacks convergence guarantees. In contrast, Deep Equilibrium (DEQ) models treat the hidden layers of deep networks as the solution to a fixed-point problem and models them as infinite-depth networks, naturally consistent with the optimization. Under the framework of DEQ, we propose a Deep Equilibrium Convolutional Sparse Coding (DECSC) framework that unifies local spatial-spectral correlations, nonlocal spatial self-similarities, and global spatial consistency for robust HSI denoising. Within the convolutional sparse coding (CSC) framework, we enforce shared 2D convolutional sparse representation to ensure global spatial consistency across bands, while unshared 3D convolutional sparse representation captures local spatial-spectral details. To further exploit nonlocal self-similarities, a transformer block is embedded after the 2D CSC. Additionally, a detail enhancement module is integrated with the 3D CSC to promote image detail preservation. We formulate the proximal gradient descent of the CSC model as a fixed-point problem and transform the iterative updates into a learnable network architecture within the framework of DEQ. Experimental results demonstrate that our DECSC method achieves superior denoising performance compared to state-of-the-art methods.", "date": "2025-08-21", "categories": []}
{"id": "2509.06741", "title": "Event Spectroscopy: Event-based Multispectral and Depth Sensing using Structured Light", "abstract": "Uncrewed aerial vehicles (UAVs) are increasingly deployed in forest environments for tasks such as environmental monitoring and search and rescue, which require safe navigation through dense foliage and precise data collection. Traditional sensing approaches, including passive multispectral and RGB imaging, suffer from latency, poor depth resolution, and strong dependence on ambient light - especially under forest canopies. In this work, we present a novel event spectroscopy system that simultaneously enables high-resolution, low-latency depth reconstruction with integrated multispectral imaging using a single sensor. Depth is reconstructed using structured light, and by modulating the wavelength of the projected structured light, our system captures spectral information in controlled bands between 650 nm and 850 nm. We demonstrate up to $60\\%$ improvement in RMSE over commercial depth sensors and validate the spectral accuracy against a reference spectrometer and commercial multispectral cameras, demonstrating comparable performance. A portable version limited to RGB (3 wavelengths) is used to collect real-world depth and spectral data from a Masoala Rainforest. We demonstrate the use of this prototype for color image reconstruction and material differentiation between leaves and branches using spectral and depth data. Our results show that adding depth (available at no extra effort with our setup) to material differentiation improves the accuracy by over $30\\%$ compared to color-only method. Our system, tested in both lab and real-world rainforest environments, shows strong performance in depth estimation, RGB reconstruction, and material differentiation - paving the way for lightweight, integrated, and robust UAV perception and data collection in complex natural environments.", "date": "2025-09-08", "categories": []}
{"id": "2512.18367", "title": "PSI3D: Plug-and-Play 3D Stochastic Inference with Slice-wise Latent Diffusion Prior", "abstract": "Diffusion models are highly expressive image priors for Bayesian inverse problems. However, most diffusion models cannot operate on large-scale, high-dimensional data due to high training and inference costs. In this work, we introduce a Plug-and-play algorithm for 3D stochastic inference with latent diffusion prior (PSI3D) to address massive ($1024\\times 1024\\times 128$) volumes. Specifically, we formulate a Markov chain Monte Carlo approach to reconstruct each two-dimensional (2D) slice by sampling from a 2D latent diffusion model. To enhance inter-slice consistency, we also incorporate total variation (TV) regularization stochastically along the concatenation axis. We evaluate our performance on optical coherence tomography (OCT) super-resolution. Our method significantly improves reconstruction quality for large-scale scientific imaging compared to traditional and learning-based baselines, while providing robust and credible reconstructions.", "date": "2025-12-20", "categories": []}
{"id": "2512.19095", "title": "Mamba-Based Modality Disentanglement Network for Multi-Contrast MRI Reconstruction", "abstract": "Magnetic resonance imaging (MRI) is a cornerstone of modern clinical diagnosis, offering unparalleled soft-tissue contrast without ionizing radiation. However, prolonged scan times remain a major barrier to patient throughput and comfort. Existing accelerated MRI techniques often struggle with two key challenges: (1) failure to effectively utilize inherent K-space prior information, leading to persistent aliasing artifacts from zero-filled inputs; and (2) contamination of target reconstruction quality by irrelevant information when employing multi-contrast fusion strategies. To overcome these challenges, we present MambaMDN, a dual-domain framework for multi-contrast MRI reconstruction. Our approach first employs fully-sampled reference K-space data to complete the undersampled target data, generating structurally aligned but modality-mixed inputs. Subsequently, we develop a Mamba-based modality disentanglement network to extract and remove reference-specific features from the mixed representation. Furthermore, we introduce an iterative refinement mechanism to progressively enhance reconstruction accuracy through repeated feature purification. Extensive experiments demonstrate that MambaMDN can significantly outperform existing multi-contrast reconstruction methods.", "date": "2025-12-22", "categories": []}
{"id": "2512.19489", "title": "Rethinking Coupled Tensor Analysis for Hyperspectral Super-Resolution: Recoverable Modeling Under Endmember Variability", "abstract": "This work revisits the hyperspectral super-resolution (HSR) problem, i.e., fusing a pair of spatially co-registered hyperspectral (HSI) and multispectral (MSI) images to recover a super-resolution image (SRI) that enhances the spatial resolution of the HSI. Coupled tensor decomposition (CTD)-based methods have gained traction in this domain, offering recoverability guarantees under various assumptions. Existing models such as canonical polyadic decomposition (CPD) and Tucker decomposition provide strong expressive power but lack physical interpretability. The block-term decomposition model with rank-$(L_r, L_r, 1)$ terms (the LL1 model) yields interpretable factors under the linear mixture model (LMM) of spectral images, but LMM assumptions are often violated in practice -- primarily due to nonlinear effects such as endmember variability (EV). To address this, we propose modeling spectral images using a more flexible block-term tensor decomposition with rank-$(L_r, M_r, N_r)$ terms (the LMN model). This modeling choice retains interpretability, subsumes CPD, Tucker, and LL1 as special cases, and robustly accounts for non-ideal effects such as EV, offering a balanced tradeoff between expressiveness and interpretability for HSR. Importantly, under the LMN model for HSI and MSI, recoverability of the SRI can still be established under proper conditions -- providing strong theoretical support. Extensive experiments on synthetic and real datasets further validate the effectiveness and robustness of the proposed method compared with existing CTD-based approaches.", "date": "2025-12-22", "categories": []}
{"id": "2510.17330", "title": "CharDiff-LP: A Diffusion Model with Character-Level Guidance for License Plate Image Restoration", "abstract": "License plate image restoration is important not only as a preprocessing step for license plate recognition but also for enhancing evidential value, improving visual clarity, and enabling broader reuse of license plate images. We propose a novel diffusion-based framework with character-level guidance, CharDiff-LP, which effectively restores and recognizes severely degraded license plate images captured under realistic conditions. CharDiff-LP leverages fine-grained character-level priors extracted through external segmentation and Optical Character Recognition (OCR) modules tailored for low-quality license plate images. For precise and focused guidance, CharDiff-LP incorporates a novel Character-guided Attention through Region-wise Masking (CHARM) module, which ensures that each character's guidance is restricted to its own region, thereby avoiding interference with other regions. In experiments, CharDiff-LP significantly outperformed baseline restoration models in both restoration quality and recognition accuracy, achieving a 28.3% relative reduction in character error rate (CER) on the Roboflow-LP dataset compared with the best-performing baseline.", "date": "2025-10-20", "categories": []}
{"id": "2512.17578", "title": "3One2: One-step Regression Plus One-step Diffusion for One-hot Modulation in Dual-path Video Snapshot Compressive Imaging", "abstract": "Video snapshot compressive imaging (SCI) captures dynamic scene sequences through a two-dimensional (2D) snapshot, fundamentally relying on optical modulation for hardware compression and the corresponding software reconstruction. While mainstream video SCI using random binary modulation has demonstrated success, it inevitably results in temporal aliasing during compression. One-hot modulation, activating only one sub-frame per pixel, provides a promising solution for achieving perfect temporal decoupling, thereby alleviating issues associated with aliasing. However, no algorithms currently exist to fully exploit this potential. To bridge this gap, we propose an algorithm specifically designed for one-hot masks. First, leveraging the decoupling properties of one-hot modulation, we transform the reconstruction task into a generative video inpainting problem and introduce a stochastic differential equation (SDE) of the forward process that aligns with the hardware compression process. Next, we identify limitations of the pure diffusion method for video SCI and propose a novel framework that combines one-step regression initialization with one-step diffusion refinement. Furthermore, to mitigate the spatial degradation caused by one-hot modulation, we implement a dual optical path at the hardware level, utilizing complementary information from another path to enhance the inpainted video. To our knowledge, this is the first work integrating diffusion into video SCI reconstruction. Experiments conducted on synthetic datasets and real scenes demonstrate the effectiveness of our method.", "date": "2025-12-19", "categories": []}
{"id": "2512.16609", "title": "Hazedefy: A Lightweight Real-Time Image and Video Dehazing Pipeline for Practical Deployment", "abstract": "This paper introduces Hazedefy, a lightweight and application-focused dehazing pipeline intended for real-time video and live camera feed enhancement. Hazedefy prioritizes computational simplicity and practical deployability on consumer-grade hardware, building upon the Dark Channel Prior (DCP) concept and the atmospheric scattering model. Key elements include gamma-adaptive reconstruction, a fast transmission approximation with lower bounds for numerical stability, a stabilized atmospheric light estimator based on fractional top-pixel averaging, and an optional color balance stage. The pipeline is suitable for mobile and embedded applications, as experimental demonstrations on real-world images and videos show improved visibility and contrast without requiring GPU acceleration.", "date": "2025-12-18", "categories": []}
{"id": "2511.18806", "title": "TPG-INR: Target Prior-Guided Implicit 3D CT Reconstruction for Enhanced Sparse-view Imaging", "abstract": "X-ray imaging, based on penetration, enables detailed visualization of internal structures. Building on this capability, existing implicit 3D reconstruction methods have adapted the NeRF model and its variants for internal CT reconstruction. However, these approaches often neglect the significance of objects' anatomical priors for implicit learning, limiting both reconstruction precision and learning efficiency, particularly in ultra-sparse view scenarios. To address these challenges, we propose a novel 3D CT reconstruction framework that employs a 'target prior' derived from the object's projection data to enhance implicit learning. Our approach integrates positional and structural encoding to facilitate voxel-wise implicit reconstruction, utilizing the target prior to guide voxel sampling and enrich structural encoding. This dual strategy significantly boosts both learning efficiency and reconstruction quality. Additionally, we introduce a CUDA-based algorithm for rapid estimation of high-quality 3D target priors from sparse-view projections. Experiments utilizing projection data from a complex abdominal dataset demonstrate that the proposed model substantially enhances learning efficiency, outperforming the current leading model, NAF, by a factor of ten. In terms of reconstruction quality, it also exceeds the most accurate model, NeRP, achieving PSNR improvements of 3.57 dB, 5.42 dB, and 5.70 dB with 10, 20, and 30 projections, respectively. The code is available at https://github.com/qlcao171/TPG-INR.", "date": "2025-11-24", "categories": []}
{"id": "2512.15221", "title": "SLCFormer: Spectral-Local Context Transformer with Physics-Grounded Flare Synthesis for Nighttime Flare Removal", "abstract": "Lens flare is a common nighttime artifact caused by strong light sources scattering within camera lenses, leading to hazy streaks, halos, and glare that degrade visual quality. However, existing methods usually fail to effectively address nonuniform scattered flares, which severely reduces their applicability to complex real-world scenarios with diverse lighting conditions. To address this issue, we propose SLCFormer, a novel spectral-local context transformer framework for effective nighttime lens flare removal. SLCFormer integrates two key modules: the Frequency Fourier and Excitation Module (FFEM), which captures efficient global contextual representations in the frequency domain to model flare characteristics, and the Directionally-Enhanced Spatial Module (DESM) for local structural enhancement and directional features in the spatial domain for precise flare removal. Furthermore, we introduce a ZernikeVAE-based scatter flare generation pipeline to synthesize physically realistic scatter flares with spatially varying PSFs, bridging optical physics and data-driven training. Extensive experiments on the Flare7K++ dataset demonstrate that our method achieves state-of-the-art performance, outperforming existing approaches in both quantitative metrics and perceptual visual quality, and generalizing robustly to real nighttime scenes with complex flare artifacts.", "date": "2025-12-17", "categories": []}
{"id": "2504.07959", "title": "CCMNet: Leveraging Calibrated Color Correction Matrices for Cross-Camera Color Constancy", "abstract": "Computational color constancy, or white balancing, is a key module in a camera's image signal processor (ISP) that corrects color casts from scene lighting. Because this operation occurs in the camera-specific raw color space, white balance algorithms must adapt to different cameras. This paper introduces a learning-based method for cross-camera color constancy that generalizes to new cameras without retraining. Our method leverages pre-calibrated color correction matrices (CCMs) available on ISPs that map the camera's raw color space to a standard space (e.g., CIE XYZ). Our method uses these CCMs to transform predefined illumination colors (i.e., along the Planckian locus) into the test camera's raw space. The mapped illuminants are encoded into a compact camera fingerprint embedding (CFE) that enables the network to adapt to unseen cameras. To prevent overfitting due to limited cameras and CCMs during training, we introduce a data augmentation technique that interpolates between cameras and their CCMs. Experimental results across multiple datasets and backbones show that our method achieves state-of-the-art cross-camera color constancy while remaining lightweight and relying only on data readily available in camera ISPs.", "date": "2025-04-10", "categories": []}
{"id": "2512.11624", "title": "Fast and Explicit: Slice-to-Volume Reconstruction via 3D Gaussian Primitives with Analytic Point Spread Function Modeling", "abstract": "Recovering high-fidelity 3D images from sparse or degraded 2D images is a fundamental challenge in medical imaging, with broad applications ranging from 3D ultrasound reconstruction to MRI super-resolution. In the context of fetal MRI, high-resolution 3D reconstruction of the brain from motion-corrupted low-resolution 2D acquisitions is a prerequisite for accurate neurodevelopmental diagnosis. While implicit neural representations (INRs) have recently established state-of-the-art performance in self-supervised slice-to-volume reconstruction (SVR), they suffer from a critical computational bottleneck: accurately modeling the image acquisition physics requires expensive stochastic Monte Carlo sampling to approximate the point spread function (PSF). In this work, we propose a shift from neural network based implicit representations to Gaussian based explicit representations. By parameterizing the HR 3D image volume as a field of anisotropic Gaussian primitives, we leverage the property of Gaussians being closed under convolution and thus derive a \\textit{closed-form analytical solution} for the forward model. This formulation reduces the previously intractable acquisition integral to an exact covariance addition ($\\mathbf{\\Sigma}_{obs} = \\mathbf{\\Sigma}_{HR} + \\mathbf{\\Sigma}_{PSF}$), effectively bypassing the need for compute-intensive stochastic sampling while ensuring exact gradient propagation. We demonstrate that our approach matches the reconstruction quality of self-supervised state-of-the-art SVR frameworks while delivering a 5$\\times$--10$\\times$ speed-up on neonatal and fetal data. With convergence often reached in under 30 seconds, our framework paves the way towards translation into clinical routine of real-time fetal 3D MRI. Code will be public at {https://github.com/m-dannecker/Gaussian-Primitives-for-Fast-SVR}.", "date": "2025-12-12", "categories": []}
{"id": "2512.14028", "title": "Robust Single-shot Structured Light 3D Imaging via Neural Feature Decoding", "abstract": "We consider the problem of active 3D imaging using single-shot structured light systems, which are widely employed in commercial 3D sensing devices such as Apple Face ID and Intel RealSense. Traditional structured light methods typically decode depth correspondences through pixel-domain matching algorithms, resulting in limited robustness under challenging scenarios like occlusions, fine-structured details, and non-Lambertian surfaces. Inspired by recent advances in neural feature matching, we propose a learning-based structured light decoding framework that performs robust correspondence matching within feature space rather than the fragile pixel domain. Our method extracts neural features from the projected patterns and captured infrared (IR) images, explicitly incorporating their geometric priors by building cost volumes in feature space, achieving substantial performance improvements over pixel-domain decoding approaches. To further enhance depth quality, we introduce a depth refinement module that leverages strong priors from large-scale monocular depth estimation models, improving fine detail recovery and global structural coherence. To facilitate effective learning, we develop a physically-based structured light rendering pipeline, generating nearly one million synthetic pattern-image pairs with diverse objects and materials for indoor settings. Experiments demonstrate that our method, trained exclusively on synthetic data with multiple structured light patterns, generalizes well to real-world indoor environments, effectively processes various pattern types without retraining, and consistently outperforms both commercial structured light systems and passive stereo RGB-based depth estimation methods. Project page: https://namisntimpot.github.io/NSLweb/.", "date": "2025-12-16", "categories": []}
{"id": "2512.14435", "title": "Score-Based Turbo Message Passing for Plug-and-Play Compressive Imaging", "abstract": "Message-passing algorithms have been adapted for compressive imaging by incorporating various off-the-shelf image denoisers. However, these denoisers rely largely on generic or hand-crafted priors and often fall short in accurately capturing the complex statistical structure of natural images. As a result, traditional plug-and-play (PnP) methods often lead to suboptimal reconstruction, especially in highly underdetermined regimes. Recently, score-based generative models have emerged as a powerful framework for accurately characterizing sophisticated image distribution. Yet, their direct use for posterior sampling typically incurs prohibitive computational complexity. In this paper, by exploiting the close connection between score-based generative modeling and empirical Bayes denoising, we devise a message-passing framework that integrates a score-based minimum mean-squared error (MMSE) denoiser for compressive image recovery. The resulting algorithm, named score-based turbo message passing (STMP), combines the fast convergence of message passing with the expressive power of score-based generative priors. For practical systems with quantized measurements, we further propose quantized STMP (Q-STMP), which augments STMP with a component-wise MMSE dequantization module. We demonstrate that the asymptotic performance of STMP and Q-STMP can be accurately predicted by a set of state-evolution (SE) equations. Experiments on the FFHQ dataset demonstrate that STMP strikes a significantly better performance-complexity tradeoff compared with competing baselines, and that Q-STMP remains robust even under 1-bit quantization. Remarkably, both STMP and Q-STMP typically converge within 10 iterations.", "date": "2025-12-16", "categories": []}
{"id": "2212.14801", "title": "ExReg: Wide-range Photo Exposure Correction via a Multi-dimensional Regressor with Attention", "abstract": "Photo exposure correction is widely investigated, but fewer studies focus on correcting under- and over-exposed images simultaneously. Three issues remain open to handle and correct both under- and over-exposed images in a unified way. First, a locally-adaptive exposure adjustment may be more flexible instead of learning a global mapping. Second, it is an ill-posed problem to determine the suitable exposure values locally. Third, photos with the same content but different exposures may not reach consistent adjustment results. To this end, we proposed a novel exposure correction network, ExReg, to address the challenges by formulating exposure correction as a multi-dimensional regression process. Given an input image, a compact multi-exposure generation network is introduced to generate images with different exposure conditions for multi-dimensional regression and exposure correction in the next stage. An auxiliary module is designed to predict the region-wise exposure values, guiding the proposed Encoder-Decoder ANP (Attentive Neural Processes) to regress the final corrected image. The experimental results show that ExReg can generate well-exposed results and outperform the SOTA method in PSNR for extensive exposure problems. Furthermore, the processing speed, with 0.05 seconds per image on an RTX 3090, is efficient. When tested on the same image under various exposure levels, ExReg also yields results that are visually consistent and physically accurate.", "date": "2022-12-14", "categories": []}
{"id": "2407.21600", "title": "Robust Simultaneous Multislice MRI Reconstruction Using Slice-Wise Learned Generative Diffusion Priors", "abstract": "Simultaneous multislice (SMS) imaging is a powerful technique for accelerating magnetic resonance imaging (MRI) acquisitions. However, SMS reconstruction remains challenging due to complex signal interactions between and within the excited slices. In this study, we introduce ROGER, a robust SMS MRI reconstruction method based on deep generative priors. Utilizing denoising diffusion probabilistic models (DDPM), ROGER begins with Gaussian noise and gradually recovers individual slices through reverse diffusion iterations while enforcing data consistency from measured k-space data within the readout concatenation framework. The posterior sampling procedure is designed such that the DDPM training can be performed on single-slice images without requiring modifications for SMS tasks. Additionally, our method incorporates a low-frequency enhancement (LFE) module to address the practical issue that SMS-accelerated fast spin echo (FSE) and echo planar imaging (EPI) sequences cannot easily embed fully-sampled autocalibration signals. Extensive experiments on both retrospectively and prospectively accelerated datasets demonstrate that ROGER consistently outperforms existing methods, enhancing both anatomical and functional imaging with strong out-of-distribution generalization. The source code and sample data for ROGER are available at https://github.com/Solor-pikachu/ROGER.", "date": "2024-07-31", "categories": []}
{"id": "2409.18303", "title": "Deep-ER: Deep Learning ECCENTRIC Reconstruction for fast high-resolution neurometabolic imaging", "abstract": "Introduction: Altered neurometabolism is an important pathological mechanism in many neurological diseases and brain cancer, which can be mapped non-invasively by Magnetic Resonance Spectroscopic Imaging (MRSI). Advanced MRSI using non-cartesian compressed-sense acquisition enables fast high-resolution metabolic imaging but has lengthy reconstruction times that limits throughput and needs expert user interaction. Here, we present a robust and efficient Deep Learning reconstruction to obtain high-quality metabolic maps.\n  Methods: Fast high-resolution whole-brain metabolic imaging was performed at 3.4 mm$^3$ isotropic resolution with acquisition times between 4:11-9:21 min:s using ECCENTRIC pulse sequence on a 7T MRI scanner. Data were acquired in a high-resolution phantom and 27 human participants, including 22 healthy volunteers and 5 glioma patients. A deep neural network using recurring interlaced convolutional layers with joint dual-space feature representation was developed for deep learning ECCENTRIC reconstruction (Deep-ER). 21 subjects were used for training and 6 subjects for testing. Deep-ER performance was compared to conventional iterative Total Generalized Variation reconstruction using image and spectral quality metrics.\n  Results: Deep-ER demonstrated 600-fold faster reconstruction than conventional methods, providing improved spatial-spectral quality and metabolite quantification with 12%-45% (P<0.05) higher signal-to-noise and 8%-50% (P<0.05) smaller Cramer-Rao lower bounds. Metabolic images clearly visualize glioma tumor heterogeneity and boundary.\n  Conclusion: Deep-ER provides efficient and robust reconstruction for sparse-sampled MRSI. The accelerated acquisition-reconstruction MRSI is compatible with high-throughput imaging workflow. It is expected that such improved performance will facilitate basic and clinical MRSI applications.", "date": "2024-09-26", "categories": []}
{"id": "2503.22140", "title": "Score-Based Turbo Message Passing for Plug-and-Play Compressive Image Recovery", "abstract": "Message passing algorithms have been tailored for compressive imaging applications by plugging in different types of off-the-shelf image denoisers. These off-the-shelf denoisers mostly rely on some generic or hand-crafted priors for denoising. Due to their insufficient accuracy in capturing the true image prior, these methods often fail to produce satisfactory results, especially in highly underdetermined scenarios. On the other hand, score-based generative modeling offers a promising way to accurately characterize the sophisticated image distribution. In this paper, by exploiting the close relation between score-based modeling and empirical Bayes-optimal denoising, we devise a message passing framework that integrates a score-based minimum mean squared error (MMSE) denoiser for compressive image recovery. Experiments on the FFHQ dataset demonstrate that our method strikes a significantly better performance-complexity tradeoff than conventional message passing, regularized linear regression, and score-based posterior sampling baselines. Remarkably, our method typically converges in fewer than 20 neural function evaluations (NFEs).", "date": "2025-03-28", "categories": []}
{"id": "2411.05771", "title": "Equivariant Test-Time Training with Operator Sketching for Imaging Inverse Problems", "abstract": "Equivariant Imaging (EI) regularization has become the de-facto technique for unsupervised training of deep imaging networks, without any need of ground-truth data. Observing that the EI-based unsupervised training paradigm currently has significant computational redundancy leading to inefficiency in high-dimensional applications, we propose a sketched EI regularization which leverages the randomized sketching techniques for acceleration. We apply our sketched EI regularization to develop an accelerated deep internal learning framework, which can be efficiently applied for test-time network adaptation. Additionally, for network adaptation tasks, we propose a parameter-efficient approach to accelerate both EI and Sketched-EI via optimizing only the normalization layers. Our numerical study on X-ray CT and multicoil magnetic resonance image reconstruction tasks demonstrate that our approach can achieve significant computational acceleration over the standard EI counterpart, especially in test-time training tasks.", "date": "2024-11-08", "categories": []}
{"id": "2512.05104", "title": "EvoIR: Towards All-in-One Image Restoration via Evolutionary Frequency Modulation", "abstract": "All-in-One Image Restoration (AiOIR) tasks often involve diverse degradation that require robust and versatile strategies. However, most existing approaches typically lack explicit frequency modeling and rely on fixed or heuristic optimization schedules, which limit the generalization across heterogeneous degradation. To address these limitations, we propose EvoIR, an AiOIR-specific framework that introduces evolutionary frequency modulation for dynamic and adaptive image restoration. Specifically, EvoIR employs the Frequency-Modulated Module (FMM) that decomposes features into high- and low-frequency branches in an explicit manner and adaptively modulates them to enhance both structural fidelity and fine-grained details. Central to EvoIR, an Evolutionary Optimization Strategy (EOS) iteratively adjusts frequency-aware objectives through a population-based evolutionary process, dynamically balancing structural accuracy and perceptual fidelity. Its evolutionary guidance further mitigates gradient conflicts across degradation and accelerates convergence. By synergizing FMM and EOS, EvoIR yields greater improvements than using either component alone, underscoring their complementary roles. Extensive experiments on multiple benchmarks demonstrate that EvoIR outperforms state-of-the-art AiOIR methods.", "date": "2025-12-04", "categories": []}
{"id": "2512.10948", "title": "ClusIR: Towards Cluster-Guided All-in-One Image Restoration", "abstract": "All-in-One Image Restoration (AiOIR) aims to recover high-quality images from diverse degradations within a unified framework. However, existing methods often fail to explicitly model degradation types and struggle to adapt their restoration behavior to complex or mixed degradations. To address these issues, we propose ClusIR, a Cluster-Guided Image Restoration framework that explicitly models degradation semantics through learnable clustering and propagates cluster-aware cues across spatial and frequency domains for adaptive restoration. Specifically, ClusIR comprises two key components: a Probabilistic Cluster-Guided Routing Mechanism (PCGRM) and a Degradation-Aware Frequency Modulation Module (DAFMM). The proposed PCGRM disentangles degradation recognition from expert activation, enabling discriminative degradation perception and stable expert routing. Meanwhile, DAFMM leverages the cluster-guided priors to perform adaptive frequency decomposition and targeted modulation, collaboratively refining structural and textural representations for higher restoration fidelity. The cluster-guided synergy seamlessly bridges semantic cues with frequency-domain modulation, empowering ClusIR to attain remarkable restoration results across a wide range of degradations. Extensive experiments on diverse benchmarks validate that ClusIR reaches competitive performance under several scenarios.", "date": "2025-12-11", "categories": []}
{"id": "2512.09016", "title": "Learning to Remove Lens Flare in Event Camera", "abstract": "Event cameras have the potential to revolutionize vision systems with their high temporal resolution and dynamic range, yet they remain susceptible to lens flare, a fundamental optical artifact that causes severe degradation. In event streams, this optical artifact forms a complex, spatio-temporal distortion that has been largely overlooked. We present E-Deflare, the first systematic framework for removing lens flare from event camera data. We first establish the theoretical foundation by deriving a physics-grounded forward model of the non-linear suppression mechanism. This insight enables the creation of the E-Deflare Benchmark, a comprehensive resource featuring a large-scale simulated training set, E-Flare-2.7K, and the first-ever paired real-world test set, E-Flare-R, captured by our novel optical system. Empowered by this benchmark, we design E-DeflareNet, which achieves state-of-the-art restoration performance. Extensive experiments validate our approach and demonstrate clear benefits for downstream tasks. Code and datasets are publicly available.", "date": "2025-12-09", "categories": []}
{"id": "2512.09871", "title": "Diffusion Posterior Sampler for Hyperspectral Unmixing with Spectral Variability Modeling", "abstract": "Linear spectral mixture models (LMM) provide a concise form to disentangle the constituent materials (endmembers) and their corresponding proportions (abundance) in a single pixel. The critical challenges are how to model the spectral prior distribution and spectral variability. Prior knowledge and spectral variability can be rigorously modeled under the Bayesian framework, where posterior estimation of Abundance is derived by combining observed data with endmember prior distribution. Considering the key challenges and the advantages of the Bayesian framework, a novel method using a diffusion posterior sampler for semiblind unmixing, denoted as DPS4Un, is proposed to deal with these challenges with the following features: (1) we view the pretrained conditional spectrum diffusion model as a posterior sampler, which can combine the learned endmember prior with observation to get the refined abundance distribution. (2) Instead of using the existing spectral library as prior, which may raise bias, we establish the image-based endmember bundles within superpixels, which are used to train the endmember prior learner with diffusion model. Superpixels make sure the sub-scene is more homogeneous. (3) Instead of using the image-level data consistency constraint, the superpixel-based data fidelity term is proposed. (4) The endmember is initialized as Gaussian noise for each superpixel region, DPS4Un iteratively updates the abundance and endmember, contributing to spectral variability modeling. The experimental results on three real-world benchmark datasets demonstrate that DPS4Un outperforms the state-of-the-art hyperspectral unmixing methods.", "date": "2025-12-10", "categories": []}
{"id": "2512.08229", "title": "Geometry-Aware Sparse Depth Sampling for High-Fidelity RGB-D Depth Completion in Robotic Systems", "abstract": "Accurate three-dimensional perception is essential for modern industrial robotic systems that perform manipulation, inspection, and navigation tasks. RGB-D and stereo vision sensors are widely used for this purpose, but the depth maps they produce are often noisy, incomplete, or biased due to sensor limitations and environmental conditions. Depth completion methods aim to generate dense, reliable depth maps from RGB images and sparse depth input. However, a key limitation in current depth completion pipelines is the unrealistic generation of sparse depth: sparse pixels are typically selected uniformly at random from dense ground-truth depth, ignoring the fact that real sensors exhibit geometry-dependent and spatially nonuniform reliability. In this work, we propose a normal-guided sparse depth sampling strategy that leverages PCA-based surface normal estimation on the RGB-D point cloud to compute a per-pixel depth reliability measure. The sparse depth samples are then drawn according to this reliability distribution. We integrate this sampling method with the Marigold-DC diffusion-based depth completion model and evaluate it on NYU Depth v2 using the standard metrics. Experiments show that our geometry-aware sparse depth improves accuracy, reduces artifacts near edges and discontinuities, and produces more realistic training conditions that better reflect real sensor behavior.", "date": "2025-12-09", "categories": []}
{"id": "2502.20333", "title": "T1-PILOT: Optimized Trajectories for T1 Mapping Acceleration", "abstract": "Cardiac T1 mapping provides critical quantitative insights into myocardial tissue composition, enabling the assessment of pathologies such as fibrosis, inflammation, and edema. However, the inherently dynamic nature of the heart imposes strict limits on acquisition times, making high-resolution T1 mapping a persistent challenge. Compressed sensing (CS) approaches have reduced scan durations by undersampling k-space and reconstructing images from partial data, and recent studies show that jointly optimizing the undersampling patterns with the reconstruction network can substantially improve performance. Still, most current T1 mapping pipelines rely on static, hand-crafted masks that do not exploit the full acceleration and accuracy potential. In this work, we introduce T1-PILOT: an end-to-end method that explicitly incorporates the T1 signal relaxation model into the sampling-reconstruction framework to guide the learning of non-Cartesian trajectories, crossframe alignment, and T1 decay estimation. Through extensive experiments on the CMRxRecon dataset, T1-PILOT significantly outperforms several baseline strategies (including learned single-mask and fixed radial or golden-angle sampling schemes), achieving higher T1 map fidelity at greater acceleration factors. In particular, we observe consistent gains in PSNR and VIF relative to existing methods, along with marked improvements in delineating finer myocardial structures. Our results highlight that optimizing sampling trajectories in tandem with the physical relaxation model leads to both enhanced quantitative accuracy and reduced acquisition times.", "date": "2025-02-27", "categories": []}
{"id": "2507.18594", "title": "DRWKV: Focusing on Object Edges for Low-Light Image Enhancement", "abstract": "Low-light image enhancement remains a challenging task, particularly in preserving object edge continuity and fine structural details under extreme illumination degradation. In this paper, we propose a novel model, DRWKV (Detailed Receptance Weighted Key Value), which integrates our proposed Global Edge Retinex (GER) theory, enabling effective decoupling of illumination and edge structures for enhanced edge fidelity. Secondly, we introduce Evolving WKV Attention, a spiral-scanning mechanism that captures spatial edge continuity and models irregular structures more effectively. Thirdly, we design the Bilateral Spectrum Aligner (Bi-SAB) and a tailored MS2-Loss to jointly align luminance and chrominance features, improving visual naturalness and mitigating artifacts. Extensive experiments on five LLIE benchmarks demonstrate that DRWKV achieves leading performance in PSNR, SSIM, and NIQE while maintaining low computational complexity. Furthermore, DRWKV enhances downstream performance in low-light multi-object tracking tasks, validating its generalization capabilities.", "date": "2025-07-24", "categories": []}
{"id": "2512.06530", "title": "On The Role of K-Space Acquisition in MRI Reconstruction Domain-Generalization", "abstract": "Recent work has established learned k-space acquisition patterns as a promising direction for improving reconstruction quality in accelerated Magnetic Resonance Imaging (MRI). Despite encouraging results, most existing research focuses on acquisition patterns optimized for a single dataset or modality, with limited consideration of their transferability across imaging domains. In this work, we demonstrate that the benefits of learned k-space sampling can extend beyond the training domain, enabling superior reconstruction performance under domain shifts. Our study presents two main contributions. First, through systematic evaluation across datasets and acquisition paradigms, we show that models trained with learned sampling patterns exhibitimproved generalization under cross-domain settings. Second, we propose a novel method that enhances domain robustness by introducing acquisition uncertainty during training-stochastically perturbing k-space trajectories to simulate variability across scanners and imaging conditions. Our results highlight the importance of treating kspace trajectory design not merely as an acceleration mechanism, but as an active degree of freedom for improving domain generalization in MRI reconstruction.", "date": "2025-12-06", "categories": []}
{"id": "2512.06977", "title": "Physics-Guided Diffusion Priors for Multi-Slice Reconstruction in Scientific Imaging", "abstract": "Accurate multi-slice reconstruction from limited measurement data is crucial to speed up the acquisition process in medical and scientific imaging. However, it remains challenging due to the ill-posed nature of the problem and the high computational and memory demands. We propose a framework that addresses these challenges by integrating partitioned diffusion priors with physics-based constraints. By doing so, we substantially reduce memory usage per GPU while preserving high reconstruction quality, outperforming both physics-only and full multi-slice reconstruction baselines for different modalities, namely Magnetic Resonance Imaging (MRI) and four-dimensional Scanning Transmission Electron Microscopy (4D-STEM). Additionally, we show that the proposed method improves in-distribution accuracy as well as strong generalization to out-of-distribution datasets.", "date": "2025-12-07", "categories": []}
{"id": "2512.07171", "title": "TIDE: Two-Stage Inverse Degradation Estimation with Guided Prior Disentanglement for Underwater Image Restoration", "abstract": "Underwater image restoration is essential for marine applications ranging from ecological monitoring to archaeological surveys, but effectively addressing the complex and spatially varying nature of underwater degradations remains a challenge. Existing methods typically apply uniform restoration strategies across the entire image, struggling to handle multiple co-occurring degradations that vary spatially and with water conditions. We introduce TIDE, a $\\underline{t}$wo stage $\\underline{i}$nverse $\\underline{d}$egradation $\\underline{e}$stimation framework that explicitly models degradation characteristics and applies targeted restoration through specialized prior decomposition. Our approach disentangles the restoration process into multiple specialized hypotheses that are adaptively fused based on local degradation patterns, followed by a progressive refinement stage that corrects residual artifacts. Specifically, TIDE decomposes underwater degradations into four key factors, namely color distortion, haze, detail loss, and noise, and designs restoration experts specialized for each. By generating specialized restoration hypotheses, TIDE balances competing degradation factors and produces natural results even in highly degraded regions. Extensive experiments across both standard benchmarks and challenging turbid water conditions show that TIDE achieves competitive performance on reference based fidelity metrics while outperforming state of the art methods on non reference perceptual quality metrics, with strong improvements in color correction and contrast enhancement. Our code is available at: https://rakesh-123-cryp.github.io/TIDE.", "date": "2025-12-08", "categories": []}
{"id": "2512.05240", "title": "IE2Video: Adapting Pretrained Diffusion Models for Event-Based Video Reconstruction", "abstract": "Continuous video monitoring in surveillance, robotics, and wearable systems faces a fundamental power constraint: conventional RGB cameras consume substantial energy through fixed-rate capture. Event cameras offer sparse, motion-driven sensing with low power consumption, but produce asynchronous event streams rather than RGB video. We propose a hybrid capture paradigm that records sparse RGB keyframes alongside continuous event streams, then reconstructs full RGB video offline -- reducing capture power consumption while maintaining standard video output for downstream applications. We introduce the Image and Event to Video (IE2Video) task: reconstructing RGB video sequences from a single initial frame and subsequent event camera data. We investigate two architectural strategies: adapting an autoregressive model (HyperE2VID) for RGB generation, and injecting event representations into a pretrained text-to-video diffusion model (LTX) via learned encoders and low-rank adaptation. Our experiments demonstrate that the diffusion-based approach achieves 33\\% better perceptual quality than the autoregressive baseline (0.283 vs 0.422 LPIPS). We validate our approach across three event camera datasets (BS-ERGB, HS-ERGB far/close) at varying sequence lengths (32-128 frames), demonstrating robust cross-dataset generalization with strong performance on unseen capture configurations.", "date": "2025-12-04", "categories": []}
{"id": "2512.05481", "title": "UniFS: Unified Multi-Contrast MRI Reconstruction via Frequency-Spatial Fusion", "abstract": "Recently, Multi-Contrast MR Reconstruction (MCMR) has emerged as a hot research topic that leverages high-quality auxiliary modalities to reconstruct undersampled target modalities of interest. However, existing methods often struggle to generalize across different k-space undersampling patterns, requiring the training of a separate model for each specific pattern, which limits their practical applicability. To address this challenge, we propose UniFS, a Unified Frequency-Spatial Fusion model designed to handle multiple k-space undersampling patterns for MCMR tasks without any need for retraining. UniFS integrates three key modules: a Cross-Modal Frequency Fusion module, an Adaptive Mask-Based Prompt Learning module, and a Dual-Branch Complementary Refinement module. These modules work together to extract domain-invariant features from diverse k-space undersampling patterns while dynamically adapt to their own variations. Another limitation of existing MCMR methods is their tendency to focus solely on spatial information while neglect frequency characteristics, or extract only shallow frequency features, thus failing to fully leverage complementary cross-modal frequency information. To relieve this issue, UniFS introduces an adaptive prompt-guided frequency fusion module for k-space learning, significantly enhancing the model's generalization performance. We evaluate our model on the BraTS and HCP datasets with various k-space undersampling patterns and acceleration factors, including previously unseen patterns, to comprehensively assess UniFS's generalizability. Experimental results across multiple scenarios demonstrate that UniFS achieves state-of-the-art performance. Our code is available at https://github.com/LIKP0/UniFS.", "date": "2025-12-05", "categories": []}
{"id": "2512.05674", "title": "Hyperspectral Unmixing with 3D Convolutional Sparse Coding and Projected Simplex Volume Maximization", "abstract": "Hyperspectral unmixing (HSU) aims to separate each pixel into its constituent endmembers and estimate their corresponding abundance fractions. This work presents an algorithm-unrolling-based network for the HSU task, named the 3D Convolutional Sparse Coding Network (3D-CSCNet), built upon a 3D CSC model. Unlike existing unrolling-based networks, our 3D-CSCNet is designed within the powerful autoencoder (AE) framework. Specifically, to solve the 3D CSC problem, we propose a 3D CSC block (3D-CSCB) derived through deep algorithm unrolling. Given a hyperspectral image (HSI), 3D-CSCNet employs the 3D-CSCB to estimate the abundance matrix. The use of 3D CSC enables joint learning of spectral and spatial relationships in the 3D HSI data cube. The estimated abundance matrix is then passed to the AE decoder to reconstruct the HSI, and the decoder weights are extracted as the endmember matrix. Additionally, we propose a projected simplex volume maximization (PSVM) algorithm for endmember estimation, and the resulting endmembers are used to initialize the decoder weights of 3D-CSCNet. Extensive experiments on three real datasets and one simulated dataset with three different signal-to-noise ratio (SNR) levels demonstrate that our 3D-CSCNet outperforms state-of-the-art methods.", "date": "2025-12-05", "categories": []}
{"id": "2512.04222", "title": "ReasonX: MLLM-Guided Intrinsic Image Decomposition", "abstract": "Intrinsic image decomposition aims to separate images into physical components such as albedo, depth, normals, and illumination. While recent diffusion- and transformer-based models benefit from paired supervision from synthetic datasets, their generalization to diverse, real-world scenarios remains challenging. We propose ReasonX, a novel framework that leverages a multimodal large language model (MLLM) as a perceptual judge providing relative intrinsic comparisons, and uses these comparisons as GRPO rewards for fine-tuning intrinsic decomposition models on unlabeled, in-the-wild images. Unlike RL methods for generative models, our framework aligns conditional intrinsic predictors by rewarding agreement between the judge's relational assessments and analytically derived relations from the model's outputs. ReasonX is model-agnostic and can be applied to different intrinsic predictors. Across multiple base architectures and modalities, ReasonX yields significant improvements, including 9-25% WHDR reduction on IIW albedo and up to 46% depth accuracy gains on ETH3D, highlighting the promise of MLLM-guided comparative supervision to bridge low- and high-level vision reasoning.", "date": "2025-12-03", "categories": []}
{"id": "2512.04283", "title": "Plug-and-Play Image Restoration with Flow Matching: A Continuous Viewpoint", "abstract": "Flow matching-based generative models have been integrated into the plug-and-play image restoration framework, and the resulting plug-and-play flow matching (PnP-Flow) model has achieved some remarkable empirical success for image restoration. However, the theoretical understanding of PnP-Flow lags its empirical success. In this paper, we derive a continuous limit for PnP-Flow, resulting in a stochastic differential equation (SDE) surrogate model of PnP-Flow. The SDE model provides two particular insights to improve PnP-Flow for image restoration: (1) It enables us to quantify the error for image restoration, informing us to improve step scheduling and regularize the Lipschitz constant of the neural network-parameterized vector field for error reduction. (2) It informs us to accelerate off-the-shelf PnP-Flow models via extrapolation, resulting in a rescaled version of the proposed SDE model. We validate the efficacy of the SDE-informed improved PnP-Flow using several benchmark tasks, including image denoising, deblurring, super-resolution, and inpainting. Numerical results show that our method significantly outperforms the baseline PnP-Flow and other state-of-the-art approaches, achieving superior performance across evaluation metrics.", "date": "2025-12-03", "categories": []}
{"id": "2512.04390", "title": "FMA-Net++: Motion- and Exposure-Aware Real-World Joint Video Super-Resolution and Deblurring", "abstract": "Real-world video restoration is plagued by complex degradations from motion coupled with dynamically varying exposure - a key challenge largely overlooked by prior works and a common artifact of auto-exposure or low-light capture. We present FMA-Net++, a framework for joint video super-resolution and deblurring that explicitly models this coupled effect of motion and dynamically varying exposure. FMA-Net++ adopts a sequence-level architecture built from Hierarchical Refinement with Bidirectional Propagation blocks, enabling parallel, long-range temporal modeling. Within each block, an Exposure Time-aware Modulation layer conditions features on per-frame exposure, which in turn drives an exposure-aware Flow-Guided Dynamic Filtering module to infer motion- and exposure-aware degradation kernels. FMA-Net++ decouples degradation learning from restoration: the former predicts exposure- and motion-aware priors to guide the latter, improving both accuracy and efficiency. To evaluate under realistic capture conditions, we introduce REDS-ME (multi-exposure) and REDS-RE (random-exposure) benchmarks. Trained solely on synthetic data, FMA-Net++ achieves state-of-the-art accuracy and temporal consistency on our new benchmarks and GoPro, outperforming recent methods in both restoration quality and inference speed, and generalizes well to challenging real-world videos.", "date": "2025-12-04", "categories": []}
{"id": "2512.04699", "title": "OmniScaleSR: Unleashing Scale-Controlled Diffusion Prior for Faithful and Realistic Arbitrary-Scale Image Super-Resolution", "abstract": "Arbitrary-scale super-resolution (ASSR) overcomes the limitation of traditional super-resolution (SR) methods that operate only at fixed scales (e.g., 4x), enabling a single model to handle arbitrary magnification. Most existing ASSR approaches rely on implicit neural representation (INR), but its regression-driven feature extraction and aggregation intrinsically limit the ability to synthesize fine details, leading to low realism. Recent diffusion-based realistic image super-resolution (Real-ISR) models leverage powerful pre-trained diffusion priors and show impressive results at the 4x setting. We observe that they can also achieve ASSR because the diffusion prior implicitly adapts to scale by encouraging high-realism generation. However, without explicit scale control, the diffusion process cannot be properly adjusted for different magnification levels, resulting in excessive hallucination or blurry outputs, especially under ultra-high scales. To address these issues, we propose OmniScaleSR, a diffusion-based realistic arbitrary-scale SR framework designed to achieve both high fidelity and high realism. We introduce explicit, diffusion-native scale control mechanisms that work synergistically with implicit scale adaptation, enabling scale-aware and content-aware modulation of the diffusion process. In addition, we incorporate multi-domain fidelity enhancement designs to further improve reconstruction accuracy. Extensive experiments on bicubic degradation benchmarks and real-world datasets show that OmniScaleSR surpasses state-of-the-art methods in both fidelity and perceptual realism, with particularly strong performance at large magnification factors. Code will be released at https://github.com/chaixinning/OmniScaleSR.", "date": "2025-12-04", "categories": []}
{"id": "2510.02208", "title": "MACS: Measurement-Aware Consistency Sampling for Inverse Problems", "abstract": "Diffusion models have emerged as powerful generative priors for solving inverse imaging problems. However, their practical deployment is hindered by the substantial computational cost of slow, multi-step sampling. Although Consistency Models (CMs) address this limitation by enabling high-quality generation in only one or a few steps, their direct application to inverse problems has remained largely unexplored. This paper introduces a modified consistency sampling framework specifically designed for inverse problems. The proposed approach regulates the sampler's stochasticity through a measurement-consistency mechanism that leverages the degradation operator, thereby enforcing fidelity to the observed data while preserving the computational efficiency of consistency-based generation. Comprehensive experiments on the Fashion-MNIST and LSUN Bedroom datasets demonstrate consistent improvements across both perceptual and pixel-level metrics, including the Fr\\'echet Inception Distance (FID), Kernel Inception Distance (KID), peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM), compared with baseline consistency and diffusion-based sampling methods. The proposed method achieves competitive or superior reconstruction quality with only a small number of sampling steps.", "date": "2025-10-02", "categories": []}
{"id": "2512.03852", "title": "Traffic Image Restoration under Adverse Weather via Frequency-Aware Mamba", "abstract": "Traffic image restoration under adverse weather conditions remains a critical challenge for intelligent transportation systems. Existing methods primarily focus on spatial-domain modeling but neglect frequency-domain priors. Although the emerging Mamba architecture excels at long-range dependency modeling through patch-wise correlation analysis, its potential for frequency-domain feature extraction remains unexplored. To address this, we propose Frequency-Aware Mamba (FAMamba), a novel framework that integrates frequency guidance with sequence modeling for efficient image restoration. Our architecture consists of two key components: (1) a Dual-Branch Feature Extraction Block (DFEB) that enhances local-global interaction via bidirectional 2D frequency-adaptive scanning, dynamically adjusting traversal paths based on sub-band texture distributions; and (2) a Prior-Guided Block (PGB) that refines texture details through wavelet-based high-frequency residual learning, enabling high-quality image reconstruction with precise details. Meanwhile, we design a novel Adaptive Frequency Scanning Mechanism (AFSM) for the Mamba architecture, which enables the Mamba to achieve frequency-domain scanning across distinct subgraphs, thereby fully leveraging the texture distribution characteristics inherent in subgraph structures. Extensive experiments demonstrate the efficiency and effectiveness of FAMamba.", "date": "2025-12-03", "categories": []}
{"id": "2507.14308", "title": "Self-Supervised Joint Reconstruction and Denoising of T2-Weighted PROPELLER MRI of the Lungs at 0.55T", "abstract": "Purpose: This study aims to improve 0.55T T2-weighted PROPELLER lung MRI through a self-supervised joint reconstruction and denoising model.\n  Methods: T2-weighted 0.55T lung MRI dataset including 44 patients with previous covid infection were used. A self-supervised learning framework was developed, where each blade of the PROPELLER acquisition was split along the readout direction into two partitions. One subset trains the unrolled reconstruction network, while the other subset is used for loss calculation, enabling self-supervised training without clean targets and leveraging matched noise statistics for denoising. For comparison, Marchenko-Pastur Principal Component Analysis (MPPCA) was performed along the coil dimension, followed by conventional parallel imaging reconstruction. The quality of the reconstructed lung MRI was assessed visually by two experienced radiologists independently.\n  Results: The proposed self-supervised model improved the clarity and structural integrity of the lung images. For cases with available CT scans, the reconstructed images demonstrated strong alignment with corresponding CT images. Additionally, the proposed model enables further scan time reduction by requiring only half the number of blades. Reader evaluations confirmed that the proposed method outperformed MPPCA-denoised images across all categories (Wilcoxon signed-rank test, p<0.001), with moderate inter-reader agreement (weighted Cohen's kappa=0.55; percentage of exact and within +/-1 point agreement=91%).\n  Conclusion: By leveraging intrinsic structural redundancies between two disjoint splits of k-space subsets, the proposed self-supervised learning model effectively reconstructs the image while suppressing the noise for 0.55T T2-weighted lung MRI with PROPELLER sampling.", "date": "2025-07-18", "categories": []}
{"id": "2508.09200", "title": "Zero-shot self-supervised learning of single breath-hold magnetic resonance cholangiopancreatography (MRCP) reconstruction", "abstract": "To investigate the feasibility of zero-shot self-supervised learning reconstruction for reducing breath-hold times in magnetic resonance cholangiopancreatography (MRCP). Breath-hold MRCP was acquired from 11 healthy volunteers on 3T scanners using an incoherent k-space sampling pattern, leading to 14-second acquisition time and an acceleration factor of R=25. Zero-shot reconstruction was compared with parallel imaging of respiratory-triggered MRCP (338s, R=3) and compressed sensing reconstruction. For two volunteers, breath-hold scans (40s, R=6) were additionally acquired and retrospectively undersampled to R=25 to compute peak signal-to-noise ratio (PSNR). To address long zero-shot training time, the n+m full stages of the zero-shot learning were divided into two parts to reduce backpropagation depth during training: 1) n frozen stages initialized with n-stage pretrained network and 2) m trainable stages initialized either randomly or m-stage pretrained network. Efficiency of our approach was assessed by varying initialization strategies and the number of trainable stages using the retrospectively undersampled data. Zero-shot reconstruction significantly improved visual image quality over compressed sensing, particularly in SNR and ductal delineation, and achieved image quality comparable to that of successful respiratory-triggered acquisitions with regular breathing patterns. Improved initializations enhanced PSNR and reduced reconstruction time. Adjusting frozen/trainable configurations demonstrated that PSNR decreased only slightly from 38.25 dB (0/13) to 37.67 dB (12/1), while training time decreased up to 6.7-fold. Zero-shot learning delivers high-fidelity MRCP reconstructions with reduced breath-hold times, and the proposed partially trainable approach offers a practical solution for translation into time-constrained clinical workflows.", "date": "2025-08-08", "categories": []}
{"id": "2511.13922", "title": "Self-Supervised Compression and Artifact Correction for Streaming Underwater Imaging Sonar", "abstract": "Real-time imaging sonar is crucial for underwater monitoring where optical sensing fails, but its use is limited by low uplink bandwidth and severe sonar-specific artifacts (speckle, motion blur, reverberation, acoustic shadows) affecting up to 98% of frames. We present SCOPE, a self-supervised framework that jointly performs compression and artifact correction without clean-noise pairs or synthetic assumptions. SCOPE combines (i) Adaptive Codebook Compression (ACC), which learns frequency-encoded latent representations tailored to sonar, with (ii) Frequency-Aware Multiscale Segmentation (FAMS), which decomposes frames into low-frequency structure and sparse high-frequency dynamics while suppressing rapidly fluctuating artifacts. A hedging training strategy further guides frequency-aware learning using low-pass proxy pairs generated without labels. Evaluated on months of in-situ ARIS sonar data, SCOPE achieves a structural similarity index (SSIM) of 0.77, representing a 40% improvement over prior self-supervised denoising baselines, at bitrates down to <= 0.0118 bpp. It reduces uplink bandwidth by more than 80% while improving downstream detection. The system runs in real time, with 3.1 ms encoding on an embedded GPU and 97 ms full multi-layer decoding on the server end. SCOPE has been deployed for months in three Pacific Northwest rivers to support real-time salmon enumeration and environmental monitoring in the wild. Results demonstrate that learning frequency-structured latents enables practical, low-bitrate sonar streaming with preserved signal details under real-world deployment conditions.", "date": "2025-11-17", "categories": []}
{"id": "2409.07417", "title": "Self-Supervised One-Step Diffusion Refinement for Snapshot Compressive Imaging", "abstract": "Snapshot compressive imaging (SCI) captures multispectral images (MSIs) using a single coded two-dimensional (2-D) measurement, but reconstructing high-fidelity MSIs from these compressed inputs remains a fundamentally ill-posed challenge. While diffusion-based reconstruction methods have recently raised the bar for quality, they face critical limitations: a lack of large-scale MSI training data, adverse domain shifts from RGB-pretrained models, and inference inefficiencies due to multi-step sampling. These drawbacks restrict their practicality in real-world applications. In contrast to existing methods, which either follow costly iterative refinement or adapt subspace-based embeddings for diffusion models (e.g. DiffSCI, PSR-SCI), we introduce a fundamentally different paradigm: a self-supervised One-Step Diffusion (OSD) framework specifically designed for SCI. The key novelty lies in using a single-step diffusion refiner to correct an initial reconstruction, eliminating iterative denoising entirely while preserving generative quality. Moreover, we adopt a self-supervised equivariant learning strategy to train both the predictor and refiner directly from raw 2-D measurements, enabling generalization to unseen domains without the need for ground-truth MSI. To further address the challenge of limited MSI data, we design a band-selection-driven distillation strategy that transfers core generative priors from large-scale RGB datasets, effectively bridging the domain gap. Extensive experiments confirm that our approach sets a new benchmark, yielding PSNR gains of 3.44 dB, 1.61 dB, and 0.28 dB on the Harvard, NTIRE, and ICVL datasets, respectively, while reducing reconstruction time by 97.5%. This remarkable improvement in efficiency and adaptability makes our method a significant advancement in SCI reconstruction, combining both accuracy and practicality for real-world deployment.", "date": "2024-09-11", "categories": []}
{"id": "2507.10009", "title": "Robust Phase-Shifting Profilometry for Arbitrary Motion", "abstract": "Phase-shifting profilometry (PSP) enables high-accuracy 3D reconstruction but remains highly susceptible to object motion. Although numerous studies have explored compensation for motion-induced errors, residual inaccuracies still persist, particularly in complex motion scenarios. In this paper, we propose a robust phase-shifting profilometry for arbitrary motion (RPSP-AM), including six-degrees-of-freedom (6-DoF) motion (translation and rotation in any direction), non-rigid deformations, and multi-target movements, achieving high-fidelity motion-error-free 3D reconstruction. We categorize motion errors into two components: 1) ghosting artifacts induced by image misalignment, and 2) ripple-like distortions induced by phase deviation. To eliminate the ghosting artifacts, we perform pixel-wise image alignment based on dense optical flow tracking. To correct ripple-like distortions, we propose a high-accuracy, low-complexity image-sequential binomial self-compensation (I-BSC) method, which performs a summation of the homogeneous fringe images weighted by binomial coefficients, exponentially reducing the ripple-like distortions with a competitive computational speed compared with the traditional four-step phase-shifting method. Extensive experimental results demonstrate that, under challenging conditions such as 6-DoF motion, non-rigid deformations, and multi-target movements, the proposed RPSP-AM outperforms state-of-the-art (SoTA) methods in compensating for both ghosting artifacts and ripple-like distortions. Our approach extends the applicability of PSP to arbitrary motion scenarios, endowing it with potential for widespread adoption in fields such as robotics, industrial inspection, and medical reconstruction.", "date": "2025-07-14", "categories": []}
{"id": "2508.00453", "title": "PIF-Net: Ill-Posed Prior Guided Multispectral and Hyperspectral Image Fusion via Invertible Mamba and Fusion-Aware LoRA", "abstract": "The goal of multispectral and hyperspectral image fusion (MHIF) is to generate high-quality images that simultaneously possess rich spectral information and fine spatial details. However, due to the inherent trade-off between spectral and spatial information and the limited availability of observations, this task is fundamentally ill-posed. Previous studies have not effectively addressed the ill-posed nature caused by data misalignment. To tackle this challenge, we propose a fusion framework named PIF-Net, which explicitly incorporates ill-posed priors to effectively fuse multispectral images and hyperspectral images. To balance global spectral modeling with computational efficiency, we design a method based on an invertible Mamba architecture that maintains information consistency during feature transformation and fusion, ensuring stable gradient flow and process reversibility. Furthermore, we introduce a novel fusion module called the Fusion-Aware Low-Rank Adaptation module, which dynamically calibrates spectral and spatial features while keeping the model lightweight. Extensive experiments on multiple benchmark datasets demonstrate that PIF-Net achieves significantly better image restoration performance than current state-of-the-art methods while maintaining model efficiency.", "date": "2025-08-01", "categories": []}
{"id": "2508.12346", "title": "MBMamba: When Memory Buffer Meets Mamba for Structure-Aware Image Deblurring", "abstract": "The Mamba architecture has emerged as a promising alternative to CNNs and Transformers for image deblurring. However, its flatten-and-scan strategy often results in local pixel forgetting and channel redundancy, limiting its ability to effectively aggregate 2D spatial information. Although existing methods mitigate this by modifying the scan strategy or incorporating local feature modules, it increase computational complexity and hinder real-time performance. In this paper, we propose a structure-aware image deblurring network without changing the original Mamba architecture. Specifically, we design a memory buffer mechanism to preserve historical information for later fusion, enabling reliable modeling of relevance between adjacent features. Additionally, we introduce an Ising-inspired regularization loss that simulates the energy minimization of the physical system's \"mutual attraction\" between pixels, helping to maintain image structure and coherence. Building on this, we develop MBMamba. Experimental results show that our method outperforms state-of-the-art approaches on widely used benchmarks.", "date": "2025-08-17", "categories": []}
{"id": "2510.20029", "title": "BrainPuzzle: Hybrid Physics and Data-Driven Reconstruction for Transcranial Ultrasound Tomography", "abstract": "Ultrasound brain imaging remains challenging due to the large difference in sound speed between the skull and brain tissues and the difficulty of coupling large probes to the skull. This work aims to achieve quantitative transcranial ultrasound by reconstructing an accurate speed-of-sound (SoS) map of the brain. Traditional physics-based full-waveform inversion (FWI) is limited by weak signals caused by skull-induced attenuation, mode conversion, and phase aberration, as well as incomplete spatial coverage since full-aperture arrays are clinically impractical. In contrast, purely data-driven methods that learn directly from raw ultrasound data often fail to model the complex nonlinear and nonlocal wave propagation through bone, leading to anatomically plausible but quantitatively biased SoS maps under low signal-to-noise and sparse-aperture conditions. To address these issues, we propose BrainPuzzle, a hybrid two-stage framework that combines physical modeling with machine learning. In the first stage, reverse time migration (time-reversal acoustics) is applied to multi-angle acquisitions to produce migration fragments that preserve structural details even under low SNR. In the second stage, a transformer-based super-resolution encoder-decoder with a graph-based attention unit (GAU) fuses these fragments into a coherent and quantitatively accurate SoS image. A partial-array acquisition strategy using a movable low-count transducer set improves feasibility and coupling, while the hybrid algorithm compensates for the missing aperture. Experiments on two synthetic datasets show that BrainPuzzle achieves superior SoS reconstruction accuracy and image completeness, demonstrating its potential for advancing quantitative ultrasound brain imaging.", "date": "2025-10-22", "categories": []}
{"id": "2511.16963", "title": "Two Heads Better than One: Dual Degradation Representation for Blind Super-Resolution", "abstract": "Previous methods have demonstrated remarkable performance in single image super-resolution (SISR) tasks with known and fixed degradation (e.g., bicubic downsampling). However, when the actual degradation deviates from these assumptions, these methods may experience significant declines in performance. In this paper, we propose a Dual Branch Degradation Extractor Network to address the blind SR problem. While some blind SR methods assume noise-free degradation and others do not explicitly consider the presence of noise in the degradation model, our approach predicts two unsupervised degradation embeddings that represent blurry and noisy information. The SR network can then be adapted to blur embedding and noise embedding in distinct ways. Furthermore, we treat the degradation extractor as a regularizer to capitalize on differences between SR and HR images. Extensive experiments on several benchmarks demonstrate our method achieves SOTA performance in the blind SR problem.", "date": "2025-11-21", "categories": []}
{"id": "2511.17138", "title": "One-Step Diffusion Transformer for Controllable Real-World Image Super-Resolution", "abstract": "Recent advances in diffusion-based real-world image super-resolution (Real-ISR) have demonstrated remarkable perceptual quality, yet the balance between fidelity and controllability remains a problem: multi-step diffusion-based methods suffer from generative diversity and randomness, resulting in low fidelity, while one-step methods lose control flexibility due to fidelity-specific finetuning. In this paper, we present ODTSR, a one-step diffusion transformer based on Qwen-Image that performs Real-ISR considering fidelity and controllability simultaneously: a newly introduced visual stream receives low-quality images (LQ) with adjustable noise (Control Noise), and the original visual stream receives LQs with consistent noise (Prior Noise), forming the Noise-hybrid Visual Stream (NVS) design. ODTSR further employs Fidelity-aware Adversarial Training (FAA) to enhance controllability and achieve one-step inference. Extensive experiments demonstrate that ODTSR not only achieves state-of-the-art (SOTA) performance on generic Real-ISR, but also enables prompt controllability on challenging scenarios such as real-world scene text image super-resolution (STISR) of Chinese characters without training on specific datasets. Codes are available at https://github.com/RedMediaTech/ODTSR.", "date": "2025-11-21", "categories": []}
{"id": "2511.20991", "title": "Wavefront-Constrained Passive Obscured Object Detection", "abstract": "Accurately localizing and segmenting obscured objects from faint light patterns beyond the field of view is highly challenging due to multiple scattering and medium-induced perturbations. Most existing methods, based on real-valued modeling or local convolutional operations, are inadequate for capturing the underlying physics of coherent light propagation. Moreover, under low signal-to-noise conditions, these methods often converge to non-physical solutions, severely compromising the stability and reliability of the observation. To address these challenges, we propose a novel physics-driven Wavefront Propagating Compensation Network (WavePCNet) to simulate wavefront propagation and enhance the perception of obscured objects. This WavePCNet integrates the Tri-Phase Wavefront Complex-Propagation Reprojection (TriWCP) to incorporate complex amplitude transfer operators to precisely constrain coherent propagation behavior, along with a momentum memory mechanism to effectively suppress the accumulation of perturbations. Additionally, a High-frequency Cross-layer Compensation Enhancement is introduced to construct frequency-selective pathways with multi-scale receptive fields and dynamically model structural consistency across layers, further boosting the model's robustness and interpretability under complex environmental conditions. Extensive experiments conducted on four physically collected datasets demonstrate that WavePCNet consistently outperforms state-of-the-art methods across both accuracy and robustness.", "date": "2025-11-26", "categories": []}
{"id": "2511.22913", "title": "Optical diffraction neural networks assisted computational ghost imaging through dynamic scattering media", "abstract": "Ghost imaging leverages a single-pixel detector with no spatial resolution to acquire object echo intensity signals, which are correlated with illumination patterns to reconstruct an image. This architecture inherently mitigates scattering interference between the object and the detector but sensitive to scattering between the light source and the object. To address this challenge, we propose an optical diffraction neural networks (ODNNs) assisted ghost imaging method for imaging through dynamic scattering media. In our scheme, a set of fixed ODNNs, trained on simulated datasets, is incorporated into the experimental optical path to actively correct random distortions induced by dynamic scattering media. Experimental validation using rotating single-layer and double-layer ground glass confirms the feasibility and effectiveness of our approach. Furthermore, our scheme can also be combined with physics-prior-based reconstruction algorithms, enabling high-quality imaging under undersampled conditions. This work demonstrates a novel strategy for imaging through dynamic scattering media, which can be extended to other imaging systems.", "date": "2025-11-28", "categories": []}
{"id": "2501.12981", "title": "UniUIR: Considering Underwater Image Restoration as An All-in-One Learner", "abstract": "Existing underwater image restoration (UIR) methods generally only handle color distortion or jointly address color and haze issues, but they often overlook the more complex degradations that can occur in underwater scenes. To address this limitation, we propose a Universal Underwater Image Restoration method, termed as UniUIR, considering the complex scenario of real-world underwater mixed distortions as an all-in-one manner. To decouple degradation-specific issues and explore the inter-correlations among various degradations in UIR task, we designed the Mamba Mixture-of-Experts module. This module enables each expert to identify distinct types of degradation and collaboratively extract task-specific priors while maintaining global feature representation based on linear complexity. Building upon this foundation, to enhance degradation representation and address the task conflicts that arise when handling multiple types of degradation, we introduce the spatial-frequency prior generator. This module extracts degradation prior information in both spatial and frequency domains, and adaptively selects the most appropriate task-specific prompts based on image content, thereby improving the accuracy of image restoration. Finally, to more effectively address complex, region-dependent distortions in UIR task, we incorporate depth information derived from a large-scale pre-trained depth prediction model, thereby enabling the network to perceive and leverage depth variations across different image regions to handle localized degradation. Extensive experiments demonstrate that UniUIR can produce more attractive results across qualitative and quantitative comparisons, and shows strong generalization than state-of-the-art methods.", "date": "2025-01-22", "categories": []}
{"id": "2511.21132", "title": "DeepRFTv2: Kernel-level Learning for Image Deblurring", "abstract": "It is well-known that if a network aims to learn how to deblur, it should understand the blur process. Blurring is naturally caused by the convolution of the sharp image with the blur kernel. Thus, allowing the network to learn the blur process in the kernel-level can significantly improve the image deblurring performance. But, current deep networks are still at the pixel-level learning stage, either performing end-to-end pixel-level restoration or stage-wise pseudo kernel-level restoration, failing to enable the deblur model to understand the essence of the blur. To this end, we propose Fourier Kernel Estimator (FKE), which considers the activation operation in Fourier space and converts the convolution problem in the spatial domain to a multiplication problem in Fourier space. Our FKE, jointly optimized with the deblur model, enables the network to learn the kernel-level blur process with low complexity and without any additional supervision. Furthermore, we change the convolution object of the kernel from ``image\" to network extracted ``feature\", whose rich semantic and structural information is more suitable to blur process learning. With the convolution of the feature and the estimated kernel, our model can learn the essence of blur in kernel-level. To further improve the efficiency of feature extraction, we design a decoupled multi-scale architecture with multiple hierarchical sub-unets with a reversible strategy, which allows better multi-scale encoding and decoding in low training memory. Extensive experiments indicate that our method achieves state-of-the-art motion deblurring results and show potential for handling other kernel-related problems. Analysis also shows our kernel estimator is able to learn physically meaningful kernels. The code will be available at https://github.com/DeepMed-Lab-ECNU/Single-Image-Deblur.", "date": "2025-11-26", "categories": []}
{"id": "2412.11578", "title": "DVP-MVS: Synergize Depth-Edge and Visibility Prior for Multi-View Stereo", "abstract": "Patch deformation-based methods have recently exhibited substantial effectiveness in multi-view stereo, due to the incorporation of deformable and expandable perception to reconstruct textureless areas. However, such approaches typically focus on exploring correlative reliable pixels to alleviate match ambiguity during patch deformation, but ignore the deformation instability caused by mistaken edge-skipping and visibility occlusion, leading to potential estimation deviation. To remedy the above issues, we propose DVP-MVS, which innovatively synergizes depth-edge aligned and cross-view prior for robust and visibility-aware patch deformation. Specifically, to avoid unexpected edge-skipping, we first utilize Depth Anything V2 followed by the Roberts operator to initialize coarse depth and edge maps respectively, both of which are further aligned through an erosion-dilation strategy to generate fine-grained homogeneous boundaries for guiding patch deformation. In addition, we reform view selection weights as visibility maps and restore visible areas by cross-view depth reprojection, then regard them as cross-view prior to facilitate visibility-aware patch deformation. Finally, we improve propagation and refinement with multi-view geometry consistency by introducing aggregated visible hemispherical normals based on view selection and local projection depth differences based on epipolar lines, respectively. Extensive evaluations on ETH3D and Tanks & Temples benchmarks demonstrate that our method can achieve state-of-the-art performance with excellent robustness and generalization.", "date": "2024-12-16", "categories": []}
{"id": "2504.14131", "title": "Transforming Hyperspectral Images Into Chemical Maps: A Novel End-to-End Deep Learning Approach", "abstract": "Current approaches to chemical map generation from hyperspectral images are based on models such as partial least squares (PLS) regression, generating pixel-wise predictions that do not consider spatial context and suffer from a high degree of noise. This study proposes an end-to-end deep learning approach using a modified version of U-Net and a custom loss function to directly obtain chemical maps from hyperspectral images, skipping all intermediate steps required for traditional pixel-wise analysis. This study compares the U-Net with the traditional PLS regression on a real dataset of pork belly samples with associated mean fat reference values. The U-Net obtains a test set root mean squared error that is 7% lower than that of PLS regression on the task of mean fat prediction. At the same time, U-Net generates fine detail chemical maps where 99.91% of the variance is spatially correlated. Conversely, only 2.37% of the variance in the PLS-generated chemical maps is spatially correlated, indicating that each pixel-wise prediction is largely independent of neighboring pixels. Additionally, while the PLS-generated chemical maps contain predictions far beyond the physically possible range of 0-100%, U-Net learns to stay inside this range. Thus, the findings of this study indicate that U-Net is superior to PLS for chemical map generation.", "date": "2025-04-19", "categories": []}
{"id": "2504.19687", "title": "Prompt Guiding Multi-Scale Adaptive Sparse Representation-driven Network for Low-Dose CT MAR", "abstract": "Low-dose CT (LDCT) is capable of reducing X-ray radiation exposure, but it will potentially degrade image quality, even yields metal artifacts at the case of metallic implants. For simultaneous LDCT reconstruction and metal artifact reduction (LDMAR), existing deep learning-based efforts face two main limitations: i) the network design neglects multi-scale and within-scale information; ii) training a distinct model for each dose necessitates significant storage space for multiple doses. To fill these gaps, we propose a prompt guiding multi-scale adaptive sparse representation-driven network, abbreviated as PMSRNet, for LDMAR task. Specifically, we construct PMSRNet inspired from multi-scale sparsifying frames, and it can simultaneously employ within-scale characteristics and cross-scale complementarity owing to an elaborated prompt guiding scale-adaptive threshold generator (PSATG) and a built multi-scale coefficient fusion module (MSFuM). The PSATG can adaptively capture multiple contextual information to generate more faithful thresholds, achieved by fusing features from local, regional, and global levels. Furthermore, we elaborate a model interpretable dual domain LDMAR framework called PDuMSRNet, and train single model with a prompt guiding strategy for multiple dose levels. We build a prompt guiding module, whose input contains dose level, metal mask and input instance, to provide various guiding information, allowing a single model to accommodate various CT dose settings. Extensive experiments at various dose levels demonstrate that the proposed methods outperform the state-of-the-art LDMAR methods.", "date": "2025-04-28", "categories": []}
{"id": "2505.19385", "title": "Advancing Limited-Angle CT Reconstruction Through Diffusion-Based Sinogram Completion", "abstract": "Limited Angle Computed Tomography (LACT) often faces significant challenges due to missing angular information. Unlike previous methods that operate in the image domain, we propose a new method that focuses on sinogram inpainting. We leverage MR-SDEs, a variant of diffusion models that characterize the diffusion process with mean-reverting stochastic differential equations, to fill in missing angular data at the projection level. Furthermore, by combining distillation with constraining the output of the model using the pseudo-inverse of the inpainting matrix, the diffusion process is accelerated and done in a step, enabling efficient and accurate sinogram completion. A subsequent post-processing module back-projects the inpainted sinogram into the image domain and further refines the reconstruction, effectively suppressing artifacts while preserving critical structural details. Quantitative experimental results demonstrate that the proposed method achieves state-of-the-art performance in both perceptual and fidelity quality, offering a promising solution for LACT reconstruction in scientific and clinical applications.", "date": "2025-05-26", "categories": []}
{"id": "2511.19539", "title": "PhysDNet: Physics-Guided Decomposition Network of Side-Scan Sonar Imagery", "abstract": "Side-scan sonar (SSS) imagery is widely used for seafloor mapping and underwater remote sensing, yet the measured intensity is strongly influenced by seabed reflectivity, terrain elevation, and acoustic path loss. This entanglement makes the imagery highly view-dependent and reduces the robustness of downstream analysis. In this letter, we present PhysDNet, a physics-guided multi-branch network that decouples SSS images into three interpretable fields: seabed reflectivity, terrain elevation, and propagation loss. By embedding the Lambertian reflection model, PhysDNet reconstructs sonar intensity from these components, enabling self-supervised training without ground-truth annotations. Experiments show that the decomposed representations preserve stable geological structures, capture physically consistent illumination and attenuation, and produce reliable shadow maps. These findings demonstrate that physics-guided decomposition provides a stable and interpretable domain for SSS analysis, improving both physical consistency and downstream tasks such as registration and shadow interpretation.", "date": "2025-11-24", "categories": []}
{"id": "2511.19941", "title": "Optimize Flip Angle Schedules In MR Fingerprinting Using Reinforcement Learning", "abstract": "Magnetic Resonance Fingerprinting (MRF) leverages transient-state signal dynamics generated by the tunable acquisition parameters, making the design of an optimal, robust sequence a complex, high-dimensional sequential decision problem, such as optimizing one of the key parameters, flip angle. Reinforcement learning (RL) offers a promising approach to automate parameter selection, to optimize pulse sequences that maximize the distinguishability of fingerprints across the parameter space. In this work, we introduce an RL framework for optimizing the flip-angle schedule in MRF and demonstrate a learned schedule exhibiting non-periodic patterns that enhances fingerprint separability. Additionally, an interesting observation is that the RL-optimized schedule may enable a reduction in the number of repetition time, potentially accelerate MRF acquisitions.", "date": "2025-11-25", "categories": []}
{"id": "2511.20296", "title": "Prompting Lipschitz-constrained network for multiple-in-one sparse-view CT reconstruction", "abstract": "Despite significant advancements in deep learning-based sparse-view computed tomography (SVCT) reconstruction algorithms, these methods still encounter two primary limitations: (i) It is challenging to explicitly prove that the prior networks of deep unfolding algorithms satisfy Lipschitz constraints due to their empirically designed nature. (ii) The substantial storage costs of training a separate model for each setting in the case of multiple views hinder practical clinical applications. To address these issues, we elaborate an explicitly provable Lipschitz-constrained network, dubbed LipNet, and integrate an explicit prompt module to provide discriminative knowledge of different sparse sampling settings, enabling the treatment of multiple sparse view configurations within a single model. Furthermore, we develop a storage-saving deep unfolding framework for multiple-in-one SVCT reconstruction, termed PromptCT, which embeds LipNet as its prior network to ensure the convergence of its corresponding iterative algorithm. In simulated and real data experiments, PromptCT outperforms benchmark reconstruction algorithms in multiple-in-one SVCT reconstruction, achieving higher-quality reconstructions with lower storage costs. On the theoretical side, we explicitly demonstrate that LipNet satisfies boundary property, further proving its Lipschitz continuity and subsequently analyzing the convergence of the proposed iterative algorithms. The data and code are publicly available at https://github.com/shibaoshun/PromptCT.", "date": "2025-11-25", "categories": []}
{"id": "2511.20551", "title": "Time-Domain Linear Model-based Framework for Passive Acoustic Mapping of Cavitation Activity", "abstract": "Passive acoustic mapping enables the spatial mapping and temporal monitoring of cavitation activity, playing a crucial role in therapeutic ultrasound applications. Most conventional beamforming methods, whether implemented in the time or frequency domains, suffer from limited axial resolution due to the absence of a reference emission onset time. While frequency-domain methods, the most efficient of which are based on the cross-spectral matrix, require long signals for accurate estimation, time-domain methods typically achieve lower spatial resolution. To address these limitations, we propose a linear model-based beamforming framework fully formulated in the time domain. The linear forward model relates a discretized spatiotemporal distribution of cavitation activity to the temporal signals recorded by a probe, explicitly accounting for time-of-flight delays dictated by the acquisition geometry. This model is then inverted using regularization techniques that exploit prior knowledge of cavitation activity in both spatial and temporal domains. Experimental results show that the proposed framework achieves enhanced or competitive cavitation map quality while using only 20\\% of the data typically required by frequency-domain methods. This highlights the substantial gain in data efficiency and the flexibility of our spatiotemporal regularization to adapt to diverse passive cavitation scenarios, outperforming state-of-the-art techniques.", "date": "2025-11-25", "categories": []}
{"id": "2409.13930", "title": "RN-SDEs: Limited-Angle CT Reconstruction with Residual Null-Space Diffusion Stochastic Differential Equations", "abstract": "Computed tomography is a widely used imaging modality with applications ranging from medical imaging to material analysis. One major challenge arises from the lack of scanning information at certain angles, resulting in distortion or artifacts in the reconstructed images. This is referred to as the Limited Angle Computed Tomography (LACT) reconstruction problem. To address this problem, we propose the use of Residual Null-Space Diffusion Stochastic Differential Equations (RN-SDEs), which are a variant of diffusion models that characterize the diffusion process with mean-reverting (MR) stochastic differential equations. To demonstrate the generalizability of RN-SDEs, we conducted experiments with two different LACT datasets, ChromSTEM and C4KC-KiTS. Through extensive experiments, we demonstrate that by leveraging learned MR-SDEs as a prior and emphasizing data consistency using Range-Null Space Decomposition (RNSD) based rectification, we can recover high-quality images from severely degraded ones and achieve state-of-the-art performance in most LACT tasks. Additionally, we present a quantitative comparison of RN-SDE with other networks, in terms of computational complexity and runtime efficiency, highlighting the superior effectiveness of our proposed approach.", "date": "2024-09-20", "categories": []}
{"id": "2511.13533", "title": "Minimax Multi-Target Conformal Prediction with Applications to Imaging Inverse Problems", "abstract": "In ill-posed imaging inverse problems, uncertainty quantification remains a fundamental challenge, especially in safety-critical applications. Recently, conformal prediction has been used to quantify the uncertainty that the inverse problem contributes to downstream tasks like image classification, image quality assessment, fat mass quantification, etc. While existing works handle only a scalar estimation target, practical applications often involve multiple targets. In response, we propose an asymptotically minimax approach to multi-target conformal prediction that provides tight prediction intervals while ensuring joint marginal coverage. We then outline how our minimax approach can be applied to multi-metric blind image quality assessment, multi-task uncertainty quantification, and multi-round measurement acquisition. Finally, we numerically demonstrate the benefits of our minimax method, relative to existing multi-target conformal prediction methods, using both synthetic and magnetic resonance imaging (MRI) data. Code is available at https://github.com/jwen307/multi_target_minimax.", "date": "2025-11-17", "categories": []}
{"id": "2511.18028", "title": "MambaX: Image Super-Resolution with State Predictive Control", "abstract": "Image super-resolution (SR) is a critical technology for overcoming the inherent hardware limitations of sensors. However, existing approaches mainly focus on directly enhancing the final resolution, often neglecting effective control over error propagation and accumulation during intermediate stages. Recently, Mamba has emerged as a promising approach that can represent the entire reconstruction process as a state sequence with multiple nodes, allowing for intermediate intervention. Nonetheless, its fixed linear mapper is limited by a narrow receptive field and restricted flexibility, which hampers its effectiveness in fine-grained images. To address this, we created a nonlinear state predictive control model \\textbf{MambaX} that maps consecutive spectral bands into a latent state space and generalizes the SR task by dynamically learning the nonlinear state parameters of control equations. Compared to existing sequence models, MambaX 1) employs dynamic state predictive control learning to approximate the nonlinear differential coefficients of state-space models; 2) introduces a novel state cross-control paradigm for multimodal SR fusion; and 3) utilizes progressive transitional learning to mitigate heterogeneity caused by domain and modality shifts. Our evaluation demonstrates the superior performance of the dynamic spectrum-state representation model in both single-image SR and multimodal fusion-based SR tasks, highlighting its substantial potential to advance spectrally generalized modeling across arbitrary dimensions and modalities.", "date": "2025-11-22", "categories": []}
{"id": "2511.18037", "title": "Hybrid Event Frame Sensors: Modeling, Calibration, and Simulation", "abstract": "Event frame hybrid sensors integrate an Active Pixel Sensor (APS) and an Event Vision Sensor (EVS) within a single chip, combining the high dynamic range and low latency of the EVS with the rich spatial intensity information from the APS. While this tight integration offers compact, temporally precise imaging, the complex circuit architecture introduces non-trivial noise patterns that remain poorly understood and unmodeled. In this work, we present the first unified, statistics-based imaging noise model that jointly describes the noise behavior of APS and EVS pixels. Our formulation explicitly incorporates photon shot noise, dark current noise, fixed-pattern noise, and quantization noise, and links EVS noise to illumination level and dark current. Based on this formulation, we further develop a calibration pipeline to estimate noise parameters from real data and offer a detailed analysis of both APS and EVS noise behaviors. Finally, we propose HESIM, a statistically grounded simulator that generates RAW frames and events under realistic, jointly calibrated noise statistics. Experiments on two hybrid sensors validate our model across multiple imaging tasks (e.g., video frame interpolation and deblurring), demonstrating strong transfer from simulation to real data.", "date": "2025-11-22", "categories": []}
{"id": "2511.18232", "title": "Parallel qMRI Reconstruction from 4x Accelerated Acquisitions", "abstract": "Magnetic Resonance Imaging (MRI) acquisitions require extensive scan times, limiting patient throughput and increasing susceptibility to motion artifacts. Accelerated parallel MRI techniques reduce acquisition time by undersampling k-space data, but require robust reconstruction methods to recover high-quality images. Traditional approaches like SENSE require both undersampled k-space data and pre-computed coil sensitivity maps. We propose an end-to-end deep learning framework that jointly estimates coil sensitivity maps and reconstructs images from only undersampled k-space measurements at 4x acceleration. Our two-module architecture consists of a Coil Sensitivity Map (CSM) estimation module and a U-Net-based MRI reconstruction module. We evaluate our method on multi-coil brain MRI data from 10 subjects with 8 echoes each, using 2x SENSE reconstructions as ground truth. Our approach produces visually smoother reconstructions compared to conventional SENSE output, achieving comparable visual quality despite lower PSNR/SSIM metrics. We identify key challenges including spatial misalignment between different acceleration factors and propose future directions for improved reconstruction quality.", "date": "2025-11-23", "categories": []}
{"id": "2511.18534", "title": "HiFi-MambaV2: Hierarchical Shared-Routed MoE for High-Fidelity MRI Reconstruction", "abstract": "Reconstructing high-fidelity MR images from undersampled k-space data requires recovering high-frequency details while maintaining anatomical coherence. We present HiFi-MambaV2, a hierarchical shared-routed Mixture-of-Experts (MoE) Mamba architecture that couples frequency decomposition with content-adaptive computation. The model comprises two core components: (i) a separable frequency-consistent Laplacian pyramid (SF-Lap) that delivers alias-resistant, stable low- and high-frequency streams; and (ii) a hierarchical shared-routed MoE that performs per-pixel top-1 sparse dispatch to shared experts and local routers, enabling effective specialization with stable cross-depth behavior. A lightweight global context path is fused into an unrolled, data-consistency-regularized backbone to reinforce long-range reasoning and preserve anatomical coherence. Evaluated on fastMRI, CC359, ACDC, M4Raw, and Prostate158, HiFi-MambaV2 consistently outperforms CNN-, Transformer-, and prior Mamba-based baselines in PSNR, SSIM, and NMSE across single- and multi-coil settings and multiple acceleration factors, consistently surpassing consistent improvements in high-frequency detail and overall structural fidelity. These results demonstrate that HiFi-MambaV2 enables reliable and robust MRI reconstruction.", "date": "2025-11-23", "categories": []}
{"id": "2511.18591", "title": "Zero-Reference Joint Low-Light Enhancement and Deblurring via Visual Autoregressive Modeling with VLM-Derived Modulation", "abstract": "Real-world dark images commonly exhibit not only low visibility and contrast but also complex noise and blur, posing significant restoration challenges. Existing methods often rely on paired data or fail to model dynamic illumination and blur characteristics, leading to poor generalization. To tackle this, we propose a generative framework based on visual autoregressive (VAR) modeling, guided by perceptual priors from the vision-language model (VLM). Specifically, to supply informative conditioning cues for VAR models, we deploy an adaptive curve estimation scheme to modulate the diverse illumination based on VLM-derived visibility scores. In addition, we integrate dynamic and spatial-frequency-aware Rotary Positional Encodings (SF-RoPE) into VAR to enhance its ability to model structures degraded by blur. Furthermore, we propose a recursive phase-domain modulation strategy that mitigates blur-induced artifacts in the phase domain via bounded iterative refinement guided by VLM-assessed blur scores. Our framework is fully unsupervised and achieves state-of-the-art performance on benchmark datasets.", "date": "2025-11-23", "categories": []}
{"id": "2511.18667", "title": "Equivariant Deep Equilibrium Models for Imaging Inverse Problems", "abstract": "Equivariant imaging (EI) enables training signal reconstruction models without requiring ground truth data by leveraging signal symmetries. Deep equilibrium models (DEQs) are a powerful class of neural networks where the output is a fixed point of a learned operator. However, training DEQs with complex EI losses requires implicit differentiation through fixed-point computations, whose implementation can be challenging. We show that backpropagation can be implemented modularly, simplifying training. Experiments demonstrate that DEQs trained with implicit differentiation outperform those trained with Jacobian-free backpropagation and other baseline methods. Additionally, we find evidence that EI-trained DEQs approximate the proximal map of an invariant prior.", "date": "2025-11-24", "categories": []}
{"id": "2511.17253", "title": "Blind Deconvolution for Color Images Using Normalized Quaternion Kernels", "abstract": "In this work, we address the challenging problem of blind deconvolution for color images. Existing methods often convert color images to grayscale or process each color channel separately, which overlooking the relationships between color channels. To handle this issue, we formulate a novel quaternion fidelity term designed specifically for color image blind deconvolution. This fidelity term leverages the properties of quaternion convolution kernel, which consists of four kernels: one that functions similarly to a non-negative convolution kernel to capture the overall blur, and three additional convolution kernels without constraints corresponding to red, green and blue channels respectively model their unknown interdependencies. In order to preserve image intensity, we propose to use the normalized quaternion kernel in the blind deconvolution process. Extensive experiments on real datasets of blurred color images show that the proposed method effectively removes artifacts and significantly improves deblurring effect, demonstrating its potential as a powerful tool for color image deconvolution.", "date": "2025-11-21", "categories": []}
{"id": "2511.17492", "title": "EvDiff: High Quality Video with an Event Camera", "abstract": "As neuromorphic sensors, event cameras asynchronously record changes in brightness as streams of sparse events with the advantages of high temporal resolution and high dynamic range. Reconstructing intensity images from events is a highly ill-posed task due to the inherent ambiguity of absolute brightness. Early methods generally follow an end-to-end regression paradigm, directly mapping events to intensity frames in a deterministic manner. While effective to some extent, these approaches often yield perceptually inferior results and struggle to scale up in model capacity and training data. In this work, we propose EvDiff, an event-based diffusion model that follows a surrogate training framework to produce high-quality videos. To reduce the heavy computational cost of high-frame-rate video generation, we design an event-based diffusion model that performs only a single forward diffusion step, equipped with a temporally consistent EvEncoder. Furthermore, our novel Surrogate Training Framework eliminates the dependence on paired event-image datasets, allowing the model to leverage large-scale image datasets for higher capacity. The proposed EvDiff is capable of generating high-quality colorful videos solely from monochromatic event streams. Experiments on real-world datasets demonstrate that our method strikes a sweet spot between fidelity and realism, outperforming existing approaches on both pixel-level and perceptual metrics.", "date": "2025-11-21", "categories": []}
{"id": "2509.03475", "title": "Solving Imaging Inverse Problems Using Plug-and-Play Denoisers: Regularization and Optimization Perspectives", "abstract": "Inverse problems lie at the heart of modern imaging science, with broad applications in areas such as medical imaging, remote sensing, and microscopy. Recent years have witnessed a paradigm shift in solving imaging inverse problems, where data-driven regularizers are used increasingly, leading to remarkably high-fidelity reconstruction. A particularly notable approach for data-driven regularization is to use learned image denoisers as implicit priors in iterative image reconstruction algorithms. This chapter presents a comprehensive overview of this powerful and emerging class of algorithms, commonly referred to as plug-and-play (PnP) methods. We begin by providing a brief background on image denoising and inverse problems, followed by a short review of traditional regularization strategies. We then explore how proximal splitting algorithms, such as the alternating direction method of multipliers (ADMM) and proximal gradient descent (PGD), can naturally accommodate learned denoisers in place of proximal operators, and under what conditions such replacements preserve convergence. The role of Tweedie's formula in connecting optimal Gaussian denoisers and score estimation is discussed, which lays the foundation for regularization-by-denoising (RED) and more recent diffusion-based posterior sampling methods. We discuss theoretical advances regarding the convergence of PnP algorithms, both within the RED and proximal settings, emphasizing the structural assumptions that the denoiser must satisfy for convergence, such as non-expansiveness, Lipschitz continuity, and local homogeneity. We also address practical considerations in algorithm design, including choices of denoiser architecture and acceleration strategies.", "date": "2025-09-03", "categories": []}
{"id": "2511.14270", "title": "Gaussian Splatting-based Low-Rank Tensor Representation for Multi-Dimensional Image Recovery", "abstract": "Tensor singular value decomposition (t-SVD) is a promising tool for multi-dimensional image representation, which decomposes a multi-dimensional image into a latent tensor and an accompanying transform matrix. However, two critical limitations of t-SVD methods persist: (1) the approximation of the latent tensor (e.g., tensor factorizations) is coarse and fails to accurately capture spatial local high-frequency information; (2) The transform matrix is composed of fixed basis atoms (e.g., complex exponential atoms in DFT and cosine atoms in DCT) and cannot precisely capture local high-frequency information along the mode-3 fibers. To address these two limitations, we propose a Gaussian Splatting-based Low-rank tensor Representation (GSLR) framework, which compactly and continuously represents multi-dimensional images. Specifically, we leverage tailored 2D Gaussian splatting and 1D Gaussian splatting to generate the latent tensor and transform matrix, respectively. The 2D and 1D Gaussian splatting are indispensable and complementary under this representation framework, which enjoys a powerful representation capability, especially for local high-frequency information. To evaluate the representation ability of the proposed GSLR, we develop an unsupervised GSLR-based multi-dimensional image recovery model. Extensive experiments on multi-dimensional image recovery demonstrate that GSLR consistently outperforms state-of-the-art methods, particularly in capturing local high-frequency information.", "date": "2025-11-18", "categories": []}
{"id": "2511.15022", "title": "Complex-Valued 2D Gaussian Representation for Computer-Generated Holography", "abstract": "We propose a new hologram representation based on structured complex-valued 2D Gaussian primitives, which replaces per-pixel information storage and reduces the parameter search space by up to 10:1. To enable end-to-end training, we develop a differentiable rasterizer for our representation, integrated with a GPU-optimized light propagation kernel in free space. Our extensive experiments show that our method achieves up to 2.5x lower VRAM usage and 50% faster optimization while producing higher-fidelity reconstructions than existing methods. We further introduce a conversion procedure that adapts our representation to practical hologram formats, including smooth and random phase-only holograms. Our experiments show that this procedure can effectively suppress noise artifacts observed in previous methods. By reducing the hologram parameter search space, our representation enables a more scalable hologram estimation in the next-generation computer-generated holography systems.", "date": "2025-11-19", "categories": []}
{"id": "2308.02340", "title": "Generative Priors for MRI Reconstruction Trained from Magnitude-Only Images Using Phase Augmentation", "abstract": "Purpose: In this work, we present a workflow to construct generic and robust generative image priors from magnitude-only images. The priors can then be used for regularization in reconstruction to improve image quality. Methods: The workflow begins with the preparation of training datasets from magnitude-only MR images. This dataset is then augmented with phase information and used to train generative priors of complex images. Finally, trained priors are evaluated using both linear and nonlinear reconstruction for compressed sensing parallel imaging with various undersampling schemes. Results: The results of our experiments demonstrate that priors trained on complex images outperform priors trained only on magnitude images. Additionally, a prior trained on a larger dataset exhibits higher robustness. Finally, we show that the generative priors are superior to L1 -wavelet regularization for compressed sensing parallel imaging with high undersampling. Conclusion: These findings stress the importance of incorporating phase information and leveraging large datasets to raise the performance and reliability of the generative priors for MRI reconstruction. Phase augmentation makes it possible to use existing image databases for training.", "date": "2023-08-04", "categories": []}
{"id": "2505.17644", "title": "Towards Prospective Medical Image Reconstruction via Knowledge-Informed Dynamic Optimal Transport", "abstract": "Medical image reconstruction from measurement data is a vital but challenging inverse problem. Deep learning approaches have achieved promising results, but often requires paired measurement and high-quality images, which is typically simulated through a forward model, i.e., retrospective reconstruction. However, training on simulated pairs commonly leads to performance degradation on real prospective data due to the retrospective-to-prospective gap caused by incomplete imaging knowledge in simulation. To address this challenge, this paper introduces imaging Knowledge-Informed Dynamic Optimal Transport (KIDOT), a novel dynamic optimal transport framework with optimality in the sense of preserving consistency with imaging physics in transport, that conceptualizes reconstruction as finding a dynamic transport path. KIDOT learns from unpaired data by modeling reconstruction as a continuous evolution path from measurements to images, guided by an imaging knowledge-informed cost function and transport equation. This dynamic and knowledge-aware approach enhances robustness and better leverages unpaired data while respecting acquisition physics. Theoretically, we demonstrate that KIDOT naturally generalizes dynamic optimal transport, ensuring its mathematical rationale and solution existence. Extensive experiments on MRI and CT reconstruction demonstrate KIDOT's superior performance.", "date": "2025-05-23", "categories": []}
{"id": "2507.02686", "title": "Learning few-step posterior samplers by unfolding and distillation of diffusion models", "abstract": "Diffusion models (DMs) have emerged as powerful image priors in Bayesian computational imaging. Two primary strategies have been proposed for leveraging DMs in this context: Plug-and-Play methods, which are zero-shot and highly flexible but rely on approximations; and specialized conditional DMs, which achieve higher accuracy and faster inference for specific tasks through supervised training. In this work, we introduce a novel framework that integrates deep unfolding and model distillation to transform a DM image prior into a few-step conditional model for posterior sampling. A central innovation of our approach is the unfolding of a Markov chain Monte Carlo (MCMC) algorithm - specifically, the recently proposed LATINO Langevin sampler (Spagnoletti et al., 2025) - representing the first known instance of deep unfolding applied to a Monte Carlo sampling scheme. We demonstrate our proposed unfolded and distilled samplers through extensive experiments and comparisons with the state of the art, where they achieve excellent accuracy and computational efficiency, while retaining the flexibility to adapt to variations in the forward model at inference time.", "date": "2025-07-03", "categories": []}
{"id": "2507.11461", "title": "Deep Equilibrium models for Poisson Imaging Inverse problems via Mirror Descent", "abstract": "Deep Equilibrium Models (DEQs) are implicit neural networks with fixed points, which have recently gained attention for learning image regularization functionals, particularly in settings involving Gaussian fidelities, where assumptions on the forward operator ensure contractiveness of standard (proximal) Gradient Descent operators. In this work, we extend the application of DEQs to Poisson inverse problems, where the data fidelity term is more appropriately modeled by the Kullback--Leibler divergence. To this end, we introduce a novel DEQ formulation based on Mirror Descent defined in terms of a tailored non-Euclidean geometry that naturally adapts with the structure of the data term. This enables the learning of neural regularizers within a principled training framework. We derive sufficient conditions and establish refined convergence results based on the Kurdyka--Lojasiewicz framework for subanalytic functions with non-closed domains to guarantee the convergence of the learned reconstruction scheme and propose computational strategies that enable both efficient training and parameter-free inference. Numerical experiments show that our method outperforms traditional model-based approaches and it is comparable to the performance of Bregman Plug-and-Play methods, while mitigating their typical drawbacks, such as time-consuming tuning of hyper-parameters. The code is publicly available at https://github.com/christiandaniele/DEQ-MD.", "date": "2025-07-15", "categories": []}
{"id": "2511.13015", "title": "Geometry Meets Light: Leveraging Geometric Priors for Universal Photometric Stereo under Limited Multi-Illumination Cues", "abstract": "Universal Photometric Stereo is a promising approach for recovering surface normals without strict lighting assumptions. However, it struggles when multi-illumination cues are unreliable, such as under biased lighting or in shadows or self-occluded regions of complex in-the-wild scenes. We propose GeoUniPS, a universal photometric stereo network that integrates synthetic supervision with high-level geometric priors from large-scale 3D reconstruction models pretrained on massive in-the-wild data. Our key insight is that these 3D reconstruction models serve as visual-geometry foundation models, inherently encoding rich geometric knowledge of real scenes. To leverage this, we design a Light-Geometry Dual-Branch Encoder that extracts both multi-illumination cues and geometric priors from the frozen 3D reconstruction model. We also address the limitations of the conventional orthographic projection assumption by introducing the PS-Perp dataset with realistic perspective projection to enable learning of spatially varying view directions. Extensive experiments demonstrate that GeoUniPS delivers state-of-the-arts performance across multiple datasets, both quantitatively and qualitatively, especially in the complex in-the-wild scenes.", "date": "2025-11-17", "categories": []}
{"id": "2511.14310", "title": "Iterative Diffusion-Refined Neural Attenuation Fields for Multi-Source Stationary CT Reconstruction: NAF Meets Diffusion Model", "abstract": "Multi-source stationary computed tomography (CT) has recently attracted attention for its ability to achieve rapid image reconstruction, making it suitable for time-sensitive clinical and industrial applications. However, practical systems are often constrained by ultra-sparse-view sampling, which significantly degrades reconstruction quality. Traditional methods struggle under ultra-sparse-view settings, where interpolation becomes inaccurate and the resulting reconstructions are unsatisfactory. To address this challenge, this study proposes Diffusion-Refined Neural Attenuation Fields (Diff-NAF), an iterative framework tailored for multi-source stationary CT under ultra-sparse-view conditions. Diff-NAF combines a Neural Attenuation Field representation with a dual-branch conditional diffusion model. The process begins by training an initial NAF using ultra-sparse-view projections. New projections are then generated through an Angle-Prior Guided Projection Synthesis strategy that exploits inter view priors, and are subsequently refined by a Diffusion-driven Reuse Projection Refinement Module. The refined projections are incorporated as pseudo-labels into the training set for the next iteration. Through iterative refinement, Diff-NAF progressively enhances projection completeness and reconstruction fidelity under ultra-sparse-view conditions, ultimately yielding high-quality CT reconstructions. Experimental results on multiple simulated 3D CT volumes and real projection data demonstrate that Diff-NAF achieves the best performance under ultra-sparse-view conditions.", "date": "2025-11-18", "categories": []}
{"id": "2501.14659", "title": "GUSLO: General and Unified Structured Light Optimization", "abstract": "Structured light (SL) 3D reconstruction captures the precise surface shape of objects, providing high-accuracy 3D data essential for industrial inspection and cultural heritage digitization. However, existing methods suffer from two key limitations: reliance on scene-specific calibration with manual parameter tuning, and optimization frameworks tailored to specific SL patterns, limiting their generalizability across varied scenarios. We propose General and Unified Structured Light Optimization (GUSLO), a novel framework addressing these issues through two coordinated innovations: (1) single-shot calibration via 2D triangulation-based interpolation that converts sparse matches into dense correspondence fields, and (2) artifact-aware photometric adaptation via explicit transfer functions, balancing generalization and color fidelity. We conduct diverse experiments covering binary, speckle, and color-coded settings. Results show that GUSLO consistently improves accuracy and cross-encoding robustness over conventional methods in challenging industrial and cultural scenarios.", "date": "2025-01-24", "categories": []}
{"id": "2505.22344", "title": "Task-Driven Implicit Representations for Automated Design of LiDAR Systems", "abstract": "Imaging system design is a complex, time-consuming, and largely manual process; LiDAR design, ubiquitous in mobile devices, autonomous vehicles, and aerial imaging platforms, adds further complexity through unique spatial and temporal sampling requirements. In this work, we propose a framework for automated, task-driven LiDAR system design under arbitrary constraints. To achieve this, we represent LiDAR configurations in a continuous six-dimensional design space and learn task-specific implicit densities in this space via flow-based generative modeling. We then synthesize new LiDAR systems by modeling sensors as parametric distributions in 6D space and fitting these distributions to our learned implicit density using expectation-maximization, enabling efficient, constraint-aware LiDAR system design. We validate our method on diverse tasks in 3D vision, enabling automated LiDAR system design across real-world-inspired applications in face scanning, robotic tracking, and object detection.", "date": "2025-05-28", "categories": []}
{"id": "2511.12066", "title": "DCA-LUT: Deep Chromatic Alignment with 5D LUT for Purple Fringing Removal", "abstract": "Purple fringing, a persistent artifact caused by Longitudinal Chromatic Aberration (LCA) in camera lenses, has long degraded the clarity and realism of digital imaging. Traditional solutions rely on complex and expensive apochromatic (APO) lens hardware and the extraction of handcrafted features, ignoring the data-driven approach. To fill this gap, we introduce DCA-LUT, the first deep learning framework for purple fringing removal. Inspired by the physical root of the problem, the spatial misalignment of RGB color channels due to lens dispersion, we introduce a novel Chromatic-Aware Coordinate Transformation (CA-CT) module, learning an image-adaptive color space to decouple and isolate fringing into a dedicated dimension. This targeted separation allows the network to learn a precise ``purple fringe channel\", which then guides the accurate restoration of the luminance channel. The final color correction is performed by a learned 5D Look-Up Table (5D LUT), enabling efficient and powerful% non-linear color mapping. To enable robust training and fair evaluation, we constructed a large-scale synthetic purple fringing dataset (PF-Synth). Extensive experiments in synthetic and real-world datasets demonstrate that our method achieves state-of-the-art performance in purple fringing removal.", "date": "2025-11-15", "categories": []}
{"id": "2511.12556", "title": "DLMMPR:Deep Learning-based Measurement Matrix for Phase Retrieval", "abstract": "This paper pioneers the integration of learning optimization into measurement matrix design for phase retrieval. We introduce the Deep Learning-based Measurement Matrix for Phase Retrieval (DLMMPR) algorithm, which parameterizes the measurement matrix within an end-to-end deep learning architecture. Synergistically augmented with subgradient descent and proximal mapping modules for robust recovery, DLMMPR's efficacy is decisively confirmed through comprehensive empirical validation across diverse noise regimes. Benchmarked against DeepMMSE and PrComplex, our method yields substantial gains in PSNR and SSIM, underscoring its superiority.", "date": "2025-11-16", "categories": []}
{"id": "2511.12939", "title": "Semi-Supervised High Dynamic Range Image Reconstructing via Bi-Level Uncertain Area Masking", "abstract": "Reconstructing high dynamic range (HDR) images from low dynamic range (LDR) bursts plays an essential role in the computational photography. Impressive progress has been achieved by learning-based algorithms which require LDR-HDR image pairs. However, these pairs are hard to obtain, which motivates researchers to delve into the problem of annotation-efficient HDR image reconstructing: how to achieve comparable performance with limited HDR ground truths (GTs). This work attempts to address this problem from the view of semi-supervised learning where a teacher model generates pseudo HDR GTs for the LDR samples without GTs and a student model learns from pseudo GTs. Nevertheless, the confirmation bias, i.e., the student may learn from the artifacts in pseudo HDR GTs, presents an impediment. To remove this impediment, an uncertainty-based masking process is proposed to discard unreliable parts of pseudo GTs at both pixel and patch levels, then the trusted areas can be learned from by the student. With this novel masking process, our semi-supervised HDR reconstructing method not only outperforms previous annotation-efficient algorithms, but also achieves comparable performance with up-to-date fully-supervised methods by using only 6.7% HDR GTs.", "date": "2025-11-17", "categories": []}
{"id": "2511.13185", "title": "Uncertainty-aware Physics-informed Neural Networks for Robust CARS-to-Raman Signal Reconstruction", "abstract": "Coherent anti-Stokes Raman scattering (CARS) spectroscopy is a powerful and rapid technique widely used in medicine, material science, and chemical analyses. However, its effectiveness is hindered by the presence of a non-resonant background that interferes with and distorts the true Raman signal. Deep learning methods have been employed to reconstruct the true Raman spectrum from measured CARS data using labeled datasets. A more recent development integrates the domain knowledge of Kramers-Kronig relationships and smoothness constraints in the form of physics-informed loss functions. However, these deterministic models lack the ability to quantify uncertainty, an essential feature for reliable deployment in high-stakes scientific and biomedical applications. In this work, we evaluate and compare various uncertainty quantification (UQ) techniques within the context of CARS-to-Raman signal reconstruction. Furthermore, we demonstrate that incorporating physics-informed constraints into these models improves their calibration, offering a promising path toward more trustworthy CARS data analysis.", "date": "2025-11-17", "categories": []}
{"id": "2511.11015", "title": "SUPER Decoder Block for Reconstruction-Aware U-Net Variants", "abstract": "Skip-connected encoder-decoder architectures (U-Net variants) are widely adopted for inverse problems but still suffer from information loss, limiting recovery of fine high-frequency details. We present Selectively Suppressed Perfect Reconstruction (SUPER), which exploits the perfect reconstruction (PR) property of wavelets to prevent information degradation while selectively suppressing (SS) redundant features. Free from rigid framelet constraints, SUPER serves as a plug-and-play decoder block for diverse U-Net variants, eliminating their intrinsic reconstruction bottlenecks and enhancing representational richness. Experiments across diverse crack benchmarks, including state-of-the-art (SOTA) models, demonstrate the structural potential of the proposed SUPER Decoder Block. Maintaining comparable computational cost, SUPER enriches representational diversity through increased parameterization. In small-scale in-domain experiments on the CrackVision12K dataset, SUPER markedly improves thin-crack segmentation performance, particularly for cracks narrower than 4 px, underscoring its advantage in high-frequency dominant settings. In smartphone image denoising on SIDD, where low-frequency components prevail, SUPER still achieves a moderate gain in PSNR, confirming its robustness across low- and high-frequency regimes. These results validate its plug-and-play generality across U-Net variants, achieving high-frequency fidelity and global coherence within a unified, reconstruction-aware framework.", "date": "2025-11-14", "categories": []}
{"id": "2511.11078", "title": "SplineSplat: 3D Ray Tracing for Higher-Quality Tomography", "abstract": "We propose a method to efficiently compute tomographic projections of a 3D volume represented by a linear combination of shifted B-splines. To do so, we propose a ray-tracing algorithm that computes 3D line integrals with arbitrary projection geometries. One of the components of our algorithm is a neural network that computes the contribution of the basis functions efficiently. In our experiments, we consider well-posed cases where the data are sufficient for accurate reconstruction without the need for regularization. We achieve higher reconstruction quality than traditional voxel-based methods.", "date": "2025-11-14", "categories": []}
{"id": "2511.09734", "title": "A Fourier-Based Global Denoising Model for Smart Artifacts Removing of Microscopy Images", "abstract": "Microscopy such as Scanning Tunneling Microscopy (STM), Atomic Force Microscopy (AFM) and Scanning Electron Microscopy (SEM) are essential tools in material imaging at micro- and nanoscale resolutions to extract physical knowledge and materials structure-property relationships. However, tuning microscopy controls (e.g. scanning speed, current setpoint, tip bias etc.) to obtain a high-quality of images is a non-trivial and time-consuming effort. On the other hand, with sub-standard images, the key features are not accurately discovered due to noise and artifacts, leading to erroneous analysis. Existing denoising models mostly build on generalizing the weak signals as noises while the strong signals are enhanced as key features, which is not always the case in microscopy images, thus can completely erase a significant amount of hidden physical information. To address these limitations, we propose a global denoising model (GDM) to smartly remove artifacts of microscopy images while preserving weaker but physically important features. The proposed model is developed based on 1) first designing a two-imaging input channel of non-pair and goal specific pre-processed images with user-defined trade-off information between two channels and 2) then integrating a loss function of pixel- and fast Fourier-transformed (FFT) based on training the U-net model. We compared the proposed GDM with the non-FFT denoising model over STM-generated images of Copper(Cu) and Silicon(Si) materials, AFM-generated Pantoea sp.YR343 bio-film images and SEM-generated plastic degradation images. We believe this proposed workflow can be extended to improve other microscopy image quality and will benefit the experimentalists with the proposed design flexibility to smartly tune via domain-experts preferences.", "date": "2025-11-12", "categories": []}
{"id": "2511.09965", "title": "Equivariant Sampling for Improving Diffusion Model-based Image Restoration", "abstract": "Recent advances in generative models, especially diffusion models, have significantly improved image restoration (IR) performance. However, existing problem-agnostic diffusion model-based image restoration (DMIR) methods face challenges in fully leveraging diffusion priors, resulting in suboptimal performance. In this paper, we address the limitations of current problem-agnostic DMIR methods by analyzing their sampling process and providing effective solutions. We introduce EquS, a DMIR method that imposes equivariant information through dual sampling trajectories. To further boost EquS, we propose the Timestep-Aware Schedule (TAS) and introduce EquS$^+$. TAS prioritizes deterministic steps to enhance certainty and sampling efficiency. Extensive experiments on benchmarks demonstrate that our method is compatible with previous problem-agnostic DMIR methods and significantly boosts their performance without increasing computational costs. Our code is available at https://github.com/FouierL/EquS.", "date": "2025-11-13", "categories": []}
{"id": "2511.10166", "title": "Physically Interpretable Multi-Degradation Image Restoration via Deep Unfolding and Explainable Convolution", "abstract": "Although image restoration has advanced significantly, most existing methods target only a single type of degradation. In real-world scenarios, images often contain multiple degradations simultaneously, such as rain, noise, and haze, requiring models capable of handling diverse degradation types. Moreover, methods that improve performance through module stacking often suffer from limited interpretability. In this paper, we propose a novel interpretability-driven approach for multi-degradation image restoration, built upon a deep unfolding network that maps the iterative process of a mathematical optimization algorithm into a learnable network structure. Specifically, we employ an improved second-order semi-smooth Newton algorithm to ensure that each module maintains clear physical interpretability. To further enhance interpretability and adaptability, we design an explainable convolution module inspired by the human brain's flexible information processing and the intrinsic characteristics of images, allowing the network to flexibly leverage learned knowledge and autonomously adjust parameters for different input. The resulting tightly integrated architecture, named InterIR, demonstrates excellent performance in multi-degradation restoration while remaining highly competitive on single-degradation tasks.", "date": "2025-11-13", "categories": []}
{"id": "2507.03184", "title": "EvRWKV: A Continuous Interactive RWKV Framework for Effective Event-Guided Low-Light Image Enhancement", "abstract": "Event cameras offer significant potential for Low-light Image Enhancement (LLIE), yet existing fusion approaches are constrained by a fundamental dilemma: early fusion struggles with modality heterogeneity, while late fusion severs crucial feature correlations. To address these limitations, we propose EvRWKV, a novel framework that enables continuous cross-modal interaction through dual-domain processing, which mainly includes a Cross-RWKV Module to capture fine-grained temporal and cross-modal dependencies, and an Event Image Spectral Fusion Enhancer (EISFE) module to perform joint adaptive frequency-domain denoising and spatial-domain alignment. This continuous interaction maintains feature consistency from low-level textures to high-level semantics. Extensive experiments on the real-world SDE and SDSD datasets demonstrate that EvRWKV significantly outperforms only image-based methods by 1.79 dB and 1.85 dB in PSNR, respectively. To further validate the practical utility of our method for downstream applications, we evaluated its impact on semantic segmentation. Experiments demonstrate that images enhanced by EvRWKV lead to a significant 35.44% improvement in mIoU.", "date": "2025-07-01", "categories": []}
{"id": "2511.09055", "title": "4KDehazeFlow: Ultra-High-Definition Image Dehazing via Flow Matching", "abstract": "Ultra-High-Definition (UHD) image dehazing faces challenges such as limited scene adaptability in prior-based methods and high computational complexity with color distortion in deep learning approaches. To address these issues, we propose 4KDehazeFlow, a novel method based on Flow Matching and the Haze-Aware vector field. This method models the dehazing process as a progressive optimization of continuous vector field flow, providing efficient data-driven adaptive nonlinear color transformation for high-quality dehazing. Specifically, our method has the following advantages: 1) 4KDehazeFlow is a general method compatible with various deep learning networks, without relying on any specific network architecture. 2) We propose a learnable 3D lookup table (LUT) that encodes haze transformation parameters into a compact 3D mapping matrix, enabling efficient inference through precomputed mappings. 3) We utilize a fourth-order Runge-Kutta (RK4) ordinary differential equation (ODE) solver to stably solve the dehazing flow field through an accurate step-by-step iterative method, effectively suppressing artifacts. Extensive experiments show that 4KDehazeFlow exceeds seven state-of-the-art methods. It delivers a 2dB PSNR increase and better performance in dense haze and color fidelity.", "date": "2025-11-12", "categories": []}
{"id": "2508.07760", "title": "Sea-Undistort: A Dataset for Through-Water Image Restoration in High Resolution Airborne Bathymetric Mapping", "abstract": "Accurate image-based bathymetric mapping in shallow waters remains challenging due to the complex optical distortions such as wave induced patterns, scattering and sunglint, introduced by the dynamic water surface, the water column properties, and solar illumination. In this work, we introduce Sea-Undistort, a comprehensive synthetic dataset of 1200 paired 512x512 through-water scenes rendered in Blender. Each pair comprises a distortion-free and a distorted view, featuring realistic water effects such as sun glint, waves, and scattering over diverse seabeds. Accompanied by per-image metadata such as camera parameters, sun position, and average depth, Sea-Undistort enables supervised training that is otherwise infeasible in real environments. We use Sea-Undistort to benchmark two state-of-the-art image restoration methods alongside an enhanced lightweight diffusion-based framework with an early-fusion sun-glint mask. When applied to real aerial data, the enhanced diffusion model delivers more complete Digital Surface Models (DSMs) of the seabed, especially in deeper areas, reduces bathymetric errors, suppresses glint and scattering, and crisply restores fine seabed details. Dataset, weights, and code are publicly available at https://www.magicbathy.eu/Sea-Undistort.html.", "date": "2025-08-11", "categories": []}
{"id": "2509.12728", "title": "Generalizable Holographic Reconstruction via Amplitude-Only Diffusion Priors", "abstract": "Phase retrieval in inline holography is a fundamental yet ill-posed inverse problem due to the nonlinear coupling between amplitude and phase in coherent imaging. We present a novel off-the-shelf solution that leverages a diffusion model trained solely on object amplitude to recover both amplitude and phase from diffraction intensities. Using a predictor-corrector sampling framework with separate likelihood gradients for amplitude and phase, our method enables complex field reconstruction without requiring ground-truth phase data for training. We validate the proposed approach through extensive simulations and experiments, demonstrating robust generalization across diverse object shapes, imaging system configurations, and modalities, including lensless setups. Notably, a diffusion prior trained on simple amplitude data (e.g., polystyrene beads) successfully reconstructs complex biological tissue structures, highlighting the method's adaptability. This framework provides a cost-effective, generalizable solution for nonlinear inverse problems in computational imaging, and establishes a foundation for broader coherent imaging applications beyond holography.", "date": "2025-09-16", "categories": []}
{"id": "2511.07479", "title": "Modulo Video Recovery via Selective Spatiotemporal Vision Transformer", "abstract": "Conventional image sensors have limited dynamic range, causing saturation in high-dynamic-range (HDR) scenes. Modulo cameras address this by folding incident irradiance into a bounded range, yet require specialized unwrapping algorithms to reconstruct the underlying signal. Unlike HDR recovery, which extends dynamic range from conventional sampling, modulo recovery restores actual values from folded samples. Despite being introduced over a decade ago, progress in modulo image recovery has been slow, especially in the use of modern deep learning techniques. In this work, we demonstrate that standard HDR methods are unsuitable for modulo recovery. Transformers, however, can capture global dependencies and spatial-temporal relationships crucial for resolving folded video frames. Still, adapting existing Transformer architectures for modulo recovery demands novel techniques. To this end, we present Selective Spatiotemporal Vision Transformer (SSViT), the first deep learning framework for modulo video reconstruction. SSViT employs a token selection strategy to improve efficiency and concentrate on the most critical regions. Experiments confirm that SSViT produces high-quality reconstructions from 8-bit folded videos and achieves state-of-the-art performance in modulo video recovery.", "date": "2025-11-09", "categories": []}
{"id": "2511.08427", "title": "An update to PYRO-NN: A Python Library for Differentiable CT Operators", "abstract": "Deep learning has brought significant advancements to X-ray Computed Tomography (CT) reconstruction, offering solutions to challenges arising from modern imaging technologies. These developments benefit from methods that combine classical reconstruction techniques with data-driven approaches. Differentiable operators play a key role in this integration by enabling end-to-end optimization and the incorporation of physical modeling within neural networks.\n  In this work, we present an updated version of PYRO-NN, a Python-based library for differentiable CT reconstruction. The updated framework extends compatibility to PyTorch and introduces native CUDA kernel support for efficient projection and back-projection operations across parallel, fan, and cone-beam geometries. Additionally, it includes tools for simulating imaging artifacts, modeling arbitrary acquisition trajectories, and creating flexible, end-to-end trainable pipelines through a high-level Python API. Code is available at: https://github.com/csyben/PYRO-NN", "date": "2025-11-11", "categories": []}
{"id": "2412.05853", "title": "Unsupervised Multi-Parameter Inverse Solving for Reducing Ring Artifacts in 3D X-Ray CBCT", "abstract": "Ring artifacts are prevalent in 3D cone-beam computed tomography (CBCT) due to non-ideal responses of X-ray detectors, substantially affecting image quality and diagnostic reliability. Existing state-of-the-art (SOTA) ring artifact reduction (RAR) methods rely on supervised learning with large-scale paired CT datasets. While effective in-domain, supervised methods tend to struggle to fully capture the physical characteristics of ring artifacts, leading to pronounced performance drops in complex real-world acquisitions. Moreover, their scalability to 3D CBCT is limited by high memory demands. In this work, we propose Riner, a new unsupervised RAR method. Based on a theoretical analysis of ring artifact formation, we reformulate RAR as a multi-parameter inverse problem, where the non-ideal responses of X-ray detectors are parameterized as solvable physical variables. Using a new differentiable forward model, Riner can jointly learn the implicit neural representation of artifact-free images and estimate the physical parameters directly from CT measurements, without external training data. Additionally, Riner is memory-friendly due to its ray-based optimization, enhancing its usability in large-scale 3D CBCT. Experiments on both simulated and real-world datasets show Riner outperforms existing SOTA supervised methods.", "date": "2024-12-08", "categories": []}
{"id": "2511.06751", "title": "Hierarchical Spatial-Frequency Aggregation for Spectral Deconvolution Imaging", "abstract": "Computational spectral imaging (CSI) achieves real-time hyperspectral imaging through co-designed optics and algorithms, but typical CSI methods suffer from a bulky footprint and limited fidelity. Therefore, Spectral Deconvolution imaging (SDI) methods based on PSF engineering have been proposed to achieve high-fidelity compact CSI design recently. However, the composite convolution-integration operations of SDI render the normal-equation coefficient matrix scene-dependent, which hampers the efficient exploitation of imaging priors and poses challenges for accurate reconstruction. To tackle the inherent data-dependent operators in SDI, we introduce a Hierarchical Spatial-Spectral Aggregation Unfolding Framework (HSFAUF). By decomposing subproblems and projecting them into the frequency domain, HSFAUF transforms nonlinear processes into linear mappings, thereby enabling efficient solutions. Furthermore, to integrate spatial-spectral priors during iterative refinement, we propose a Spatial-Frequency Aggregation Transformer (SFAT), which explicitly aggregates information across spatial and frequency domains. By integrating SFAT into HSFAUF, we develop a Transformer-based deep unfolding method, \\textbf{H}ierarchical \\textbf{S}patial-\\textbf{F}requency \\textbf{A}ggregation \\textbf{U}nfolding \\textbf{T}ransformer (HSFAUT), to solve the inverse problem of SDI. Systematic simulated and real experiments show that HSFAUT surpasses SOTA methods with cheaper memory and computational costs, while exhibiting optimal performance on different SDI systems.", "date": "2025-11-10", "categories": []}
{"id": "2511.06823", "title": "Integrating Reweighted Least Squares with Plug-and-Play Diffusion Priors for Noisy Image Restoration", "abstract": "Existing plug-and-play image restoration methods typically employ off-the-shelf Gaussian denoisers as proximal operators within classical optimization frameworks based on variable splitting. Recently, denoisers induced by generative priors have been successfully integrated into regularized optimization methods for image restoration under Gaussian noise. However, their application to non-Gaussian noise--such as impulse noise--remains largely unexplored. In this paper, we propose a plug-and-play image restoration framework based on generative diffusion priors for robust removal of general noise types, including impulse noise. Within the maximum a posteriori (MAP) estimation framework, the data fidelity term is adapted to the specific noise model. Departing from the conventional least-squares loss used for Gaussian noise, we introduce a generalized Gaussian scale mixture-based loss, which approximates a wide range of noise distributions and leads to an $\\ell_q$-norm ($0<q\\leq2$) fidelity term. This optimization problem is addressed using an iteratively reweighted least squares (IRLS) approach, wherein the proximal step involving the generative prior is efficiently performed via a diffusion-based denoiser. Experimental results on benchmark datasets demonstrate that the proposed method effectively removes non-Gaussian impulse noise and achieves superior restoration performance.", "date": "2025-11-10", "categories": []}
{"id": "2306.15599", "title": "Coupling a Recurrent Neural Network to SPAD TCSPC Systems for Real-time Fluorescence Lifetime Imaging", "abstract": "Fluorescence lifetime imaging (FLI) has been receiving increased attention in recent years as a powerful diagnostic technique in biological and medical research. However, existing FLI systems often suffer from a tradeoff between processing speed, accuracy, and robustness. In this paper, we propose a robust approach that enables fast FLI with no degradation of accuracy. The approach is based on a SPAD TCSPC system coupled to a recurrent neural network (RNN) that accurately estimates the fluorescence lifetime directly from raw timestamps without building histograms, thereby drastically reducing transfer data volumes and hardware resource utilization, thus enabling FLI acquisition at video rate. We train two variants of the RNN on a synthetic dataset and compare the results to those obtained using center-of-mass method (CMM) and least squares fitting (LS fitting). Results demonstrate that two RNN variants, gated recurrent unit (GRU) and long short-term memory (LSTM), are comparable to CMM and LS fitting in terms of accuracy, while outperforming them in background noise by a large margin. To explore the ultimate limits of the approach, we derived the Cramer-Rao lower bound of the measurement, showing that RNN yields lifetime estimations with near-optimal precision. Moreover, our FLI model, which is purely trained on synthetic datasets, works well with never-seen-before, real-world data. To demonstrate real-time operation, we have built a FLI microscope based on Piccolo, a 32x32 SPAD sensor developed in our lab. Four quantized GRU cores, capable of processing up to 4 million photons per second, are deployed on a Xilinx Kintex-7 FPGA. Powered by the GRU, the FLI setup can retrieve real-time fluorescence lifetime images at up to 10 frames per second. The proposed FLI system is promising and ideally suited for biomedical applications.", "date": "2023-06-27", "categories": []}
{"id": "2311.17643", "title": "Thera: Aliasing-Free Arbitrary-Scale Super-Resolution with Neural Heat Fields", "abstract": "Recent approaches to arbitrary-scale single image super-resolution (ASR) use neural fields to represent continuous signals that can be sampled at arbitrary resolutions. However, point-wise queries of neural fields do not naturally match the point spread function (PSF) of pixels, which may cause aliasing in the super-resolved image. Existing methods attempt to mitigate this by approximating an integral version of the field at each scaling factor, compromising both fidelity and generalization. In this work, we introduce neural heat fields, a novel neural field formulation that inherently models a physically exact PSF. Our formulation enables analytically correct anti-aliasing at any desired output resolution, and -- unlike supersampling -- at no additional cost. Building on this foundation, we propose Thera, an end-to-end ASR method that substantially outperforms existing approaches, while being more parameter-efficient and offering strong theoretical guarantees. The project page is at https://therasr.github.io.", "date": "2023-11-29", "categories": []}
{"id": "2502.04521", "title": "Generative Autoregressive Transformers for Model-Agnostic Federated MRI Reconstruction", "abstract": "While learning-based models hold great promise for MRI reconstruction, single-site models trained on limited local datasets often show poor generalization. This has motivated collaborative training across institutions via federated learning (FL)-a privacy-preserving framework that aggregates model updates instead of sharing raw data. Conventional FL requires architectural homogeneity, restricting sites from using models tailored to their resources or needs. To address this limitation, we propose FedGAT, a model-agnostic FL technique that first collaboratively trains a global generative prior for MR images, adapted from a natural image foundation model composed of a variational autoencoder (VAE) and a transformer that generates images via spatial-scale autoregression. We fine-tune the transformer module after injecting it with a lightweight site-specific prompting mechanism, keeping the VAE frozen, to efficiently adapt the model to multi-site MRI data. In a second tier, each site independently trains its preferred reconstruction model by augmenting local data with synthetic MRI data from other sites, generated by site-prompting the tuned prior. This decentralized augmentation improves generalization while preserving privacy. Experiments on multi-institutional datasets show that FedGAT outperforms state-of-the-art FL baselines in both within- and cross-site reconstruction performance under model-heterogeneous settings.", "date": "2025-02-06", "categories": []}
{"id": "2405.20559", "title": "Information-driven design of imaging systems", "abstract": "Imaging systems have traditionally been designed to mimic the human eye and produce visually interpretable measurements. Modern imaging systems, however, process raw measurements computationally before or instead of human viewing. As a result, the information content of raw measurements matters more than their visual interpretability. Despite the importance of measurement information content, current approaches for evaluating imaging system performance do not quantify it: they instead either use alternative metrics that assess specific aspects of measurement quality or assess measurements indirectly with performance on secondary tasks.\n  We developed the theoretical foundations and a practical method to directly quantify mutual information between noisy measurements and unknown objects. By fitting probabilistic models to measurements and their noise characteristics, our method estimates information by upper bounding its true value. By applying gradient-based optimization to these estimates, we also developed a technique for designing imaging systems called Information-Driven Encoder Analysis Learning (IDEAL). Our information estimates accurately captured system performance differences across four imaging domains (color photography, radio astronomy, lensless imaging, and microscopy). Systems designed with IDEAL matched the performance of those designed with end-to-end optimization, the prevailing approach that jointly optimizes hardware and image processing algorithms. These results establish mutual information as a universal performance metric for imaging systems that enables both computationally efficient design optimization and evaluation in real-world conditions.\n  A video summarizing this work can be found at: https://waller-lab.github.io/EncodingInformationWebsite/", "date": "2024-05-31", "categories": []}
{"id": "2511.02938", "title": "From Narrow to Wide: Autoencoding Transformers for Ultrasound Bandwidth Recovery", "abstract": "Conventional pulse-echo ultrasound suffers when low-cost probes deliver only narrow fractional bandwidths, elongating pulses and erasing high-frequency detail. We address this limitation by learning a data-driven mapping from band-limited to broadband spectrogram of radio-frequency (RF) lines. To this end, a variation of Tiny Vision Transform (ViT) auto-encoder is trained on simulation data using a curriculum-weighted loss. On heterogeneous speckle-cyst phantoms, the network reduces image-domain MSE by 90 percent, boosts PSNR by 6.7 dB, and raises SSIM to 0.965 compared with the narrow-band input. It also sharpens point-target rows in a completely unseen resolution phantom, demonstrating strong out-of-distribution generalisation without sacrificing frame rate or phase information. These results indicate that a purely software upgrade can endow installed narrow-band probes with broadband-like performance, potentially widening access to high-resolution ultrasound in resource-constrained settings.", "date": "2025-11-04", "categories": []}
{"id": "2511.03093", "title": "A Plug-and-Play Framework for Volumetric Light-Sheet Image Reconstruction", "abstract": "Cardiac contraction is a rapid, coordinated process that unfolds across three-dimensional tissue on millisecond timescales. Traditional optical imaging is often inadequate for capturing dynamic cellular structure in the beating heart because of a fundamental trade-off between spatial and temporal resolution. To overcome these limitations, we propose a high-performance computational imaging framework that integrates Compressive Sensing (CS) with Light-Sheet Microscopy (LSM) for efficient, low-phototoxic cardiac imaging. The system performs compressed acquisition of fluorescence signals via random binary mask coding using a Digital Micromirror Device (DMD). We propose a Plug-and-Play (PnP) framework, solved using the alternating direction method of multipliers (ADMM), which flexibly incorporates advanced denoisers, including Tikhonov, Total Variation (TV), and BM3D. To preserve structural continuity in dynamic imaging, we further introduce temporal regularization enforcing smoothness between adjacent z-slices. Experimental results on zebrafish heart imaging under high compression ratios demonstrate that the proposed method successfully reconstructs cellular structures with excellent denoising performance and image clarity, validating the effectiveness and robustness of our algorithm in real-world high-speed, low-light biological imaging scenarios.", "date": "2025-11-05", "categories": []}
{"id": "2504.00159", "title": "SonarSplat: Novel View Synthesis of Imaging Sonar via Gaussian Splatting", "abstract": "In this paper, we present SonarSplat, a novel Gaussian splatting framework for imaging sonar that demonstrates realistic novel view synthesis and models acoustic streaking phenomena. Our method represents the scene as a set of 3D Gaussians with acoustic reflectance and saturation properties. We develop a novel method to efficiently rasterize Gaussians to produce a range/azimuth image that is faithful to the acoustic image formation model of imaging sonar. In particular, we develop a novel approach to model azimuth streaking in a Gaussian splatting framework. We evaluate SonarSplat using real-world datasets of sonar images collected from an underwater robotic platform in a controlled test tank and in a real-world river environment. Compared to the state-of-the-art, SonarSplat offers improved image synthesis capabilities (+3.2 dB PSNR) and more accurate 3D reconstruction (77% lower Chamfer Distance). We also demonstrate that SonarSplat can be leveraged for azimuth streak removal.", "date": "2025-03-31", "categories": []}
{"id": "2510.26907", "title": "Generative diffusion modeling protocols for improving the Kikuchi pattern indexing in electron back-scatter diffraction", "abstract": "Electron back-scatter diffraction (EBSD) has traditionally relied upon methods such as the Hough transform and dictionary Indexing to interpret diffraction patterns and extract crystallographic orientation. However, these methods encounter significant limitations, particularly when operating at high scanning speeds, where the exposure time per pattern is decreased beyond the operating sensitivity of CCD camera. Hence the signal to noise ratio decreases for the observed pattern which makes the pattern noisy, leading to reduced indexing accuracy. This research work aims to develop generative machine learning models for the post-processing or on-the-fly processing of Kikuchi patterns which are capable of restoring noisy EBSD patterns obtained at high scan speeds. These restored patterns can be used for the determination of crystal orientations to provide reliable indexing results. We compare the performance of such generative models in enhancing the quality of patterns captured at short exposure times (high scan speeds). An interesting observation is that the methodology is not data-hungry as typical machine learning methods.", "date": "2025-10-30", "categories": []}
{"id": "2511.01804", "title": "Dynamic Reconstruction of Ultrasound-Derived Flow Fields With Physics-Informed Neural Fields", "abstract": "Blood flow is sensitive to disease and provides insight into cardiac function, making flow field analysis valuable for diagnosis. However, while safer than radiation-based imaging and more suitable for patients with medical implants, ultrasound suffers from attenuation with depth, limiting the quality of the image. Despite advances in echocardiographic particle image velocimetry (EchoPIV), accurately measuring blood velocity remains challenging due to the technique's limitations and the complexity of blood flow dynamics. Physics-informed machine learning can enhance accuracy and robustness, particularly in scenarios where noisy or incomplete data challenge purely data-driven approaches. We present a physics-informed neural field model with multi-scale Fourier Feature encoding for estimating blood flow from sparse and noisy ultrasound data without requiring ground truth supervision. We demonstrate that this model achieves consistently low mean squared error in denoising and inpainting both synthetic and real datasets, verified against reference flow fields and ground truth flow rate measurements. While physics-informed neural fields have been widely used to reconstruct medical images, applications to medical flow reconstruction are mostly prominent in Flow MRI. In this work, we adapt methods that have proven effective in other imaging modalities to address the specific challenge of ultrasound-based flow reconstruction.", "date": "2025-11-03", "categories": []}
{"id": "2510.27667", "title": "Deep learning denoising unlocks quantitative insights in operando materials microscopy", "abstract": "Operando microscopy provides direct insight into the dynamic chemical and physical processes that govern functional materials, yet measurement noise limits the effective resolution and undermines quantitative analysis. Here, we present a general framework for integrating unsupervised deep learning-based denoising into quantitative microscopy workflows across modalities and length scales. Using simulated data, we demonstrate that deep denoising preserves physical fidelity, introduces minimal bias, and reduces uncertainty in model learning with partial differential equation (PDE)-constrained optimization. Applied to experiments, denoising reveals nanoscale chemical and structural heterogeneity in scanning transmission X-ray microscopy (STXM) of lithium iron phosphate (LFP), enables automated particle segmentation and phase classification in optical microscopy of graphite electrodes, and reduces noise-induced variability by nearly 80% in neutron radiography to resolve heterogeneous lithium transport. Collectively, these results establish deep denoising as a powerful, modality-agnostic enhancement that advances quantitative operando imaging and extends the reach of previously noise-limited techniques.", "date": "2025-10-31", "categories": []}
{"id": "2510.26173", "title": "MoTDiff: High-resolution Motion Trajectory estimation from a single blurred image using Diffusion models", "abstract": "Accurate estimation of motion information is crucial in diverse computational imaging and computer vision applications. Researchers have investigated various methods to extract motion information from a single blurred image, including blur kernels and optical flow. However, existing motion representations are often of low quality, i.e., coarse-grained and inaccurate. In this paper, we propose the first high-resolution (HR) Motion Trajectory estimation framework using Diffusion models (MoTDiff). Different from existing motion representations, we aim to estimate an HR motion trajectory with high-quality from a single motion-blurred image. The proposed MoTDiff consists of two key components: 1) a new conditional diffusion framework that uses multi-scale feature maps extracted from a single blurred image as a condition, and 2) a new training method that can promote precise identification of a fine-grained motion trajectory, consistent estimation of overall shape and position of a motion path, and pixel connectivity along a motion trajectory. Our experiments demonstrate that the proposed MoTDiff can outperform state-of-the-art methods in both blind image deblurring and coded exposure photography applications.", "date": "2025-10-30", "categories": []}
{"id": "2510.26339", "title": "GLYPH-SR: Can We Achieve Both High-Quality Image Super-Resolution and High-Fidelity Text Recovery via VLM-guided Latent Diffusion Model?", "abstract": "Image super-resolution(SR) is fundamental to many vision system-from surveillance and autonomy to document analysis and retail analytics-because recovering high-frequency details, especially scene-text, enables reliable downstream perception. Scene-text, i.e., text embedded in natural images such as signs, product labels, and storefronts, often carries the most actionable information; when characters are blurred or hallucinated, optical character recognition(OCR) and subsequent decisions fail even if the rest of the image appears sharp. Yet previous SR research has often been tuned to distortion (PSNR/SSIM) or learned perceptual metrics (LIPIS, MANIQA, CLIP-IQA, MUSIQ) that are largely insensitive to character-level errors. Furthermore, studies that do address text SR often focus on simplified benchmarks with isolated characters, overlooking the challenges of text within complex natural scenes. As a result, scene-text is effectively treated as generic texture. For SR to be effective in practical deployments, it is therefore essential to explicitly optimize for both text legibility and perceptual quality. We present GLYPH-SR, a vision-language-guided diffusion framework that aims to achieve both objectives jointly. GLYPH-SR utilizes a Text-SR Fusion ControlNet(TS-ControlNet) guided by OCR data, and a ping-pong scheduler that alternates between text- and scene-centric guidance. To enable targeted text restoration, we train these components on a synthetic corpus while keeping the main SR branch frozen. Across SVT, SCUT-CTW1500, and CUTE80 at x4, and x8, GLYPH-SR improves OCR F1 by up to +15.18 percentage points over diffusion/GAN baseline (SVT x8, OpenOCR) while maintaining competitive MANIQA, CLIP-IQA, and MUSIQ. GLYPH-SR is designed to satisfy both objectives simultaneously-high readability and high visual realism-delivering SR that looks right and reds right.", "date": "2025-10-30", "categories": []}
{"id": "2510.25221", "title": "MSF-Net: Multi-Stage Feature Extraction and Fusion for Robust Photometric Stereo", "abstract": "Photometric stereo is a technique aimed at determining surface normals through the utilization of shading cues derived from images taken under different lighting conditions. However, existing learning-based approaches often fail to accurately capture features at multiple stages and do not adequately promote interaction between these features. Consequently, these models tend to extract redundant features, especially in areas with intricate details such as wrinkles and edges. To tackle these issues, we propose MSF-Net, a novel framework for extracting information at multiple stages, paired with selective update strategy, aiming to extract high-quality feature information, which is critical for accurate normal construction. Additionally, we have developed a feature fusion module to improve the interplay among different features. Experimental results on the DiLiGenT benchmark show that our proposed MSF-Net significantly surpasses previous state-of-the-art methods in the accuracy of surface normal estimation.", "date": "2025-10-29", "categories": []}
{"id": "2510.25314", "title": "Seeing Clearly and Deeply: An RGBD Imaging Approach with a Bio-inspired Monocentric Design", "abstract": "Achieving high-fidelity, compact RGBD imaging presents a dual challenge: conventional compact optics struggle with RGB sharpness across the entire depth-of-field, while software-only Monocular Depth Estimation (MDE) is an ill-posed problem reliant on unreliable semantic priors. While deep optics with elements like DOEs can encode depth, they introduce trade-offs in fabrication complexity and chromatic aberrations, compromising simplicity. To address this, we first introduce a novel bio-inspired all-spherical monocentric lens, around which we build the Bionic Monocentric Imaging (BMI) framework, a holistic co-design. This optical design naturally encodes depth into its depth-varying Point Spread Functions (PSFs) without requiring complex diffractive or freeform elements. We establish a rigorous physically-based forward model to generate a synthetic dataset by precisely simulating the optical degradation process. This simulation pipeline is co-designed with a dual-head, multi-scale reconstruction network that employs a shared encoder to jointly recover a high-fidelity All-in-Focus (AiF) image and a precise depth map from a single coded capture. Extensive experiments validate the state-of-the-art performance of the proposed framework. In depth estimation, the method attains an Abs Rel of 0.026 and an RMSE of 0.130, markedly outperforming leading software-only approaches and other deep optics systems. For image restoration, the system achieves an SSIM of 0.960 and a perceptual LPIPS score of 0.082, thereby confirming a superior balance between image fidelity and depth accuracy. This study illustrates that the integration of bio-inspired, fully spherical optics with a joint reconstruction algorithm constitutes an effective strategy for addressing the intrinsic challenges in high-performance compact RGBD imaging. Source code will be publicly available at https://github.com/ZongxiYu-ZJU/BMI.", "date": "2025-10-29", "categories": []}
{"id": "2510.24486", "title": "Fast and accurate neural reflectance transformation imaging through knowledge distillation", "abstract": "Reflectance Transformation Imaging (RTI) is very popular for its ability to visually analyze surfaces by enhancing surface details through interactive relighting, starting from only a few tens of photographs taken with a fixed camera and variable illumination. Traditional methods like Polynomial Texture Maps (PTM) and Hemispherical Harmonics (HSH) are compact and fast, but struggle to accurately capture complex reflectance fields using few per-pixel coefficients and fixed bases, leading to artifacts, especially in highly reflective or shadowed areas. The NeuralRTI approach, which exploits a neural autoencoder to learn a compact function that better approximates the local reflectance as a function of light directions, has been shown to produce superior quality at comparable storage cost. However, as it performs interactive relighting with custom decoder networks with many parameters, the rendering step is computationally expensive and not feasible at full resolution for large images on limited hardware. Earlier attempts to reduce costs by directly training smaller networks have failed to produce valid results. For this reason, we propose to reduce its computational cost through a novel solution based on Knowledge Distillation (DisK-NeuralRTI). ...", "date": "2025-10-28", "categories": []}
{"id": "2510.24579", "title": "Physics-Inspired Gaussian Kolmogorov-Arnold Networks for X-ray Scatter Correction in Cone-Beam CT", "abstract": "Cone-beam CT (CBCT) employs a flat-panel detector to achieve three-dimensional imaging with high spatial resolution. However, CBCT is susceptible to scatter during data acquisition, which introduces CT value bias and reduced tissue contrast in the reconstructed images, ultimately degrading diagnostic accuracy. To address this issue, we propose a deep learning-based scatter artifact correction method inspired by physical prior knowledge. Leveraging the fact that the observed point scatter probability density distribution exhibits rotational symmetry in the projection domain. The method uses Gaussian Radial Basis Functions (RBF) to model the point scatter function and embeds it into the Kolmogorov-Arnold Networks (KAN) layer, which provides efficient nonlinear mapping capabilities for learning high-dimensional scatter features. By incorporating the physical characteristics of the scattered photon distribution together with the complex function mapping capacity of KAN, the model improves its ability to accurately represent scatter. The effectiveness of the method is validated through both synthetic and real-scan experiments. Experimental results show that the model can effectively correct the scatter artifacts in the reconstructed images and is superior to the current methods in terms of quantitative metrics.", "date": "2025-10-28", "categories": []}
{"id": "2507.00372", "title": "Efficient Depth- and Spatially-Varying Image Simulation for Defocus Deblur", "abstract": "Modern cameras with large apertures often suffer from a shallow depth of field, resulting in blurry images of objects outside the focal plane. This limitation is particularly problematic for fixed-focus cameras, such as those used in smart glasses, where adding autofocus mechanisms is challenging due to form factor and power constraints. Due to unmatched optical aberrations and defocus properties unique to each camera system, deep learning models trained on existing open-source datasets often face domain gaps and do not perform well in real-world settings. In this paper, we propose an efficient and scalable dataset synthesis approach that does not rely on fine-tuning with real-world data. Our method simultaneously models depth-dependent defocus and spatially varying optical aberrations, addressing both computational complexity and the scarcity of high-quality RGB-D datasets. Experimental results demonstrate that a network trained on our low resolution synthetic images generalizes effectively to high resolution (12MP) real-world images across diverse scenes.", "date": "2025-07-01", "categories": []}
{"id": "2510.21787", "title": "Mismatch reconstruction theory for unknown measurement matrix in imaging through multimode fiber bending", "abstract": "Multimode fiber imaging requires strict matching between measurement value and measurement matrix to achieve image reconstruction. However, in practical applications, the measurement matrix often cannot be obtained due to unknown system configuration or difficulty in real-time alignment after arbitrary fiber bending, resulting in the failure of traditional reconstruction algorithms. This paper presents a novel mismatch reconstruction theory for solving the problem of image reconstruction when measurement matrix is unknown. We first propose mismatch equation and design matched and calibration solution algorithms to construct a new measurement matrix. In addition, we also provide a detailed proof of these equations and algorithms in the appendix. The experimental results show that under low noise levels, constructed matrix can be used for matched pair in traditional reconstruction algorithms, and reconstruct the original image successfully. Then, we analyze the impact of noise, computational precision and orthogonality on reconstruction performance. The results show that proposed algorithms have a certain degree of robustness. Finally, we discuss the limitations and potential applications of this theory. The code is available: https://github.com/yanglebupt/mismatch-solution.", "date": "2025-10-19", "categories": []}
{"id": "2510.21815", "title": "HDR Image Reconstruction using an Unsupervised Fusion Model", "abstract": "High Dynamic Range (HDR) imaging aims to reproduce the wide range of brightness levels present in natural scenes, which the human visual system can perceive but conventional digital cameras often fail to capture due to their limited dynamic range. To address this limitation, we propose a deep learning-based multi-exposure fusion approach for HDR image generation. The method takes a set of differently exposed Low Dynamic Range (LDR) images, typically an underexposed and an overexposed image, and learns to fuse their complementary information using a convolutional neural network (CNN). The underexposed image preserves details in bright regions, while the overexposed image retains information in dark regions; the network effectively combines these to reconstruct a high-quality HDR output. The model is trained in an unsupervised manner, without relying on ground-truth HDR images, making it practical for real-world applications where such data is unavailable. We evaluate our results using the Multi-Exposure Fusion Structural Similarity Index Measure (MEF-SSIM) and demonstrate that our approach achieves superior visual quality compared to existing fusion methods. A customized loss function is further introduced to improve reconstruction fidelity and optimize model performance.", "date": "2025-10-21", "categories": []}
{"id": "2510.22154", "title": "Frequency-Spatial Interaction Driven Network for Low-Light Image Enhancement", "abstract": "Low-light image enhancement (LLIE) aims at improving the perception or interpretability of an image captured in an environment with poor illumination. With the advent of deep learning, the LLIE technique has achieved significant breakthroughs. However, existing LLIE methods either ignore the important role of frequency domain information or fail to effectively promote the propagation and flow of information, limiting the LLIE performance. In this paper, we develop a novel frequency-spatial interaction-driven network (FSIDNet) for LLIE based on two-stage architecture. To be specific, the first stage is designed to restore the amplitude of low-light images to improve the lightness, and the second stage devotes to restore phase information to refine fine-grained structures. Considering that Frequency domain and spatial domain information are complementary and both favorable for LLIE, we further develop two frequency-spatial interaction blocks which mutually amalgamate the complementary spatial and frequency information to enhance the capability of the model. In addition, we construct the Information Exchange Module (IEM) to associate two stages by adequately incorporating cross-stage and cross-scale features to effectively promote the propagation and flow of information in the two-stage network structure. Finally, we conduct experiments on several widely used benchmark datasets (i.e., LOL-Real, LSRW-Huawei, etc.), which demonstrate that our method achieves the excellent performance in terms of visual results and quantitative metrics while preserving good model efficiency.", "date": "2025-10-25", "categories": []}
{"id": "2510.22534", "title": "SRSR: Enhancing Semantic Accuracy in Real-World Image Super-Resolution with Spatially Re-Focused Text-Conditioning", "abstract": "Existing diffusion-based super-resolution approaches often exhibit semantic ambiguities due to inaccuracies and incompleteness in their text conditioning, coupled with the inherent tendency for cross-attention to divert towards irrelevant pixels. These limitations can lead to semantic misalignment and hallucinated details in the generated high-resolution outputs. To address these, we propose a novel, plug-and-play spatially re-focused super-resolution (SRSR) framework that consists of two core components: first, we introduce Spatially Re-focused Cross-Attention (SRCA), which refines text conditioning at inference time by applying visually-grounded segmentation masks to guide cross-attention. Second, we introduce a Spatially Targeted Classifier-Free Guidance (STCFG) mechanism that selectively bypasses text influences on ungrounded pixels to prevent hallucinations. Extensive experiments on both synthetic and real-world datasets demonstrate that SRSR consistently outperforms seven state-of-the-art baselines in standard fidelity metrics (PSNR and SSIM) across all datasets, and in perceptual quality measures (LPIPS and DISTS) on two real-world benchmarks, underscoring its effectiveness in achieving both high semantic fidelity and perceptual quality in super-resolution.", "date": "2025-10-26", "categories": []}
{"id": "2510.22577", "title": "From Pixels to Views: Learning Angular-Aware and Physics-Consistent Representations for Light Field Microscopy", "abstract": "Light field microscopy (LFM) has become an emerging tool in neuroscience for large-scale neural imaging in vivo, notable for its single-exposure volumetric imaging, broad field of view, and high temporal resolution. However, learning-based 3D reconstruction in XLFM remains underdeveloped due to two core challenges: the absence of standardized datasets and the lack of methods that can efficiently model its angular-spatial structure while remaining physically grounded. We address these challenges by introducing three key contributions. First, we construct the XLFM-Zebrafish benchmark, a large-scale dataset and evaluation suite for XLFM reconstruction. Second, we propose Masked View Modeling for Light Fields (MVN-LF), a self-supervised task that learns angular priors by predicting occluded views, improving data efficiency. Third, we formulate the Optical Rendering Consistency Loss (ORC Loss), a differentiable rendering constraint that enforces alignment between predicted volumes and their PSF-based forward projections. On the XLFM-Zebrafish benchmark, our method improves PSNR by 7.7% over state-of-the-art baselines.", "date": "2025-10-26", "categories": []}
{"id": "2510.22605", "title": "Projection Embedded Diffusion Bridge for CT Reconstruction from Incomplete Data", "abstract": "Reconstructing CT images from incomplete projection data remains challenging due to the ill-posed nature of the problem. Diffusion bridge models have recently shown promise in restoring clean images from their corresponding Filtered Back Projection (FBP) reconstructions, but incorporating data consistency into these models remains largely underexplored. Incorporating data consistency can improve reconstruction fidelity by aligning the reconstructed image with the observed projection data, and can enhance detail recovery by integrating structural information contained in the projections. In this work, we propose the Projection Embedded Diffusion Bridge (PEDB). PEDB introduces a novel reverse stochastic differential equation (SDE) to sample from the distribution of clean images conditioned on both the FBP reconstruction and the incomplete projection data. By explicitly conditioning on the projection data in sampling the clean images, PEDB naturally incorporates data consistency. We embed the projection data into the score function of the reverse SDE. Under certain assumptions, we derive a tractable expression for the posterior score. In addition, we introduce a free parameter to control the level of stochasticity in the reverse process. We also design a discretization scheme for the reverse SDE to mitigate discretization error. Extensive experiments demonstrate that PEDB achieves strong performance in CT reconstruction from three types of incomplete data, including sparse-view, limited-angle, and truncated projections. For each of these types, PEDB outperforms evaluated state-of-the-art diffusion bridge models across standard, noisy, and domain-shift evaluations.", "date": "2025-10-26", "categories": []}
{"id": "2411.14269", "title": "Guided MRI Reconstruction via Schr\\\"odinger Bridge", "abstract": "Magnetic Resonance Imaging (MRI) is an inherently multi-contrast modality, where cross-contrast priors can be exploited to improve image reconstruction from undersampled data. Recently, diffusion models have shown remarkable performance in MRI reconstruction. However, they still struggle to effectively utilize such priors, mainly because existing methods rely on feature-level fusion in image or latent spaces, which lacks explicit structural correspondence and thus leads to suboptimal performance. To address this issue, we propose $\\mathbf{I}^2$SB-Inversion, a multi-contrast guided reconstruction framework based on the Schr\\\"odinger Bridge (SB). The proposed method performs pixel-wise translation between paired contrasts, providing explicit structural constraints between the guidance and target images. Furthermore, an Inversion strategy is introduced to correct inter-modality misalignment, which often occurs in guided reconstruction, thereby mitigating artifacts and improving reconstruction accuracy. Experiments on paired T1- and T2-weighted datasets demonstrate that $\\mathbf{I}^2$SB-Inversion achieves a high acceleration factor of up to 14.4 and consistently outperforms existing methods in both quantitative and qualitative evaluations.", "date": "2024-11-21", "categories": []}
{"id": "2510.07961", "title": "Latent Harmony: Synergistic Unified UHD Image Restoration via Latent Space Regularization and Controllable Refinement", "abstract": "Ultra-High Definition (UHD) image restoration faces a trade-off between computational efficiency and high-frequency detail retention. While Variational Autoencoders (VAEs) improve efficiency via latent-space processing, their Gaussian constraint often discards degradation-specific high-frequency information, hurting reconstruction fidelity. To overcome this, we propose Latent Harmony, a two-stage framework that redefines VAEs for UHD restoration by jointly regularizing the latent space and enforcing high-frequency-aware reconstruction.In Stage One, we introduce LH-VAE, which enhances semantic robustness through visual semantic constraints and progressive degradation perturbations, while latent equivariance strengthens high-frequency reconstruction.Stage Two jointly trains this refined VAE with a restoration model using High-Frequency Low-Rank Adaptation (HF-LoRA): an encoder LoRA guided by a fidelity-oriented high-frequency alignment loss to recover authentic details, and a decoder LoRA driven by a perception-oriented loss to synthesize realistic textures. Both LoRA modules are trained via alternating optimization with selective gradient propagation to preserve the pretrained latent structure.At inference, a tunable parameter {\\alpha} enables flexible fidelity-perception trade-offs.Experiments show Latent Harmony achieves state-of-the-art performance across UHD and standard-resolution tasks, effectively balancing efficiency, perceptual quality, and reconstruction accuracy.", "date": "2025-10-09", "categories": []}
{"id": "2510.20972", "title": "Thermal Polarimetric Multi-view Stereo", "abstract": "This paper introduces a novel method for detailed 3D shape reconstruction utilizing thermal polarization cues. Unlike state-of-the-art methods, the proposed approach is independent of illumination and material properties. In this paper, we formulate a general theory of polarization observation and show that long-wave infrared (LWIR) polarimetric imaging is free from the ambiguities that affect visible polarization analyses. Subsequently, we propose a method for recovering detailed 3D shapes using multi-view thermal polarimetric images. Experimental results demonstrate that our approach effectively reconstructs fine details in transparent, translucent, and heterogeneous objects, outperforming existing techniques.", "date": "2025-10-23", "categories": []}
{"id": "2510.21100", "title": "HistRetinex: Optimizing Retinex model in Histogram Domain for Efficient Low-Light Image Enhancement", "abstract": "Retinex-based low-light image enhancement methods are widely used due to their excellent performance. However, most of them are time-consuming for large-sized images. This paper extends the Retinex model from the spatial domain to the histogram domain, and proposes a novel histogram-based Retinex model for fast low-light image enhancement, named HistRetinex. Firstly, we define the histogram location matrix and the histogram count matrix, which establish the relationship among histograms of the illumination, reflectance and the low-light image. Secondly, based on the prior information and the histogram-based Retinex model, we construct a novel two-level optimization model. Through solving the optimization model, we give the iterative formulas of the illumination histogram and the reflectance histogram, respectively. Finally, we enhance the low-light image through matching its histogram with the one provided by HistRetinex. Experimental results demonstrate that the HistRetinex outperforms existing enhancement methods in both visibility and performance metrics, while executing 1.86 seconds on 1000*664 resolution images, achieving a minimum time saving of 6.67 seconds.", "date": "2025-10-24", "categories": []}
{"id": "2510.21437", "title": "Anisotropic Pooling for LUT-realizable CNN Image Restoration", "abstract": "Table look-up realization of image restoration CNNs has the potential of achieving competitive image quality while being much faster and resource frugal than the straightforward CNN implementation. The main technical challenge facing the LUT-based CNN algorithm designers is to manage the table size without overly restricting the receptive field. The prevailing strategy is to reuse the table for small pixel patches of different orientations (apparently assuming a degree of isotropy) and then fuse the look-up results. The fusion is currently done by average pooling, which we find being ill suited to anisotropic signal structures. To alleviate the problem, we investigate and discuss anisotropic pooling methods to replace naive averaging for improving the performance of the current LUT-realizable CNN restoration methods. First, we introduce the method of generalized median pooling which leads to measurable gains over average pooling. We then extend this idea by learning data-dependent pooling coefficients for each orientation, so that they can adaptively weigh the contributions of differently oriented pixel patches. Experimental results on various restoration benchmarks show that our anisotropic pooling strategy yields both perceptually and numerically superior results compared to existing LUT-realizable CNN methods.", "date": "2025-10-24", "categories": []}
{"id": "2505.15961", "title": "Super-Resolution with Structured Motion", "abstract": "We consider the limits of super-resolution using imaging constraints. Due to various theoretical and practical limitations, reconstruction-based methods have been largely restricted to small increases in resolution. In addition, motion-blur is usually seen as a nuisance that impedes super-resolution. We show that by using high-precision motion information, sparse image priors, and convex optimization, it is possible to increase resolution by large factors. A key operation in super-resolution is deconvolution with a box. In general, convolution with a box is not invertible. However, we obtain perfect reconstructions of sparse signals using convex optimization. We also show that motion blur can be helpful for super-resolution. We demonstrate that using pseudo-random motion it is possible to reconstruct a high-resolution target using a single low-resolution image. We present numerical experiments with simulated data and results with real data captured by a camera mounted on a computer controlled stage.", "date": "2025-05-21", "categories": []}
{"id": "2505.17581", "title": "MODEM: A Morton-Order Degradation Estimation Mechanism for Adverse Weather Image Recovery", "abstract": "Restoring images degraded by adverse weather remains a significant challenge due to the highly non-uniform and spatially heterogeneous nature of weather-induced artifacts, e.g., fine-grained rain streaks versus widespread haze. Accurately estimating the underlying degradation can intuitively provide restoration models with more targeted and effective guidance, enabling adaptive processing strategies. To this end, we propose a Morton-Order Degradation Estimation Mechanism (MODEM) for adverse weather image restoration. Central to MODEM is the Morton-Order 2D-Selective-Scan Module (MOS2D), which integrates Morton-coded spatial ordering with selective state-space models to capture long-range dependencies while preserving local structural coherence. Complementing MOS2D, we introduce a Dual Degradation Estimation Module (DDEM) that disentangles and estimates both global and local degradation priors. These priors dynamically condition the MOS2D modules, facilitating adaptive and context-aware restoration. Extensive experiments and ablation studies demonstrate that MODEM achieves state-of-the-art results across multiple benchmarks and weather types, highlighting its effectiveness in modeling complex degradation dynamics. Our code will be released at https://github.com/hainuo-wang/MODEM.git.", "date": "2025-05-23", "categories": []}
{"id": "2510.20550", "title": "From Cheap to Pro: A Learning-based Adaptive Camera Parameter Network for Professional-Style Imaging", "abstract": "Consumer-grade camera systems often struggle to maintain stable image quality under complex illumination conditions such as low light, high dynamic range, and backlighting, as well as spatial color temperature variation. These issues lead to underexposure, color casts, and tonal inconsistency, which degrade the performance of downstream vision tasks. To address this, we propose ACamera-Net, a lightweight and scene-adaptive camera parameter adjustment network that directly predicts optimal exposure and white balance from RAW inputs. The framework consists of two modules: ACamera-Exposure, which estimates ISO to alleviate underexposure and contrast loss, and ACamera-Color, which predicts correlated color temperature and gain factors for improved color consistency. Optimized for real-time inference on edge devices, ACamera-Net can be seamlessly integrated into imaging pipelines. Trained on diverse real-world data with annotated references, the model generalizes well across lighting conditions. Extensive experiments demonstrate that ACamera-Net consistently enhances image quality and stabilizes perception outputs, outperforming conventional auto modes and lightweight baselines without relying on additional image enhancement modules.", "date": "2025-10-23", "categories": []}
{"id": "2412.01463", "title": "Learning Differential Pyramid Representation for Tone Mapping", "abstract": "Existing tone mapping methods operate on downsampled inputs and rely on handcrafted pyramids to recover high-frequency details. These designs typically fail to preserve fine textures and structural fidelity in complex HDR scenes. Furthermore, most methods lack an effective mechanism to jointly model global tone consistency and local contrast enhancement, leading to globally flat or locally inconsistent outputs such as halo artifacts. We present the Differential Pyramid Representation Network (DPRNet), an end-to-end framework for high-fidelity tone mapping. At its core is a learnable differential pyramid that generalizes traditional Laplacian and Difference-of-Gaussian pyramids through content-aware differencing operations across scales. This allows DPRNet to adaptively capture high-frequency variations under diverse luminance and contrast conditions. To enforce perceptual consistency, DPRNet incorporates global tone perception and local tone tuning modules operating on downsampled inputs, enabling efficient yet expressive tone adaptation. Finally, an iterative detail enhancement module progressively restores the full-resolution output in a coarse-to-fine manner, reinforcing structure and sharpness. Experiments show that DPRNet achieves state-of-the-art results, improving PSNR by 2.39 dB on the 4K HDR+ dataset and 3.01 dB on the 4K HDRI Haven dataset, while producing perceptually coherent, detail-preserving results. \\textit{We provide an anonymous online demo at https://xxxxxxdprnet.github.io/DPRNet/.", "date": "2024-12-02", "categories": []}
{"id": "2509.08967", "title": "Physics-informed waveform inversion using pretrained wavefield neural operators", "abstract": "Full waveform inversion (FWI) is crucial for reconstructing high-resolution subsurface models, but it is often hindered, considering the limited data, by its null space resulting in low-resolution models, and more importantly, by its computational cost, especially if needed for real-time applications. Recent attempts to accelerate FWI using learned wavefield neural operators have shown promise in efficiency and differentiability, but typically suffer from noisy and unstable inversion performance. To address these limitations, we introduce a novel physics-informed FWI framework to enhance the inversion in accuracy while maintaining the efficiency of neural operator-based FWI. Instead of relying only on the L2 norm objective function via automatic differentiation, resulting in noisy model reconstruction, we integrate a physics constraint term in the loss function of FWI, improving the quality of the inverted velocity models. Specifically, starting with an initial model to simulate wavefields and then evaluating the loss over how much the resulting wavefield obeys the physical laws (wave equation) and matches the recorded data, we achieve a reduction in noise and artifacts. Numerical experiments using the OpenFWI and Overthrust models demonstrate our method's superior performance, offering cleaner and more accurate subsurface velocity than vanilla approaches. Considering the efficiency of the approach compared to FWI, this advancement represents a significant step forward in the practical application of FWI for real-time subsurface monitoring.", "date": "2025-09-10", "categories": []}
{"id": "2510.19472", "title": "Predicting before Reconstruction: A generative prior framework for MRI acceleration", "abstract": "Recent advancements in artificial intelligence have created transformative capabilities in image synthesis and generation, enabling diverse research fields to innovate at revolutionary speed and spectrum. In this study, we leverage this generative power to introduce a new paradigm for accelerating Magnetic Resonance Imaging (MRI), introducing a shift from image reconstruction to proactive predictive imaging. Despite being a cornerstone of modern patient care, MRI's lengthy acquisition times limit clinical throughput. Our novel framework addresses this challenge by first predicting a target contrast image, which then serves as a data-driven prior for reconstructing highly under-sampled data. This informative prior is predicted by a generative model conditioned on diverse data sources, such as other contrast images, previously scanned images, acquisition parameters, patient information. We demonstrate this approach with two key applications: (1) reconstructing FLAIR images using predictions from T1w and/or T2w scans, and (2) reconstructing T1w images using predictions from previously acquired T1w scans. The framework was evaluated on internal and multiple public datasets (total 14,921 scans; 1,051,904 slices), including multi-channel k-space data, for a range of high acceleration factors (x4, x8 and x12). The results demonstrate that our prediction-prior reconstruction method significantly outperforms other approaches, including those with alternative or no prior information. Through this framework we introduce a fundamental shift from image reconstruction towards a new paradigm of predictive imaging.", "date": "2025-10-22", "categories": []}
{"id": "2510.18400", "title": "Bayesian Fully-Connected Tensor Network for Hyperspectral-Multispectral Image Fusion", "abstract": "Tensor decomposition is a powerful tool for data analysis and has been extensively employed in the field of hyperspectral-multispectral image fusion (HMF). Existing tensor decomposition-based fusion methods typically rely on disruptive data vectorization/reshaping or impose rigid constraints on the arrangement of factor tensors, hindering the preservation of spatial-spectral structures and the modeling of cross-dimensional correlations. Although recent advances utilizing the Fully-Connected Tensor Network (FCTN) decomposition have partially alleviated these limitations, the process of reorganizing data into higher-order tensors still disrupts the intrinsic spatial-spectral structure. Furthermore, these methods necessitate extensive manual parameter tuning and exhibit limited robustness against noise and spatial degradation. To alleviate these issues, we propose the Bayesian FCTN (BFCTN) method. Within this probabilistic framework, a hierarchical sparse prior that characterizing the sparsity of physical elements, establishes connections between the factor tensors. This framework explicitly models the intrinsic physical coupling among spatial structures, spectral signatures, and local scene homogeneity. For model learning, we develop a parameter estimation method based on Variational Bayesian inference (VB) and the Expectation-Maximization (EM) algorithm, which significantly reduces the need for manual parameter tuning. Extensive experiments demonstrate that BFCTN not only achieves state-of-the-art fusion accuracy and strong robustness but also exhibits practical applicability in complex real-world scenarios.", "date": "2025-10-21", "categories": []}
{"id": "2510.18851", "title": "DP$^2$O-SR: Direct Perceptual Preference Optimization for Real-World Image Super-Resolution", "abstract": "Benefiting from pre-trained text-to-image (T2I) diffusion models, real-world image super-resolution (Real-ISR) methods can synthesize rich and realistic details. However, due to the inherent stochasticity of T2I models, different noise inputs often lead to outputs with varying perceptual quality. Although this randomness is sometimes seen as a limitation, it also introduces a wider perceptual quality range, which can be exploited to improve Real-ISR performance. To this end, we introduce Direct Perceptual Preference Optimization for Real-ISR (DP$^2$O-SR), a framework that aligns generative models with perceptual preferences without requiring costly human annotations. We construct a hybrid reward signal by combining full-reference and no-reference image quality assessment (IQA) models trained on large-scale human preference datasets. This reward encourages both structural fidelity and natural appearance. To better utilize perceptual diversity, we move beyond the standard best-vs-worst selection and construct multiple preference pairs from outputs of the same model. Our analysis reveals that the optimal selection ratio depends on model capacity: smaller models benefit from broader coverage, while larger models respond better to stronger contrast in supervision. Furthermore, we propose hierarchical preference optimization, which adaptively weights training pairs based on intra-group reward gaps and inter-group diversity, enabling more efficient and stable learning. Extensive experiments across both diffusion- and flow-based T2I backbones demonstrate that DP$^2$O-SR significantly improves perceptual quality and generalizes well to real-world benchmarks.", "date": "2025-10-21", "categories": []}
{"id": "2409.13482", "title": "Invertible ResNets for Inverse Imaging Problems: Competitive Performance with Provable Regularization Properties", "abstract": "Learning-based methods have demonstrated remarkable performance in solving inverse problems, particularly in image reconstruction tasks. Despite their success, these approaches often lack theoretical guarantees, which are crucial in sensitive applications such as medical imaging. Recent works by Arndt et al addressed this gap by analyzing a data-driven reconstruction method based on invertible residual networks (iResNets). They revealed that, under reasonable assumptions, this approach constitutes a convergent regularization scheme. However, the performance of the reconstruction method was only validated on academic toy problems and small-scale iResNet architectures. In this work, we address this gap by evaluating the performance of iResNets on two real-world imaging tasks: a linear blurring operator and a nonlinear diffusion operator. To do so, we compare the performance of iResNets against state-of-the-art neural networks, revealing their competitiveness at the expense of longer training times. Moreover, we numerically demonstrate the advantages of the iResNet's inherent stability and invertibility by showcasing increased robustness across various scenarios as well as interpretability of the learned operator, thereby reducing the black-box nature of the reconstruction scheme.", "date": "2024-09-20", "categories": []}
{"id": "2503.12593", "title": "Fourier-Based 3D Multistage Transformer for Aberration Correction in Multicellular Specimens", "abstract": "High-resolution tissue imaging is often compromised by sample-induced optical aberrations that degrade resolution and contrast. While wavefront sensor-based adaptive optics (AO) can measure these aberrations, such hardware solutions are typically complex, expensive to implement, and slow when serially mapping spatially varying aberrations across large fields of view. Here, we introduce AOViFT (Adaptive Optical Vision Fourier Transformer) -- a machine learning-based aberration sensing framework built around a 3D multistage Vision Transformer that operates on Fourier domain embeddings. AOViFT infers aberrations and restores diffraction-limited performance in puncta-labeled specimens with substantially reduced computational cost, training time, and memory footprint compared to conventional architectures or real-space networks. We validated AOViFT on live gene-edited zebrafish embryos, demonstrating its ability to correct spatially varying aberrations using either a deformable mirror or post-acquisition deconvolution. By eliminating the need for the guide star and wavefront sensing hardware and simplifying the experimental workflow, AOViFT lowers technical barriers for high-resolution volumetric microscopy across diverse biological samples.", "date": "2025-03-16", "categories": []}
{"id": "2505.08909", "title": "Learning Cocoercive Conservative Denoisers via Helmholtz Decomposition for Poisson Inverse Problems", "abstract": "Plug-and-play (PnP) methods with deep denoisers have shown impressive results in imaging problems. They typically require strong convexity or smoothness of the fidelity term and a (residual) non-expansive denoiser for convergence. These assumptions, however, are violated in Poisson inverse problems, and non-expansiveness can hinder denoising performance. To address these challenges, we propose a cocoercive conservative (CoCo) denoiser, which may be (residual) expansive, leading to improved denoising. By leveraging the generalized Helmholtz decomposition, we introduce a novel training strategy that combines Hamiltonian regularization to promote conservativeness and spectral regularization to ensure cocoerciveness. We prove that CoCo denoiser is a proximal operator of a weakly convex function, enabling a restoration model with an implicit weakly convex prior. The global convergence of PnP methods to a stationary point of this restoration model is established. Extensive experimental results demonstrate that our approach outperforms closely related methods in both visual quality and quantitative metrics.", "date": "2025-05-13", "categories": []}
{"id": "2510.16321", "title": "Time-Embedded Algorithm Unrolling for Computational MRI", "abstract": "Algorithm unrolling methods have proven powerful for solving the regularized least squares problem in computational magnetic resonance imaging (MRI). These approaches unfold an iterative algorithm with a fixed number of iterations, typically alternating between a neural network-based proximal operator for regularization, a data fidelity operation and auxiliary updates with learnable parameters. While the connection to optimization methods dictate that the proximal operator network should be shared across unrolls, this can introduce artifacts or blurring. Heuristically, practitioners have shown that using distinct networks may be beneficial, but this significantly increases the number of learnable parameters, making it challenging to prevent overfitting. To address these shortcomings, by taking inspirations from proximal operators with varying thresholds in approximate message passing (AMP) and the success of time-embedding in diffusion models, we propose a time-embedded algorithm unrolling scheme for inverse problems. Specifically, we introduce a novel perspective on the iteration-dependent proximal operation in vector AMP (VAMP) and the subsequent Onsager correction in the context of algorithm unrolling, framing them as a time-embedded neural network. Similarly, the scalar weights in the data fidelity operation and its associated Onsager correction are cast as time-dependent learnable parameters. Our extensive experiments on the fastMRI dataset, spanning various acceleration rates and datasets, demonstrate that our method effectively reduces aliasing artifacts and mitigates noise amplification, achieving state-of-the-art performance. Furthermore, we show that our time-embedding strategy extends to existing algorithm unrolling approaches, enhancing reconstruction quality without increasing the computational complexity significantly.", "date": "2025-10-18", "categories": []}
{"id": "2510.16664", "title": "HYDRA: HYbrid knowledge Distillation and spectral Reconstruction Algorithm for high channel hyperspectral camera applications", "abstract": "Hyperspectral images (HSI) promise to support a range of new applications in computer vision. Recent research has explored the feasibility of generalizable Spectral Reconstruction (SR), the problem of recovering a HSI from a natural three-channel color image in unseen scenarios.\n  However, previous Multi-Scale Attention (MSA) works have only demonstrated sufficient generalizable results for very sparse spectra, while modern HSI sensors contain hundreds of channels.\n  This paper introduces a novel approach to spectral reconstruction via our HYbrid knowledge Distillation and spectral Reconstruction Architecture (HYDRA).\n  Using a Teacher model that encapsulates latent hyperspectral image data and a Student model that learns mappings from natural images to the Teacher's encoded domain, alongside a novel training method, we achieve high-quality spectral reconstruction.\n  This addresses key limitations of prior SR models, providing SOTA performance across all metrics, including an 18\\% boost in accuracy, and faster inference times than current SOTA models at various channel depths.", "date": "2025-10-18", "categories": []}
{"id": "2510.12493", "title": "BSGS: Bi-stage 3D Gaussian Splatting for Camera Motion Deblurring", "abstract": "3D Gaussian Splatting has exhibited remarkable capabilities in 3D scene reconstruction. However, reconstructing high-quality 3D scenes from motion-blurred images caused by camera motion poses a significant challenge.The performance of existing 3DGS-based deblurring methods are limited due to their inherent mechanisms, such as extreme dependence on the accuracy of camera poses and inability to effectively control erroneous Gaussian primitives densification caused by motion blur. To solve these problems, we introduce a novel framework, Bi-Stage 3D Gaussian Splatting, to accurately reconstruct 3D scenes from motion-blurred images. BSGS contains two stages. First, Camera Pose Refinement roughly optimizes camera poses to reduce motion-induced distortions. Second, with fixed rough camera poses, Global RigidTransformation further corrects motion-induced blur distortions. To alleviate multi-subframe gradient conflicts, we propose a subframe gradient aggregation strategy to optimize both stages. Furthermore, a space-time bi-stage optimization strategy is introduced to dynamically adjust primitive densification thresholds and prevent premature noisy Gaussian generation in blurred regions. Comprehensive experiments verify the effectiveness of our proposed deblurring method and show its superiority over the state of the arts.Our source code is available at https://github.com/wsxujm/bsgs", "date": "2025-10-14", "categories": []}
{"id": "2510.13452", "title": "Near-Infrared Hyperspectral Imaging Applications in Food Analysis -- Improving Algorithms and Methodologies", "abstract": "This thesis investigates the application of near-infrared hyperspectral imaging (NIR-HSI) for food quality analysis. The investigation is conducted through four studies operating with five research hypotheses. For several analyses, the studies compare models based on convolutional neural networks (CNNs) and partial least squares (PLS). Generally, joint spatio-spectral analysis with CNNs outperforms spatial analysis with CNNs and spectral analysis with PLS when modeling parameters where chemical and physical visual information are relevant. When modeling chemical parameters with a 2-dimensional (2D) CNN, augmenting the CNN with an initial layer dedicated to performing spectral convolution enhances its predictive performance by learning a spectral preprocessing similar to that applied by domain experts. Still, PLS-based spectral modeling performs equally well for analysis of the mean content of chemical parameters in samples and is the recommended approach. Modeling the spatial distribution of chemical parameters with NIR-HSI is limited by the ability to obtain spatially resolved reference values. Therefore, a study used bulk mean references for chemical map generation of fat content in pork bellies. A PLS-based approach gave non-smooth chemical maps and pixel-wise predictions outside the range of 0-100\\%. Conversely, a 2D CNN augmented with a spectral convolution layer mitigated all issues arising with PLS. The final study attempted to model barley's germinative capacity by analyzing NIR spectra, RGB images, and NIR-HSI images. However, the results were inconclusive due to the dataset's low degree of germination. Additionally, this thesis has led to the development of two open-sourced Python packages. The first facilitates fast PLS-based modeling, while the second facilitates very fast cross-validation of PLS and other classical machine learning models with a new algorithm.", "date": "2025-10-15", "categories": []}
{"id": "2506.05864", "title": "CryoFastAR: Fast Cryo-EM Ab Initio Reconstruction Made Easy", "abstract": "Pose estimation from unordered images is fundamental for 3D reconstruction, robotics, and scientific imaging. Recent geometric foundation models, such as DUSt3R, enable end-to-end dense 3D reconstruction but remain underexplored in scientific imaging fields like cryo-electron microscopy (cryo-EM) for near-atomic protein reconstruction. In cryo-EM, pose estimation and 3D reconstruction from unordered particle images still depend on time-consuming iterative optimization, primarily due to challenges such as low signal-to-noise ratios (SNR) and distortions from the contrast transfer function (CTF). We introduce CryoFastAR, the first geometric foundation model that can directly predict poses from Cryo-EM noisy images for Fast ab initio Reconstruction. By integrating multi-view features and training on large-scale simulated cryo-EM data with realistic noise and CTF modulations, CryoFastAR enhances pose estimation accuracy and generalization. To enhance training stability, we propose a progressive training strategy that first allows the model to extract essential features under simpler conditions before gradually increasing difficulty to improve robustness. Experiments show that CryoFastAR achieves comparable quality while significantly accelerating inference over traditional iterative approaches on both synthetic and real datasets.", "date": "2025-06-06", "categories": []}
{"id": "2510.12123", "title": "Hardware-aware Coding Function Design for Compressive Single-Photon 3D Cameras", "abstract": "Single-photon cameras are becoming increasingly popular in time-of-flight 3D imaging because they can time-tag individual photons with extreme resolution. However, their performance is susceptible to hardware limitations, such as system bandwidth, maximum laser power, sensor data rates, and in-sensor memory and compute resources. Compressive histograms were recently introduced as a solution to the challenge of data rates through an online in-sensor compression of photon timestamp data. Although compressive histograms work within limited in-sensor memory and computational resources, they underperform when subjected to real-world illumination hardware constraints. To address this, we present a constrained optimization approach for designing practical coding functions for compressive single-photon 3D imaging. Using gradient descent, we jointly optimize an illumination and coding matrix (i.e., the coding functions) that adheres to hardware constraints. We show through extensive simulations that our coding functions consistently outperform traditional coding designs under both bandwidth and peak power constraints. This advantage is particularly pronounced in systems constrained by peak power. Finally, we show that our approach adapts to arbitrary parameterized impulse responses by evaluating it on a real-world system with a non-ideal impulse response function.", "date": "2025-10-14", "categories": []}
{"id": "2510.12646", "title": "Zero-Shot CFC: Fast Real-World Image Denoising based on Cross-Frequency Consistency", "abstract": "Zero-shot denoisers address the dataset dependency of deep-learning-based denoisers, enabling the denoising of unseen single images. Nonetheless, existing zero-shot methods suffer from long training times and rely on the assumption of noise independence and a zero-mean property, limiting their effectiveness in real-world denoising scenarios where noise characteristics are more complicated. This paper proposes an efficient and effective method for real-world denoising, the Zero-Shot denoiser based on Cross-Frequency Consistency (ZSCFC), which enables training and denoising with a single noisy image and does not rely on assumptions about noise distribution. Specifically, image textures exhibit position similarity and content consistency across different frequency bands, while noise does not. Based on this property, we developed cross-frequency consistency loss and an ultralight network to realize image denoising. Experiments on various real-world image datasets demonstrate that our ZSCFC outperforms other state-of-the-art zero-shot methods in terms of computational efficiency and denoising performance.", "date": "2025-10-14", "categories": []}
{"id": "2510.12778", "title": "Wavefront Coding for Accommodation-Invariant Near-Eye Displays", "abstract": "We present a new computational near-eye display method that addresses the vergence-accommodation conflict problem in stereoscopic displays through accommodation-invariance. Our system integrates a refractive lens eyepiece with a novel wavefront coding diffractive optical element, operating in tandem with a pre-processing convolutional neural network. We employ end-to-end learning to jointly optimize the wavefront-coding optics and the image pre-processing module. To implement this approach, we develop a differentiable retinal image formation model that accounts for limiting aperture and chromatic aberrations introduced by the eye optics. We further integrate the neural transfer function and the contrast sensitivity function into the loss model to account for related perceptual effects. To tackle off-axis distortions, we incorporate position dependency into the pre-processing module. In addition to conducting rigorous analysis based on simulations, we also fabricate the designed diffractive optical element and build a benchtop setup, demonstrating accommodation-invariance for depth ranges of up to four diopters.", "date": "2025-10-14", "categories": []}
{"id": "2510.12788", "title": "Efficient Real-World Deblurring using Single Images: AIM 2025 Challenge Report", "abstract": "This paper reviews the AIM 2025 Efficient Real-World Deblurring using Single Images Challenge, which aims to advance in efficient real-blur restoration. The challenge is based on a new test set based on the well known RSBlur dataset. Pairs of blur and degraded images in this dataset are captured using a double-camera system. Participant were tasked with developing solutions to effectively deblur these type of images while fulfilling strict efficiency constraints: fewer than 5 million model parameters and a computational budget under 200 GMACs. A total of 71 participants registered, with 4 teams finally submitting valid solutions. The top-performing approach achieved a PSNR of 31.1298 dB, showcasing the potential of efficient methods in this domain. This paper provides a comprehensive overview of the challenge, compares the proposed solutions, and serves as a valuable reference for researchers in efficient real-world image deblurring.", "date": "2025-10-14", "categories": []}
{"id": "2505.22313", "title": "Large-Area Fabrication-Aware Computational Diffractive Optics", "abstract": "Differentiable optics, as an emerging paradigm that jointly optimizes optics and (optional) image processing algorithms, has made innovative optical designs possible across a broad range of applications. Many of these systems utilize diffractive optical components (DOEs) for holography, PSF engineering, or wavefront shaping. Existing approaches have, however, mostly remained limited to laboratory prototypes, owing to a large quality gap between simulation and manufactured devices. We aim at lifting the fundamental technical barriers to the practical use of learned diffractive optical systems. To this end, we propose a fabrication-aware design pipeline for diffractive optics fabricated by direct-write grayscale lithography followed by nano-imprinting replication, which is directly suited for inexpensive mass production of large area designs. We propose a super-resolved neural lithography model that can accurately predict the 3D geometry generated by the fabrication process. This model can be seamlessly integrated into existing differentiable optics frameworks, enabling fabrication-aware, end-to-end optimization of computational optical systems. To tackle the computational challenges, we also devise tensor-parallel compute framework centered on distributing large-scale FFT computation across many GPUs. As such, we demonstrate large scale diffractive optics designs up to 32.16 mm $\\times$ 21.44 mm, simulated on grids of up to 128,640 by 85,760 feature points. We find adequate agreement between simulation and fabricated prototypes for applications such as holography and PSF engineering. We also achieve high image quality from an imaging system comprised only of a single DOE, with images processed only by a Wiener filter utilizing the simulation PSF. We believe our findings lift the fabrication limitations for real-world applications of diffractive optics and differentiable optical design.", "date": "2025-05-28", "categories": []}
{"id": "2506.00716", "title": "Fovea Stacking: Imaging with Dynamic Localized Aberration Correction", "abstract": "The desire for cameras with smaller form factors has recently lead to a push for exploring computational imaging systems with reduced optical complexity such as a smaller number of lens elements. Unfortunately such simplified optical systems usually suffer from severe aberrations, especially in off-axis regions, which can be difficult to correct purely in software. In this paper we introduce Fovea Stacking , a new type of imaging system that utilizes emerging dynamic optical components called deformable phase plates (DPPs) for localized aberration correction anywhere on the image sensor. By optimizing DPP deformations through a differentiable optical model, off-axis aberrations are corrected locally, producing a foveated image with enhanced sharpness at the fixation point - analogous to the eye's fovea. Stacking multiple such foveated images, each with a different fixation point, yields a composite image free from aberrations. To efficiently cover the entire field of view, we propose joint optimization of DPP deformations under imaging budget constraints. Due to the DPP device's non-linear behavior, we introduce a neural network-based control model for improved alignment between simulation-hardware performance. We further demonstrated that for extended depth-of-field imaging, fovea stacking outperforms traditional focus stacking in image quality. By integrating object detection or eye-tracking, the system can dynamically adjust the lens to track the object of interest-enabling real-time foveated video suitable for downstream applications such as surveillance or foveated virtual reality displays", "date": "2025-05-31", "categories": []}
{"id": "2506.08793", "title": "A PDE-Based Image Dehazing Method via Atmospheric Scattering Theory", "abstract": "This paper introduces a novel partial differential equation (PDE) framework for single-image dehazing. We embed the atmospheric scattering model into a PDE featuring edge-preserving diffusion and a nonlocal operator to maintain both local details and global structures. A key innovation is an adaptive regularization mechanism guided by the dark channel prior, which adjusts smoothing strength based on haze density. The framework's mathematical well-posedness is rigorously established by proving the existence and uniqueness of its weak solution in $H_0^1(\\Omega)$. An efficient, GPU-accelerated fixed-point solver is used for implementation. Experiments confirm our method achieves effective haze removal while preserving high image fidelity, offering a principled alternative to purely data-driven techniques.", "date": "2025-06-10", "categories": []}
{"id": "2507.16385", "title": "STAR: A Benchmark for Astronomical Star Fields Super-Resolution", "abstract": "Super-resolution (SR) advances astronomical imaging by enabling cost-effective high-resolution capture, crucial for detecting faraway celestial objects and precise structural analysis. However, existing datasets for astronomical SR (ASR) exhibit three critical limitations: flux inconsistency, object-crop setting, and insufficient data diversity, significantly impeding ASR development. We propose STAR, a large-scale astronomical SR dataset containing 54,738 flux-consistent star field image pairs covering wide celestial regions. These pairs combine Hubble Space Telescope high-resolution observations with physically faithful low-resolution counterparts generated through a flux-preserving data generation pipeline, enabling systematic development of field-level ASR models. To further empower the ASR community, STAR provides a novel Flux Error (FE) to evaluate SR models in physical view. Leveraging this benchmark, we propose a Flux-Invariant Super Resolution (FISR) model that could accurately infer the flux-consistent high-resolution images from input photometry, suppressing several SR state-of-the-art methods by 24.84% on a novel designed flux consistency metric, showing the priority of our method for astrophysics. Extensive experiments demonstrate the effectiveness of our proposed method and the value of our dataset. Code and models are available at https://github.com/GuoCheng12/STAR.", "date": "2025-07-22", "categories": []}
{"id": "2510.03356", "title": "Learned Display Radiance Fields with Lensless Cameras", "abstract": "Calibrating displays is a basic and regular task that content creators must perform to maintain optimal visual experience, yet it remains a troublesome issue. Measuring display characteristics from different viewpoints often requires specialized equipment and a dark room, making it inaccessible to most users. To avoid specialized hardware requirements in display calibrations, our work co-designs a lensless camera and an Implicit Neural Representation based algorithm for capturing display characteristics from various viewpoints. More specifically, our pipeline enables efficient reconstruction of light fields emitted from a display from a viewing cone of 46.6{\\deg} X 37.6{\\deg}. Our emerging pipeline paves the initial steps towards effortless display calibration and characterization.", "date": "2025-10-02", "categories": []}
{"id": "2510.10083", "title": "Enabling High-Quality In-the-Wild Imaging from Severely Aberrated Metalens Bursts", "abstract": "We tackle the challenge of robust, in-the-wild imaging using ultra-thin nanophotonic metalens cameras. Meta-lenses, composed of planar arrays of nanoscale scatterers, promise dramatic reductions in size and weight compared to conventional refractive optics. However, severe chromatic aberration, pronounced light scattering, narrow spectral bandwidth, and low light efficiency continue to limit their practical adoption. In this work, we present an end-to-end solution for in-the-wild imaging that pairs a metalens several times thinner than conventional optics with a bespoke multi-image restoration framework optimized for practical metalens cameras. Our method centers on a lightweight convolutional network paired with a memory-efficient burst fusion algorithm that adaptively corrects noise, saturation clipping, and lens-induced distortions across rapid sequences of extremely degraded metalens captures. Extensive experiments on diverse, real-world handheld captures demonstrate that our approach consistently outperforms existing burst-mode and single-image restoration techniques.These results point toward a practical route for deploying metalens-based cameras in everyday imaging applications.", "date": "2025-10-11", "categories": []}
{"id": "2510.10612", "title": "UltraScatter: Ray-Based Simulation of Ultrasound Scattering", "abstract": "Traditional ultrasound simulation methods solve wave equations numerically, achieving high accuracy but at substantial computational cost. Faster alternatives based on convolution with precomputed impulse responses remain relatively slow, often requiring several minutes to generate a full B-mode image. We introduce UltraScatter, a probabilistic ray tracing framework that models ultrasound scattering efficiently and realistically. Tissue is represented as a volumetric field of scattering probability and scattering amplitude, and ray interactions are simulated via free-flight delta tracking. Scattered rays are traced to the transducer, with phase information incorporated through a linear time-of-flight model. Integrated with plane-wave imaging and beamforming, our parallelized ray tracing architecture produces B-mode images within seconds. Validation with phantom data shows realistic speckle and inclusion patterns, positioning UltraScatter as a scalable alternative to wave-based methods.", "date": "2025-10-12", "categories": []}
{"id": "2411.15295", "title": "Frequency-Guided Posterior Sampling for Diffusion-Based Image Restoration", "abstract": "Image restoration aims to recover high-quality images from degraded observations. When the degradation process is known, the recovery problem can be formulated as an inverse problem, and in a Bayesian context, the goal is to sample a clean reconstruction given the degraded observation. Recently, modern pretrained diffusion models have been used for image restoration by modifying their sampling procedure to account for the degradation process. However, these methods often rely on certain approximations that can lead to significant errors and compromised sample quality. In this paper, we provide the first rigorous analysis of this approximation error for linear inverse problems under distributional assumptions on the space of natural images, demonstrating cases where previous works can fail dramatically. Motivated by our theoretical insights, we propose a simple modification to existing diffusion-based restoration methods. Our approach introduces a time-varying low-pass filter in the frequency domain of the measurements, progressively incorporating higher frequencies during the restoration process. We develop an adaptive curriculum for this frequency schedule based on the underlying data distribution. Our method significantly improves performance on challenging image restoration tasks including motion deblurring and image dehazing.", "date": "2024-11-22", "categories": []}
{"id": "2506.02680", "title": "Solving Inverse Problems with FLAIR", "abstract": "Flow-based latent generative models such as Stable Diffusion 3 are able to generate images with remarkable quality, even enabling photorealistic text-to-image generation. Their impressive performance suggests that these models should also constitute powerful priors for inverse imaging problems, but that approach has not yet led to comparable fidelity. There are several key obstacles: (i) the data likelihood term is usually intractable; (ii) learned generative models cannot be directly conditioned on the distorted observations, leading to conflicting objectives between data likelihood and prior; and (iii) the reconstructions can deviate from the observed data. We present FLAIR, a novel, training-free variational framework that leverages flow-based generative models as prior for inverse problems. To that end, we introduce a variational objective for flow matching that is agnostic to the type of degradation, and combine it with deterministic trajectory adjustments to guide the prior towards regions which are more likely under the posterior. To enforce exact consistency with the observed data, we decouple the optimization of the data fidelity and regularization terms. Moreover, we introduce a time-dependent calibration scheme in which the strength of the regularization is modulated according to off-line accuracy estimates. Results on standard imaging benchmarks demonstrate that FLAIR consistently outperforms existing diffusion- and flow-based methods in terms of reconstruction quality and sample diversity. Our code is available at https://inverseflair.github.io/.", "date": "2025-06-03", "categories": []}
{"id": "2510.08641", "title": "Interlaced dynamic XCT reconstruction with spatio-temporal implicit neural representations", "abstract": "In this work, we investigate the use of spatio-temporalImplicit Neural Representations (INRs) for dynamic X-ray computed tomography (XCT) reconstruction under interlaced acquisition schemes. The proposed approach combines ADMM-based optimization with INCODE, a conditioning framework incorporating prior knowledge, to enable efficient convergence. We evaluate our method under diverse acquisition scenarios, varying the severity of global undersampling, spatial complexity (quantified via spatial information), and noise levels. Across all settings, our model achieves strong performance and outperforms Time-Interlaced Model-Based Iterative Reconstruction (TIMBIR), a state-of-the-art model-based iterative method. In particular, we show that the inductive bias of the INR provides good robustness to moderate noise levels, and that introducing explicit noise modeling through a weighted least squares data fidelity term significantly improves performance in more challenging regimes. The final part of this work explores extensions toward a practical reconstruction framework. We demonstrate the modularity of our approach by explicitly modeling detector non-idealities, incorporating ring artifact correction directly within the reconstruction process. Additionally, we present a proof-of-concept 4D volumetric reconstruction by jointly optimizing over batched axial slices, an approach which opens up the possibilities for massive parallelization, a critical feature for processing large-scale datasets.", "date": "2025-10-09", "categories": []}
{"id": "2510.09205", "title": "3D Reconstruction from Transient Measurements with Time-Resolved Transformer", "abstract": "Transient measurements, captured by the timeresolved systems, are widely employed in photon-efficient reconstruction tasks, including line-of-sight (LOS) and non-line-of-sight (NLOS) imaging. However, challenges persist in their 3D reconstruction due to the low quantum efficiency of sensors and the high noise levels, particularly for long-range or complex scenes. To boost the 3D reconstruction performance in photon-efficient imaging, we propose a generic Time-Resolved Transformer (TRT) architecture. Different from existing transformers designed for high-dimensional data, TRT has two elaborate attention designs tailored for the spatio-temporal transient measurements. Specifically, the spatio-temporal self-attention encoders explore both local and global correlations within transient data by splitting or downsampling input features into different scales. Then, the spatio-temporal cross attention decoders integrate the local and global features in the token space, resulting in deep features with high representation capabilities. Building on TRT, we develop two task-specific embodiments: TRT-LOS for LOS imaging and TRT-NLOS for NLOS imaging. Extensive experiments demonstrate that both embodiments significantly outperform existing methods on synthetic data and real-world data captured by different imaging systems. In addition, we contribute a large-scale, high-resolution synthetic LOS dataset with various noise levels and capture a set of real-world NLOS measurements using a custom-built imaging system, enhancing the data diversity in this field. Code and datasets are available at https://github.com/Depth2World/TRT.", "date": "2025-10-10", "categories": []}
{"id": "2506.19445", "title": "Deblurring in the Wild: A Real-World Image Deblurring Dataset from Smartphone High-Speed Videos", "abstract": "We introduce the largest real-world image deblurring dataset constructed from smartphone slow-motion videos. Using 240 frames captured over one second, we simulate realistic long-exposure blur by averaging frames to produce blurry images, while using the temporally centered frame as the sharp reference. Our dataset contains over 42,000 high-resolution blur-sharp image pairs, making it approximately 10 times larger than widely used datasets, with 8 times the amount of different scenes, including indoor and outdoor environments, with varying object and camera motions. We benchmark multiple state-of-the-art (SOTA) deblurring models on our dataset and observe significant performance degradation, highlighting the complexity and diversity of our benchmark. Our dataset serves as a challenging new benchmark to facilitate robust and generalizable deblurring models.", "date": "2025-06-24", "categories": []}
{"id": "2510.07741", "title": "UltraLED: Learning to See Everything in Ultra-High Dynamic Range Scenes", "abstract": "Ultra-high dynamic range (UHDR) scenes exhibit significant exposure disparities between bright and dark regions. Such conditions are commonly encountered in nighttime scenes with light sources. Even with standard exposure settings, a bimodal intensity distribution with boundary peaks often emerges, making it difficult to preserve both highlight and shadow details simultaneously. RGB-based bracketing methods can capture details at both ends using short-long exposure pairs, but are susceptible to misalignment and ghosting artifacts. We found that a short-exposure image already retains sufficient highlight detail. The main challenge of UHDR reconstruction lies in denoising and recovering information in dark regions. In comparison to the RGB images, RAW images, thanks to their higher bit depth and more predictable noise characteristics, offer greater potential for addressing this challenge. This raises a key question: can we learn to see everything in UHDR scenes using only a single short-exposure RAW image? In this study, we rely solely on a single short-exposure frame, which inherently avoids ghosting and motion blur, making it particularly robust in dynamic scenes. To achieve that, we introduce UltraLED, a two-stage framework that performs exposure correction via a ratio map to balance dynamic range, followed by a brightness-aware RAW denoiser to enhance detail recovery in dark regions. To support this setting, we design a 9-stop bracketing pipeline to synthesize realistic UHDR images and contribute a corresponding dataset based on diverse scenes, using only the shortest exposure as input for reconstruction. Extensive experiments show that UltraLED significantly outperforms existing single-frame approaches. Our code and dataset are made publicly available at https://srameo.github.io/projects/ultraled.", "date": "2025-10-09", "categories": []}
{"id": "2510.08358", "title": "SPICE: Simple and Practical Image Clarification and Enhancement", "abstract": "We introduce a simple and efficient method to enhance and clarify images. More specifically, we deal with low light image enhancement and clarification of hazy imagery (hazy/foggy images, images containing sand dust, and underwater images). Our method involves constructing an image filter to simulate low-light or hazy conditions and deriving approximate reverse filters to minimize distortions in the enhanced images. Experimental results show that our approach is highly competitive and often surpasses state-of-the-art techniques in handling extremely dark images and in enhancing hazy images. A key advantage of our approach lies in its simplicity: Our method is implementable with just a few lines of MATLAB code.", "date": "2025-10-09", "categories": []}
{"id": "2510.06335", "title": "Conditional Denoising Diffusion Model-Based Robust MR Image Reconstruction from Highly Undersampled Data", "abstract": "Magnetic Resonance Imaging (MRI) is a critical tool in modern medical diagnostics, yet its prolonged acquisition time remains a critical limitation, especially in time-sensitive clinical scenarios. While undersampling strategies can accelerate image acquisition, they often result in image artifacts and degraded quality. Recent diffusion models have shown promise for reconstructing high-fidelity images from undersampled data by learning powerful image priors; however, most existing approaches either (i) rely on unsupervised score functions without paired supervision or (ii) apply data consistency only as a post-processing step. In this work, we introduce a conditional denoising diffusion framework with iterative data-consistency correction, which differs from prior methods by embedding the measurement model directly into every reverse diffusion step and training the model on paired undersampled-ground truth data. This hybrid design bridges generative flexibility with explicit enforcement of MRI physics. Experiments on the fastMRI dataset demonstrate that our framework consistently outperforms recent state-of-the-art deep learning and diffusion-based methods in SSIM, PSNR, and LPIPS, with LPIPS capturing perceptual improvements more faithfully. These results demonstrate that integrating conditional supervision with iterative consistency updates yields substantial improvements in both pixel-level fidelity and perceptual realism, establishing a principled and practical advance toward robust, accelerated MRI reconstruction.", "date": "2025-10-07", "categories": []}
{"id": "2510.06601", "title": "AIM 2025 Challenge on Real-World RAW Image Denoising", "abstract": "We introduce the AIM 2025 Real-World RAW Image Denoising Challenge, aiming to advance efficient and effective denoising techniques grounded in data synthesis. The competition is built upon a newly established evaluation benchmark featuring challenging low-light noisy images captured in the wild using five different DSLR cameras. Participants are tasked with developing novel noise synthesis pipelines, network architectures, and training methodologies to achieve high performance across different camera models. Winners are determined based on a combination of performance metrics, including full-reference measures (PSNR, SSIM, LPIPS), and non-reference ones (ARNIQA, TOPIQ). By pushing the boundaries of camera-agnostic low-light RAW image denoising trained on synthetic data, the competition promotes the development of robust and practical models aligned with the rapid progress in digital photography. We expect the competition outcomes to influence multiple domains, from image restoration to night-time autonomous driving.", "date": "2025-10-08", "categories": []}
{"id": "2510.06757", "title": "Transforming Noise Distributions with Histogram Matching: Towards a Single Denoiser for All", "abstract": "Supervised Gaussian denoisers exhibit limited generalization when confronted with out-of-distribution noise, due to the diverse distributional characteristics of different noise types. To bridge this gap, we propose a histogram matching approach that transforms arbitrary noise towards a target Gaussian distribution with known intensity. Moreover, a mutually reinforcing cycle is established between noise transformation and subsequent denoising. This cycle progressively refines the noise to be converted, making it approximate the real noise, thereby enhancing the noise transformation effect and further improving the denoising performance. We tackle specific noise complexities: local histogram matching handles signal-dependent noise, intrapatch permutation processes channel-related noise, and frequency-domain histogram matching coupled with pixel-shuffle down-sampling breaks spatial correlation. By applying these transformations, a single Gaussian denoiser gains remarkable capability to handle various out-of-distribution noises, including synthetic noises such as Poisson, salt-and-pepper and repeating pattern noises, as well as complex real-world noises. Extensive experiments demonstrate the superior generalization and effectiveness of our method.", "date": "2025-10-08", "categories": []}
{"id": "2402.04480", "title": "SAMCIRT: A Simultaneous Reconstruction and Affine Motion Compensation Technique for Four Dimensional Computed Tomography (4DCT)", "abstract": "The majority of the recent iterative approaches in 4DCT not only rely on nested iterations, thereby increasing computational complexity and constraining potential acceleration, but also fail to provide a theoretical proof of convergence for their proposed iterative schemes. On the other hand, the latest MATLAB and Python image processing toolboxes lack the implementation of analytic adjoints of affine motion operators for 3D object volumes, which does not allow gradient methods using exact derivatives towards affine motion parameters. In this work, we propose the Simultaneous Affine Motion-Compensated Image Reconstruction Technique (SAMCIRT)- an efficient iterative reconstruction scheme that combines image reconstruction and affine motion estimation in a single update step, based on the analytic adjoints of the motion operators then exact partial derivatives with respect to both the reconstruction and the affine motion parameters. Moreover, we prove the separated Lipschitz continuity of the objective function and its associated functions, including the gradient, which supports the convergence of our proposed iterative scheme, despite the non-convexity of the objective function with respect to the affine motion parameters. Results from simulation and real experiments show that our method outperforms the state-of-the-art CT reconstruction with affine motion correction methods in computational feasibility and projection distance. In particular, this allows accurate reconstruction for a real, nonstationary diamond, showing a novel application of 4DCT.", "date": "2024-02-07", "categories": []}
{"id": "2510.05926", "title": "A Warm-basis Method for Bridging Learning and Iteration: a Case Study in Fluorescence Molecular Tomography", "abstract": "Fluorescence Molecular Tomography (FMT) is a widely used non-invasive optical imaging technology in biomedical research. It usually faces significant accuracy challenges in depth reconstruction, and conventional iterative methods struggle with poor $z$-resolution even with advanced regularization. Supervised learning approaches can improve recovery accuracy but rely on large, high-quality paired training dataset that is often impractical to acquire in practice. This naturally raises the question of how learning-based approaches can be effectively combined with iterative schemes to yield more accurate and stable algorithms. In this work, we present a novel warm-basis iterative projection method (WB-IPM) and establish its theoretical underpinnings. The method is able to achieve significantly more accurate reconstructions than the learning-based and iterative-based methods. In addition, it allows a weaker loss function depending solely on the directional component of the difference between ground truth and neural network output, thereby substantially reducing the training effort. These features are justified by our error analysis as well as simulated and real-data experiments.", "date": "2025-10-07", "categories": []}
{"id": "2302.02030", "title": "Learning the Night Sky with Deep Generative Priors", "abstract": "Recovering sharper images from blurred observations, referred to as deconvolution, is an ill-posed problem where classical approaches often produce unsatisfactory results. In ground-based astronomy, combining multiple exposures to achieve images with higher signal-to-noise ratios is complicated by the variation of point-spread functions across exposures due to atmospheric effects. We develop an unsupervised multi-frame method for denoising, deblurring, and coadding images inspired by deep generative priors. We use a carefully chosen convolutional neural network architecture that combines information from multiple observations, regularizes the joint likelihood over these observations, and allows us to impose desired constraints, such as non-negativity of pixel values in the sharp, restored image. With an eye towards the Rubin Observatory, we analyze 4K by 4K Hyper Suprime-Cam exposures and obtain preliminary results which yield promising restored images and extracted source lists.", "date": "2023-02-03", "categories": []}
{"id": "2312.07784", "title": "Robust MRI Reconstruction by Smoothed Unrolling (SMUG)", "abstract": "As the popularity of deep learning (DL) in the field of magnetic resonance imaging (MRI) continues to rise, recent research has indicated that DL-based MRI reconstruction models might be excessively sensitive to minor input disturbances, including worst-case additive perturbations. This sensitivity often leads to unstable, aliased images. This raises the question of how to devise DL techniques for MRI reconstruction that can be robust to train-test variations. To address this problem, we propose a novel image reconstruction framework, termed Smoothed Unrolling (SMUG), which advances a deep unrolling-based MRI reconstruction model using a randomized smoothing (RS)-based robust learning approach. RS, which improves the tolerance of a model against input noises, has been widely used in the design of adversarial defense approaches for image classification tasks. Yet, we find that the conventional design that applies RS to the entire DL-based MRI model is ineffective. In this paper, we show that SMUG and its variants address the above issue by customizing the RS process based on the unrolling architecture of a DL-based MRI reconstruction model. Compared to the vanilla RS approach, we show that SMUG improves the robustness of MRI reconstruction with respect to a diverse set of instability sources, including worst-case and random noise perturbations to input measurements, varying measurement sampling rates, and different numbers of unrolling steps. Furthermore, we theoretically analyze the robustness of our method in the presence of perturbations.", "date": "2023-12-12", "categories": []}
{"id": "2504.07696", "title": "Conformalized Generative Bayesian Imaging: An Uncertainty Quantification Framework for Computational Imaging", "abstract": "Uncertainty quantification plays an important role in achieving trustworthy and reliable learning-based computational imaging. Recent advances in generative modeling and Bayesian neural networks have enabled the development of uncertainty-aware image reconstruction methods. Current generative model-based methods seek to quantify the inherent (aleatoric) uncertainty on the underlying image for given measurements by learning to sample from the posterior distribution of the underlying image. On the other hand, Bayesian neural network-based approaches aim to quantify the model (epistemic) uncertainty on the parameters of a deep neural network-based reconstruction method by approximating the posterior distribution of those parameters. Unfortunately, an ongoing need for an inversion method that can jointly quantify complex aleatoric uncertainty and epistemic uncertainty patterns still persists. In this paper, we present a scalable framework that can quantify both aleatoric and epistemic uncertainties. The proposed framework accepts an existing generative model-based posterior sampling method as an input and introduces an epistemic uncertainty quantification capability through Bayesian neural networks with latent variables and deep ensembling. Furthermore, by leveraging the conformal prediction methodology, the proposed framework can be easily calibrated to ensure rigorous uncertainty quantification. We evaluated the proposed framework on magnetic resonance imaging, computed tomography, and image inpainting problems and showed that the epistemic and aleatoric uncertainty estimates produced by the proposed framework display the characteristic features of true epistemic and aleatoric uncertainties. Furthermore, our results demonstrated that the use of conformal prediction on top of the proposed framework enables marginal coverage guarantees consistent with frequentist principles.", "date": "2025-04-10", "categories": []}
{"id": "2506.11893", "title": "Measurement-Aligned Sampling for Inverse Problem", "abstract": "Diffusion models provide a powerful way to incorporate complex prior information for solving inverse problems. However, existing methods struggle to correctly incorporate guidance from conflicting signals in the prior and measurement, and often failed to maximizing the consistency to the measurement, especially in the challenging setting of non-Gaussian or unknown noise. To address these issues, we propose Measurement-Aligned Sampling (MAS), a novel framework for linear inverse problem solving that flexibly balances prior and measurement information. MAS unifies and extends existing approaches such as DDNM, TMPD, while generalizing to handle both known Gaussian noise and unknown or non-Gaussian noise types. Extensive experiments demonstrate that MAS consistently outperforms state-of-the-art methods across a variety of tasks, while maintaining relatively low computational cost.", "date": "2025-06-13", "categories": []}
{"id": "2506.18217", "title": "Shape from Polarization of Thermal Emission and Reflection", "abstract": "Shape estimation for transparent objects is challenging due to their complex light transport. To circumvent these difficulties, we leverage the Shape from Polarization (SfP) technique in the Long-Wave Infrared (LWIR) spectrum, where most materials are opaque and emissive. While a few prior studies have explored LWIR SfP, these attempts suffered from significant errors due to inadequate polarimetric modeling, particularly the neglect of reflection. Addressing this gap, we formulated a polarization model that explicitly accounts for the combined effects of emission and reflection. Based on this model, we estimated surface normals using not only a direct model-based method but also a learning-based approach employing a neural network trained on a physically-grounded synthetic dataset. Furthermore, we modeled the LWIR polarimetric imaging process, accounting for inherent systematic errors to ensure accurate polarimetry. We implemented a prototype system and created ThermoPol, the first real-world benchmark dataset for LWIR SfP. Through comprehensive experiments, we demonstrated the high accuracy and broad applicability of our method across various materials, including those transparent in the visible spectrum.", "date": "2025-06-23", "categories": []}
{"id": "2510.03431", "title": "Application of a Virtual Imaging Framework for Investigating a Deep Learning-Based Reconstruction Method for 3D Quantitative Photoacoustic Computed Tomography", "abstract": "Quantitative photoacoustic computed tomography (qPACT) is a promising imaging modality for estimating physiological parameters such as blood oxygen saturation. However, developing robust qPACT reconstruction methods remains challenging due to computational demands, modeling difficulties, and experimental uncertainties. Learning-based methods have been proposed to address these issues but remain largely unvalidated. Virtual imaging (VI) studies are essential for validating such methods early in development, before proceeding to less-controlled phantom or in vivo studies. Effective VI studies must employ ensembles of stochastically generated numerical phantoms that accurately reflect relevant anatomy and physiology. Yet, most prior VI studies for qPACT relied on overly simplified phantoms. In this work, a realistic VI testbed is employed for the first time to assess a representative 3D learning-based qPACT reconstruction method for breast imaging. The method is evaluated across subject variability and physical factors such as measurement noise and acoustic aberrations, offering insights into its strengths and limitations.", "date": "2025-10-03", "categories": []}
{"id": "2510.04069", "title": "Diffusion Low Rank Hybrid Reconstruction for Sparse View Medical Imaging", "abstract": "This work presents TV-LoRA, a novel method for low-dose sparse-view CT reconstruction that combines a diffusion generative prior (NCSN++ with SDE modeling) and multi-regularization constraints, including anisotropic TV and nuclear norm (LoRA), within an ADMM framework. To address ill-posedness and texture loss under extremely sparse views, TV-LoRA integrates generative and physical constraints, and utilizes a 2D slice-based strategy with FFT acceleration and tensor-parallel optimization for efficient inference. Experiments on AAPM-2016, CTHD, and LIDC datasets with $N_{\\mathrm{view}}=8,4,2$ show that TV-LoRA consistently surpasses benchmarks in SSIM, texture recovery, edge clarity, and artifact suppression, demonstrating strong robustness and generalizability. Ablation studies confirm the complementary effects of LoRA regularization and diffusion priors, while the FFT-PCG module provides a speedup. Overall, Diffusion + TV-LoRA achieves high-fidelity, efficient 3D CT reconstruction and broad clinical applicability in low-dose, sparse-sampling scenarios.", "date": "2025-10-05", "categories": []}
{"id": "2310.07504", "title": "PtychoDV: Vision Transformer-Based Deep Unrolling Network for Ptychographic Image Reconstruction", "abstract": "Ptychography is an imaging technique that captures multiple overlapping snapshots of a sample, illuminated coherently by a moving localized probe. The image recovery from ptychographic data is generally achieved via an iterative algorithm that solves a nonlinear phase retrieval problem derived from measured diffraction patterns. However, these iterative approaches have high computational cost. In this paper, we introduce PtychoDV, a novel deep model-based network designed for efficient, high-quality ptychographic image reconstruction. PtychoDV comprises a vision transformer that generates an initial image from the set of raw measurements, taking into consideration their mutual correlations. This is followed by a deep unrolling network that refines the initial image using learnable convolutional priors and the ptychography measurement model. Experimental results on simulated data demonstrate that PtychoDV is capable of outperforming existing deep learning methods for this problem, and significantly reduces computational cost compared to iterative methodologies, while maintaining competitive performance.", "date": "2023-10-11", "categories": []}
{"id": "2412.11108", "title": "Plug-and-Play Priors as a Score-Based Method", "abstract": "Plug-and-play (PnP) methods are extensively used for solving imaging inverse problems by integrating physical measurement models with pre-trained deep denoisers as priors. Score-based diffusion models (SBMs) have recently emerged as a powerful framework for image generation by training deep denoisers to represent the score of the image prior. While both PnP and SBMs use deep denoisers, the score-based nature of PnP is unexplored in the literature due to its distinct origins rooted in proximal optimization. This letter introduces a novel view of PnP as a score-based method, a perspective that enables the re-use of powerful SBMs within classical PnP algorithms without retraining. We present a set of mathematical relationships for adapting popular SBMs as priors within PnP. We show that this approach enables a direct comparison between PnP and SBM-based reconstruction methods using the same neural network as the prior. Code is available at https://github.com/wustl-cig/score_pnp.", "date": "2024-12-15", "categories": []}
{"id": "2403.01993", "title": "Physics-Informed Learning for Time-Resolved Angiographic Contrast Agent Concentration Reconstruction", "abstract": "Three-dimensional Digital Subtraction Angiography (3D-DSA) is a well-established X-ray-based technique for visualizing vascular anatomy. Recently, four-dimensional DSA (4D-DSA) reconstruction algorithms have been developed to enable the visualization of volumetric contrast flow dynamics through time-series of volumes. . This reconstruction problem is ill-posed mainly due to vessel overlap in the projection direction and geometric vessel foreshortening, which leads to information loss in the recorded projection images. However, knowledge about the underlying fluid dynamics can be leveraged to constrain the solution space. In our work, we implicitly include this information in a neural network-based model that is trained on a dataset of image-based blood flow simulations. The model predicts the spatially averaged contrast agent concentration for each centerline point of the vasculature over time, lowering the overall computational demand. The trained network enables the reconstruction of relative contrast agent concentrations with a mean absolute error of 0.02 $\\pm$ 0.02 and a mean absolute percentage error of 5.31 % $\\pm$ 9.25 %. Moreover, the network is robust to varying degrees of vessel overlap and vessel foreshortening. Our approach demonstrates the potential of the integration of machine learning and blood flow simulations in time-resolved angiographic flow reconstruction.", "date": "2024-03-04", "categories": []}
{"id": "2509.25075", "title": "GEM: 3D Gaussian Splatting for Efficient and Accurate Cryo-EM Reconstruction", "abstract": "Cryo-electron microscopy (cryo-EM) has become a central tool for high-resolution structural biology, yet the massive scale of datasets (often exceeding 100k particle images) renders 3D reconstruction both computationally expensive and memory intensive. Traditional Fourier-space methods are efficient but lose fidelity due to repeated transforms, while recent real-space approaches based on neural radiance fields (NeRFs) improve accuracy but incur cubic memory and computation overhead. Therefore, we introduce GEM, a novel cryo-EM reconstruction framework built on 3D Gaussian Splatting (3DGS) that operates directly in real-space while maintaining high efficiency. Instead of modeling the entire density volume, GEM represents proteins with compact 3D Gaussians, each parameterized by only 11 values. To further improve the training efficiency, we designed a novel gradient computation to 3D Gaussians that contribute to each voxel. This design substantially reduced both memory footprint and training cost. On standard cryo-EM benchmarks, GEM achieves up to 48% faster training and 12% lower memory usage compared to state-of-the-art methods, while improving local resolution by as much as 38.8%. These results establish GEM as a practical and scalable paradigm for cryo-EM reconstruction, unifying speed, efficiency, and high-resolution accuracy. Our code is available at https://github.com/UNITES-Lab/GEM.", "date": "2025-09-29", "categories": []}
{"id": "2510.01640", "title": "Joint Deblurring and 3D Reconstruction for Macrophotography", "abstract": "Macro lens has the advantages of high resolution and large magnification, and 3D modeling of small and detailed objects can provide richer information. However, defocus blur in macrophotography is a long-standing problem that heavily hinders the clear imaging of the captured objects and high-quality 3D reconstruction of them. Traditional image deblurring methods require a large number of images and annotations, and there is currently no multi-view 3D reconstruction method for macrophotography. In this work, we propose a joint deblurring and 3D reconstruction method for macrophotography. Starting from multi-view blurry images captured, we jointly optimize the clear 3D model of the object and the defocus blur kernel of each pixel. The entire framework adopts a differentiable rendering method to self-supervise the optimization of the 3D model and the defocus blur kernel. Extensive experiments show that from a small number of multi-view images, our proposed method can not only achieve high-quality image deblurring but also recover high-fidelity 3D appearance.", "date": "2025-10-02", "categories": []}
{"id": "2502.21120", "title": "SEE: See Everything Every Time -- Adaptive Brightness Adjustment for Broad Light Range Images via Events", "abstract": "Event cameras, with a high dynamic range exceeding $120dB$, significantly outperform traditional embedded cameras, robustly recording detailed changing information under various lighting conditions, including both low- and high-light situations. However, recent research on utilizing event data has primarily focused on low-light image enhancement, neglecting image enhancement and brightness adjustment across a broader range of lighting conditions, such as normal or high illumination. Based on this, we propose a novel research question: how to employ events to enhance and adaptively adjust the brightness of images captured under broad lighting conditions? To investigate this question, we first collected a new dataset, SEE-600K, consisting of 610,126 images and corresponding events across 202 scenarios, each featuring an average of four lighting conditions with over a 1000-fold variation in illumination. Subsequently, we propose a framework that effectively utilizes events to smoothly adjust image brightness through the use of prompts. Our framework captures color through sensor patterns, uses cross-attention to model events as a brightness dictionary, and adjusts the image's dynamic range to form a broad light-range representation (BLR), which is then decoded at the pixel level based on the brightness prompt. Experimental results demonstrate that our method not only performs well on the low-light enhancement dataset but also shows robust performance on broader light-range image enhancement using the SEE-600K dataset. Additionally, our approach enables pixel-level brightness adjustment, providing flexibility for post-processing and inspiring more imaging applications. The dataset and source code are publicly available at: https://github.com/yunfanLu/SEE.", "date": "2025-02-28", "categories": []}
{"id": "2503.11043", "title": "InverseBench: Benchmarking Plug-and-Play Diffusion Priors for Inverse Problems in Physical Sciences", "abstract": "Plug-and-play diffusion priors (PnPDP) have emerged as a promising research direction for solving inverse problems.\n  However, current studies primarily focus on natural image restoration, leaving the performance of these algorithms in scientific inverse problems largely unexplored. To address this gap, we introduce \\textsc{InverseBench}, a framework that evaluates diffusion models across five distinct scientific inverse problems. These problems present unique structural challenges that differ from existing benchmarks, arising from critical scientific applications such as optical tomography, medical imaging, black hole imaging, seismology, and fluid dynamics. With \\textsc{InverseBench}, we benchmark 14 inverse problem algorithms that use plug-and-play diffusion priors against strong, domain-specific baselines, offering valuable new insights into the strengths and weaknesses of existing algorithms. To facilitate further research and development, we open-source the codebase, along with datasets and pre-trained models, at https://devzhk.github.io/InverseBench/.", "date": "2025-03-14", "categories": []}
{"id": "2509.26203", "title": "Self-supervised learning for phase retrieval", "abstract": "In recent years, deep neural networks have emerged as a solution for inverse imaging problems. These networks are generally trained using pairs of images: one degraded and the other of high quality, the latter being called 'ground truth'. However, in medical and scientific imaging, the lack of fully sampled data limits supervised learning. Recent advances have made it possible to reconstruct images from measurement data alone, eliminating the need for references. However, these methods remain limited to linear problems, excluding non-linear problems such as phase retrieval. We propose a self-supervised method that overcomes this limitation in the case of phase retrieval by using the natural invariance of images to translations.", "date": "2025-09-30", "categories": []}
{"id": "2509.18497", "title": "Differentiable Light Transport with Gaussian Surfels via Adapted Radiosity for Efficient Relighting and Geometry Reconstruction", "abstract": "Radiance fields have gained tremendous success with applications ranging from novel view synthesis to geometry reconstruction, especially with the advent of Gaussian splatting. However, they sacrifice modeling of material reflective properties and lighting conditions, leading to significant geometric ambiguities and the inability to easily perform relighting. One way to address these limitations is to incorporate physically-based rendering, but it has been prohibitively expensive to include full global illumination within the inner loop of the optimization. Therefore, previous works adopt simplifications that make the whole optimization with global illumination effects efficient but less accurate. In this work, we adopt Gaussian surfels as the primitives and build an efficient framework for differentiable light transport, inspired from the classic radiosity theory. The whole framework operates in the coefficient space of spherical harmonics, enabling both diffuse and specular materials. We extend the classic radiosity into non-binary visibility and semi-opaque primitives, propose novel solvers to efficiently solve the light transport, and derive the backward pass for gradient optimizations, which is more efficient than auto-differentiation. During inference, we achieve view-independent rendering where light transport need not be recomputed under viewpoint changes, enabling hundreds of FPS for global illumination effects, including view-dependent reflections using a spherical harmonics representation. Through extensive qualitative and quantitative experiments, we demonstrate superior geometry reconstruction, view synthesis and relighting than previous inverse rendering baselines, or data-driven baselines given relatively sparse datasets with known or unknown lighting conditions.", "date": "2025-09-23", "categories": []}
{"id": "2509.23601", "title": "VAMamba: An Efficient Visual Adaptive Mamba for Image Restoration", "abstract": "Recent Mamba-based image restoration methods have achieved promising results but remain\n  limited by fixed scanning patterns and inefficient feature utilization. Conventional Mamba\n  architectures rely on predetermined paths that cannot adapt to diverse degradations, constraining\n  both restoration performance and computational efficiency. To overcome these limitations, we\n  propose VAMamba, a Visual Adaptive Mamba framework with two key innovations. First,\n  QCLAM(Queue-basedCacheLow-rankAdaptiveMemory)enhancesfeaturelearningthrougha\n  FIFO cache that stores historical representations. Similarity between current LoRA-adapted and\n  cached features guides intelligent fusion, enabling dynamic reuse while effectively controlling\n  memorygrowth.Second, GPS-SS2D(GreedyPathScanSS2D)introducesadaptive scanning. A\n  Vision Transformer generates score maps to estimate pixel importance, and a greedy strategy de termines optimal forward and backward scanning paths. These learned trajectories replace rigid\n  patterns, enabling SS2D to perform targeted feature extraction. The integration of QCLAM and\n  GPS-SS2D allows VAMamba to adaptively focus on degraded regions while maintaining high\n  computational efficiency. Extensive experiments across diverse restoration tasks demonstrate\n  that VAMamba consistently outperforms existing approaches in both restoration quality and\n  efficiency, establishing new benchmarks for adaptive image restoration. Our code is available\n  at https://github.com/WaterHQH/VAMamba.", "date": "2025-09-28", "categories": []}
{"id": "2509.24580", "title": "SAIP: A Plug-and-Play Scale-adaptive Module in Diffusion-based Inverse Problems", "abstract": "Solving inverse problems with diffusion models has shown promise in tasks such as image restoration. A common approach is to formulate the problem in a Bayesian framework and sample from the posterior by combining the prior score with the likelihood score. Since the likelihood term is often intractable, estimators like DPS, DMPS, and $\\pi$GDM are widely adopted. However, these methods rely on a fixed, manually tuned scale to balance prior and likelihood contributions. Such a static design is suboptimal, as the ideal balance varies across timesteps and tasks, limiting performance and generalization. To address this issue, we propose SAIP, a plug-and-play module that adaptively refines the scale at each timestep without retraining or altering the diffusion backbone. SAIP integrates seamlessly into existing samplers and consistently improves reconstruction quality across diverse image restoration tasks, including challenging scenarios.", "date": "2025-09-29", "categories": []}
{"id": "2509.25044", "title": "A Scalable Distributed Framework for Multimodal GigaVoxel Image Registration", "abstract": "In this work, we propose FFDP, a set of IO-aware non-GEMM fused kernels supplemented with a distributed framework for image registration at unprecedented scales. Image registration is an inverse problem fundamental to biomedical and life sciences, but algorithms have not scaled in tandem with image acquisition capabilities. Our framework complements existing model parallelism techniques proposed for large-scale transformer training by optimizing non-GEMM bottlenecks and enabling convolution-aware tensor sharding. We demonstrate unprecedented capabilities by performing multimodal registration of a 100 micron ex-vivo human brain MRI volume at native resolution - an inverse problem more than 570x larger than a standard clinical datum in about a minute using only 8 A6000 GPUs. FFDP accelerates existing state-of-the-art optimization and deep learning registration pipelines by upto 6 - 7x while reducing peak memory consumption by 20 - 59%. Comparative analysis on a 250 micron dataset shows that FFDP can fit upto 64x larger problems than existing SOTA on a single GPU, and highlights both the performance and efficiency gains of FFDP compared to SOTA image registration methods.", "date": "2025-09-29", "categories": []}
{"id": "2509.17427", "title": "Single-Image Depth from Defocus with Coded Aperture and Diffusion Posterior Sampling", "abstract": "We propose a single-snapshot depth-from-defocus (DFD) reconstruction method for coded-aperture imaging that replaces hand-crafted priors with a learned diffusion prior used purely as regularization. Our optimization framework enforces measurement consistency via a differentiable forward model while guiding solutions with the diffusion prior in the denoised image domain, yielding higher accuracy and stability than classical optimization. Unlike U-Net-style regressors, our approach requires no paired defocus-RGBD training data and does not tie training to a specific camera configuration. Experiments on comprehensive simulations and a prototype camera demonstrate consistently strong RGBD reconstructions across noise levels, outperforming both U-Net baselines and a classical coded-aperture DFD method.", "date": "2025-09-22", "categories": []}
{"id": "2509.21376", "title": "In silico Deep Learning Protocols for Label-Free Super-Resolution Microscopy: A Comparative Study of Network Architectures and SNR Dependence", "abstract": "The field of optical microscopy spans across numerous industries and research domains, ranging from education to healthcare, quality inspection and analysis. Nonetheless, a key limitation often cited by optical microscopists refers to the limit of its lateral resolution (typically defined as ~200nm), with potential circumventions involving either costly external modules (e.g. confocal scan heads, etc) and/or specialized techniques [e.g. super-resolution (SR) fluorescent microscopy]. Addressing these challenges in a normal (non-specialist) context thus remains an aspect outside the scope of most microscope users & facilities. This study thus seeks to evaluate an alternative & economical approach to achieving SR optical microscopy, involving non-fluorescent phase-modulated microscopical modalities such as Zernike phase contrast (PCM) and differential interference contrast (DIC) microscopy. Two in silico deep neural network (DNN) architectures which we developed previously (termed O-Net and Theta-Net) are assessed on their abilities to resolve a custom-fabricated test target containing nanoscale features calibrated via atomic force microscopy (AFM). The results of our study demonstrate that although both O-Net and Theta-Net seemingly performed well when super-resolving these images, they were complementary (rather than competing) approaches to be considered for image SR, particularly under different image signal-to-noise ratios (SNRs). High image SNRs favoured the application of O-Net models, while low SNRs inclined preferentially towards Theta-Net models. These findings demonstrate the importance of model architectures (in conjunction with the source image SNR) on model performance and the SR quality of the generated images where DNN models are utilized for non-fluorescent optical nanoscopy, even where the same training dataset & number of epochs are being used.", "date": "2025-09-23", "categories": []}
{"id": "2509.20417", "title": "Optimal Transport Based Hyperspectral Unmixing for Highly Mixed Observations", "abstract": "We propose a novel approach based on optimal transport (OT) for tackling the problem of highly mixed data in blind hyperspectral unmixing. Our method constrains the distribution of the estimated abundance matrix to resemble a targeted Dirichlet distribution more closely. The novelty lies in using OT to measure the discrepancy between the targeted and true abundance distributions, which we incorporate as a regularization term in our optimization problem. We demonstrate the efficiency of our method through a case study involving an unsupervised deep learning approach. Our experiments show that the proposed approach allows for a better estimation of the endmembers in the presence of highly mixed data, while displaying robustness to the choice of target abundance distribution.", "date": "2025-09-24", "categories": []}
{"id": "2412.10031", "title": "FM2S: Towards Spatially-Correlated Noise Modeling in Zero-Shot Fluorescence Microscopy Image Denoising", "abstract": "Fluorescence microscopy image (FMI) denoising faces critical challenges due to the compound mixed Poisson-Gaussian noise with strong spatial correlation and the impracticality of acquiring paired noisy/clean data in dynamic biomedical scenarios. While supervised methods trained on synthetic noise (e.g., Gaussian/Poisson) suffer from out-of-distribution generalization issues, existing self-supervised approaches degrade under real FMI noise due to oversimplified noise assumptions and computationally intensive deep architectures. In this paper, we propose Fluorescence Micrograph to Self (FM2S), a zero-shot denoiser that achieves efficient FMI denoising through three key innovations: 1) A noise injection module that ensures training data sufficiency through adaptive Poisson-Gaussian synthesis while preserving spatial correlation and global statistics of FMI noise for robust model generalization; 2) A two-stage progressive learning strategy that first recovers structural priors via pre-denoised targets then refines high-frequency details through noise distribution alignment; 3) An ultra-lightweight network (3.5k parameters) enabling rapid convergence with 270$\\times$ faster training and inference than SOTAs. Extensive experiments across FMI datasets demonstrate FM2S's superiority: It outperforms CVF-SID by 1.4dB PSNR on average while requiring 0.1% parameters of AP-BSN. Notably, FM2S maintains stable performance across varying noise levels, proving its practicality for microscopy platforms with diverse sensor characteristics. Code and datasets will be released.", "date": "2024-12-13", "categories": []}
{"id": "2509.18402", "title": "Measurement Score-Based MRI Reconstruction with Automatic Coil Sensitivity Estimation", "abstract": "Diffusion-based inverse problem solvers (DIS) have recently shown outstanding performance in compressed-sensing parallel MRI reconstruction by combining diffusion priors with physical measurement models. However, they typically rely on pre-calibrated coil sensitivity maps (CSMs) and ground truth images, making them often impractical: CSMs are difficult to estimate accurately under heavy undersampling and ground-truth images are often unavailable. We propose Calibration-free Measurement Score-based diffusion Model (C-MSM), a new method that eliminates these dependencies by jointly performing automatic CSM estimation and self-supervised learning of measurement scores directly from k-space data. C-MSM reconstructs images by approximating the full posterior distribution through stochastic sampling over partial measurement posterior scores, while simultaneously estimating CSMs. Experiments on the multi-coil brain fastMRI dataset show that C-MSM achieves reconstruction performance close to DIS with clean diffusion priors -- even without access to clean training data and pre-calibrated CSMs.", "date": "2025-09-22", "categories": []}
{"id": "2508.13479", "title": "AIM 2025 challenge on Inverse Tone Mapping Report: Methods and Results", "abstract": "This paper presents a comprehensive review of the AIM 2025 Challenge on Inverse Tone Mapping (ITM). The challenge aimed to push forward the development of effective ITM algorithms for HDR image reconstruction from single LDR inputs, focusing on perceptual fidelity and numerical consistency. A total of \\textbf{67} participants submitted \\textbf{319} valid results, from which the best five teams were selected for detailed analysis. This report consolidates their methodologies and performance, with the lowest PU21-PSNR among the top entries reaching 29.22 dB. The analysis highlights innovative strategies for enhancing HDR reconstruction quality and establishes strong benchmarks to guide future research in inverse tone mapping.", "date": "2025-08-19", "categories": []}
{"id": "2504.15756", "title": "DSDNet: Raw Domain Demoir\\'eing via Dual Color-Space Synergy", "abstract": "With the rapid advancement of mobile imaging, capturing screens using smartphones has become a prevalent practice in distance learning and conference recording. However, moir\\'e artifacts, caused by frequency aliasing between display screens and camera sensors, are further amplified by the image signal processing pipeline, leading to severe visual degradation. Existing sRGB domain demoir\\'eing methods struggle with irreversible information loss, while recent two-stage raw domain approaches suffer from information bottlenecks and inference inefficiency. To address these limitations, we propose a single-stage raw domain demoir\\'eing framework, Dual-Stream Demoir\\'eing Network (DSDNet), which leverages the synergy of raw and YCbCr images to remove moir\\'e while preserving luminance and color fidelity. Specifically, to guide luminance correction and moir\\'e removal, we design a raw-to-YCbCr mapping pipeline and introduce the Synergic Attention with Dynamic Modulation (SADM) module. This module enriches the raw-to-sRGB conversion with cross-domain contextual features. Furthermore, to better guide color fidelity, we develop a Luminance-Chrominance Adaptive Transformer (LCAT), which decouples luminance and chrominance representations. Extensive experiments demonstrate that DSDNet outperforms state-of-the-art methods in both visual quality and quantitative evaluation and achieves an inference speed $\\mathrm{\\textbf{2.4x}}$ faster than the second-best method, highlighting its practical advantages. We provide an anonymous online demo at https://xxxxxxxxdsdnet.github.io/DSDNet/.", "date": "2025-04-22", "categories": []}
{"id": "2509.15422", "title": "Analysis Plug-and-Play Methods for Imaging Inverse Problems", "abstract": "Plug-and-Play Priors (PnP) is a popular framework for solving imaging inverse problems by integrating learned priors in the form of denoisers trained to remove Gaussian noise from images. In standard PnP methods, the denoiser is applied directly in the image domain, serving as an implicit prior on natural images. This paper considers an alternative analysis formulation of PnP, in which the prior is imposed on a transformed representation of the image, such as its gradient. Specifically, we train a Gaussian denoiser to operate in the gradient domain, rather than on the image itself. Conceptually, this is an extension of total variation (TV) regularization to learned TV regularization. To incorporate this gradient-domain prior in image reconstruction algorithms, we develop two analysis PnP algorithms based on half-quadratic splitting (APnP-HQS) and the alternating direction method of multipliers (APnP-ADMM). We evaluate our approach on image deblurring and super-resolution, demonstrating that the analysis formulation achieves performance comparable to image-domain PnP algorithms.", "date": "2025-09-18", "categories": []}
{"id": "2509.16106", "title": "PRISM: Probabilistic and Robust Inverse Solver with Measurement-Conditioned Diffusion Prior for Blind Inverse Problems", "abstract": "Diffusion models are now commonly used to solve inverse problems in computational imaging. However, most diffusion-based inverse solvers require complete knowledge of the forward operator to be used. In this work, we introduce a novel probabilistic and robust inverse solver with measurement-conditioned diffusion prior (PRISM) to effectively address blind inverse problems. PRISM offers a technical advancement over current methods by incorporating a powerful measurement-conditioned diffusion model into a theoretically principled posterior sampling scheme. Experiments on blind image deblurring validate the effectiveness of the proposed method, demonstrating the superior performance of PRISM over state-of-the-art baselines in both image and blur kernel recovery.", "date": "2025-09-19", "categories": []}
{"id": "2509.14442", "title": "Indoor Airflow Imaging Using Physics-Informed Background-Oriented Schlieren Tomography", "abstract": "We develop a framework for non-invasive volumetric indoor airflow estimation from a single viewpoint using background-oriented schlieren (BOS) measurements and physics-informed reconstruction. Our framework utilizes a light projector that projects a pattern onto a target back-wall and a camera that observes small distortions in the light pattern. While the single-view BOS tomography problem is severely ill-posed, our proposed framework addresses this using: (1) improved ray tracing, (2) a physics-based light rendering approach and loss formulation, and (3) a physics-based regularization using a physics-informed neural network (PINN) to ensure that the reconstructed airflow is consistent with the governing equations for buoyancy-driven flows.", "date": "2025-09-17", "categories": []}
{"id": "2509.13863", "title": "LamiGauss: Pitching Radiative Gaussian for Sparse-View X-ray Laminography Reconstruction", "abstract": "X-ray Computed Laminography (CL) is essential for non-destructive inspection of plate-like structures in applications such as microchips and composite battery materials, where traditional computed tomography (CT) struggles due to geometric constraints. However, reconstructing high-quality volumes from laminographic projections remains challenging, particularly under highly sparse-view acquisition conditions. In this paper, we propose a reconstruction algorithm, namely LamiGauss, that combines Gaussian Splatting radiative rasterization with a dedicated detector-to-world transformation model incorporating the laminographic tilt angle. LamiGauss leverages an initialization strategy that explicitly filters out common laminographic artifacts from the preliminary reconstruction, preventing redundant Gaussians from being allocated to false structures and thereby concentrating model capacity on representing the genuine object. Our approach effectively optimizes directly from sparse projections, enabling accurate and efficient reconstruction with limited data. Extensive experiments on both synthetic and real datasets demonstrate the effectiveness and superiority of the proposed method over existing techniques. LamiGauss uses only 3$\\%$ of full views to achieve superior performance over the iterative method optimized on a full dataset.", "date": "2025-09-17", "categories": []}
{"id": "2212.08328", "title": "MEIL-NeRF: Memory-Efficient Incremental Learning of Neural Radiance Fields", "abstract": "Hinged on the representation power of neural networks, neural radiance fields (NeRF) have recently emerged as one of the promising and widely applicable methods for 3D object and scene representation. However, NeRF faces challenges in practical applications, such as large-scale scenes and edge devices with a limited amount of memory, where data needs to be processed sequentially. Under such incremental learning scenarios, neural networks are known to suffer catastrophic forgetting: easily forgetting previously seen data after training with new data. We observe that previous incremental learning algorithms are limited by either low performance or memory scalability issues. As such, we develop a Memory-Efficient Incremental Learning algorithm for NeRF (MEIL-NeRF). MEIL-NeRF takes inspiration from NeRF itself in that a neural network can serve as a memory that provides the pixel RGB values, given rays as queries. Upon the motivation, our framework learns which rays to query NeRF to extract previous pixel values. The extracted pixel values are then used to train NeRF in a self-distillation manner to prevent catastrophic forgetting. As a result, MEIL-NeRF demonstrates constant memory consumption and competitive performance.", "date": "2022-12-16", "categories": []}
{"id": "2412.13389", "title": "Marigold-DC: Zero-Shot Monocular Depth Completion with Guided Diffusion", "abstract": "Depth completion upgrades sparse depth measurements into dense depth maps guided by a conventional image. Existing methods for this highly ill-posed task operate in tightly constrained settings and tend to struggle when applied to images outside the training domain or when the available depth measurements are sparse, irregularly distributed, or of varying density. Inspired by recent advances in monocular depth estimation, we reframe depth completion as an image-conditional depth map generation guided by sparse measurements. Our method, Marigold-DC, builds on a pretrained latent diffusion model for monocular depth estimation and injects the depth observations as test-time guidance via an optimization scheme that runs in tandem with the iterative inference of denoising diffusion. The method exhibits excellent zero-shot generalization across a diverse range of environments and handles even extremely sparse guidance effectively. Our results suggest that contemporary monocular depth priors greatly robustify depth completion: it may be better to view the task as recovering dense depth from (dense) image pixels, guided by sparse depth; rather than as inpainting (sparse) depth, guided by an image. Project website: https://MarigoldDepthCompletion.github.io/", "date": "2024-12-18", "categories": []}
{"id": "2506.22426", "title": "Single-shot HDR using conventional image sensor shutter functions and optical randomization", "abstract": "High-dynamic-range (HDR) imaging is an essential technique for overcoming the dynamic range limits of image sensors. The classic method relies on multiple exposures, which slows capture time, resulting in motion artifacts when imaging dynamic scenes. Single-shot HDR imaging alleviates this issue by encoding HDR data into a single exposure, then computationally recovering it. Many established methods use strong image priors to recover improperly exposed image detail. These approaches struggle with extended highlight regions. We utilize the global reset release (GRR) shutter mode of an off-the-shelf sensor. GRR shutter mode applies a longer exposure time to rows closer to the bottom of the sensor. We use optics that relay a randomly permuted (shuffled) image onto the sensor, effectively creating spatially randomized exposures across the scene. The exposure diversity allows us to recover HDR data by solving an optimization problem with a simple total variation image prior. In simulation, we demonstrate that our method outperforms other single-shot methods when many sensor pixels are saturated (10% or more), and is competitive at a modest saturation (1%). Finally, we demonstrate a physical lab prototype that uses an off-the-shelf random fiber bundle for the optical shuffling. The fiber bundle is coupled to a low-cost commercial sensor operating in GRR shutter mode. Our prototype achieves a dynamic range of up to 73dB using an 8-bit sensor with 48dB dynamic range.", "date": "2025-06-27", "categories": []}
{"id": "2509.11087", "title": "SH-SAS: An Implicit Neural Representation for Complex Spherical-Harmonic Scattering Fields for 3D Synthetic Aperture Sonar", "abstract": "Synthetic aperture sonar (SAS) reconstruction requires recovering both the spatial distribution of acoustic scatterers and their direction-dependent response. Time-domain backprojection is the most common 3D SAS reconstruction algorithm, but it does not model directionality and can suffer from sampling limitations, aliasing, and occlusion. Prior neural volumetric methods applied to synthetic aperture sonar treat each voxel as an isotropic scattering density, not modeling anisotropic returns. We introduce SH-SAS, an implicit neural representation that expresses the complex acoustic scattering field as a set of spherical harmonic (SH) coefficients. A multi-resolution hash encoder feeds a lightweight MLP that outputs complex SH coefficients up to a specified degree L. The zeroth-order coefficient acts as an isotropic scattering field, which also serves as the density term, while higher orders compactly capture directional scattering with minimal parameter overhead. Because the model predicts the complex amplitude for any transmit-receive baseline, training is performed directly from 1-D time-of-flight signals without the need to beamform intermediate images for supervision. Across synthetic and real SAS (both in-air and underwater) benchmarks, results show that SH-SAS performs better in terms of 3D reconstruction quality and geometric metrics than previous methods.", "date": "2025-09-14", "categories": []}
{"id": "2509.09880", "title": "Automated Tuning for Diffusion Inverse Problem Solvers without Generative Prior Retraining", "abstract": "Diffusion/score-based models have recently emerged as powerful generative priors for solving inverse problems, including accelerated MRI reconstruction. While their flexibility allows decoupling the measurement model from the learned prior, their performance heavily depends on carefully tuned data fidelity weights, especially under fast sampling schedules with few denoising steps. Existing approaches often rely on heuristics or fixed weights, which fail to generalize across varying measurement conditions and irregular timestep schedules. In this work, we propose Zero-shot Adaptive Diffusion Sampling (ZADS), a test-time optimization method that adaptively tunes fidelity weights across arbitrary noise schedules without requiring retraining of the diffusion prior. ZADS treats the denoising process as a fixed unrolled sampler and optimizes fidelity weights in a self-supervised manner using only undersampled measurements. Experiments on the fastMRI knee dataset demonstrate that ZADS consistently outperforms both traditional compressed sensing and recent diffusion-based methods, showcasing its ability to deliver high-fidelity reconstructions across varying noise schedules and acquisition settings.", "date": "2025-09-11", "categories": []}
{"id": "2509.10098", "title": "Polarization Denoising and Demosaicking: Dataset and Baseline Method", "abstract": "A division-of-focal-plane (DoFP) polarimeter enables us to acquire images with multiple polarization orientations in one shot and thus it is valuable for many applications using polarimetric information. The image processing pipeline for a DoFP polarimeter entails two crucial tasks: denoising and demosaicking. While polarization demosaicking for a noise-free case has increasingly been studied, the research for the joint task of polarization denoising and demosaicking is scarce due to the lack of a suitable evaluation dataset and a solid baseline method. In this paper, we propose a novel dataset and method for polarization denoising and demosaicking. Our dataset contains 40 real-world scenes and three noise-level conditions, consisting of pairs of noisy mosaic inputs and noise-free full images. Our method takes a denoising-then-demosaicking approach based on well-accepted signal processing components to offer a reproducible method. Experimental results demonstrate that our method exhibits higher image reconstruction performance than other alternative methods, offering a solid baseline.", "date": "2025-09-12", "categories": []}
{"id": "2509.09352", "title": "Texture-aware Intrinsic Image Decomposition with Model- and Learning-based Priors", "abstract": "This paper aims to recover the intrinsic reflectance layer and shading layer given a single image. Though this intrinsic image decomposition problem has been studied for decades, it remains a significant challenge in cases of complex scenes, i.e. spatially-varying lighting effect and rich textures. In this paper, we propose a novel method for handling severe lighting and rich textures in intrinsic image decomposition, which enables to produce high-quality intrinsic images for real-world images. Specifically, we observe that previous learning-based methods tend to produce texture-less and over-smoothing intrinsic images, which can be used to infer the lighting and texture information given a RGB image. In this way, we design a texture-guided regularization term and formulate the decomposition problem into an optimization framework, to separate the material textures and lighting effect. We demonstrate that combining the novel texture-aware prior can produce superior results to existing approaches.", "date": "2025-09-11", "categories": []}
{"id": "2509.09365", "title": "Plug-and-play Diffusion Models for Image Compressive Sensing with Data Consistency Projection", "abstract": "We explore the connection between Plug-and-Play (PnP) methods and Denoising Diffusion Implicit Models (DDIM) for solving ill-posed inverse problems, with a focus on single-pixel imaging. We begin by identifying key distinctions between PnP and diffusion models-particularly in their denoising mechanisms and sampling procedures. By decoupling the diffusion process into three interpretable stages: denoising, data consistency enforcement, and sampling, we provide a unified framework that integrates learned priors with physical forward models in a principled manner. Building upon this insight, we propose a hybrid data-consistency module that linearly combines multiple PnP-style fidelity terms. This hybrid correction is applied directly to the denoised estimate, improving measurement consistency without disrupting the diffusion sampling trajectory. Experimental results on single-pixel imaging tasks demonstrate that our method achieves better reconstruction quality.", "date": "2025-09-11", "categories": []}
{"id": "2509.08712", "title": "Computational Imaging for Enhanced Computer Vision", "abstract": "This paper presents a comprehensive survey of computational imaging (CI) techniques and their transformative impact on computer vision (CV) applications. Conventional imaging methods often fail to deliver high-fidelity visual data in challenging conditions, such as low light, motion blur, or high dynamic range scenes, thereby limiting the performance of state-of-the-art CV systems. Computational imaging techniques, including light field imaging, high dynamic range (HDR) imaging, deblurring, high-speed imaging, and glare mitigation, address these limitations by enhancing image acquisition and reconstruction processes. This survey systematically explores the synergies between CI techniques and core CV tasks, including object detection, depth estimation, optical flow, face recognition, and keypoint detection. By analyzing the relationships between CI methods and their practical contributions to CV applications, this work highlights emerging opportunities, challenges, and future research directions. We emphasize the potential for task-specific, adaptive imaging pipelines that improve robustness, accuracy, and efficiency in real-world scenarios, such as autonomous navigation, surveillance, augmented reality, and robotics.", "date": "2025-09-10", "categories": []}
{"id": "2509.05541", "title": "Cryo-EM as a Stochastic Inverse Problem", "abstract": "Cryo-electron microscopy (Cryo-EM) enables high-resolution imaging of biomolecules, but structural heterogeneity remains a major challenge in 3D reconstruction. Traditional methods assume a discrete set of conformations, limiting their ability to recover continuous structural variability. In this work, we formulate cryo-EM reconstruction as a stochastic inverse problem (SIP) over probability measures, where the observed images are modeled as the push-forward of an unknown distribution over molecular structures via a random forward operator. We pose the reconstruction problem as the minimization of a variational discrepancy between observed and simulated image distributions, using statistical distances such as the KL divergence and the Maximum Mean Discrepancy. The resulting optimization is performed over the space of probability measures via a Wasserstein gradient flow, which we numerically solve using particles to represent and evolve conformational ensembles. We validate our approach using synthetic examples, including a realistic protein model, which demonstrates its ability to recover continuous distributions over structural states. We analyze the connection between our formulation and Maximum A Posteriori (MAP) approaches, which can be interpreted as instances of the discretize-then-optimize (DTO) framework. We further provide a consistency analysis, establishing conditions under which DTO methods, such as MAP estimation, converge to the solution of the underlying infinite-dimensional continuous problem. Beyond cryo-EM, the framework provides a general methodology for solving SIPs involving random forward operators.", "date": "2025-09-05", "categories": []}
{"id": "2509.05645", "title": "Stereovision Image Processing for Planetary Navigation Maps with Semi-Global Matching and Superpixel Segmentation", "abstract": "Mars exploration requires precise and reliable terrain models to ensure safe rover navigation across its unpredictable and often hazardous landscapes. Stereoscopic vision serves a critical role in the rover's perception, allowing scene reconstruction by generating precise depth maps through stereo matching. State-of-the-art Martian planetary exploration uses traditional local block-matching, aggregates cost over square windows, and refines disparities via smoothness constraints. However, this method often struggles with low-texture images, occlusion, and repetitive patterns because it considers only limited neighbouring pixels and lacks a wider understanding of scene context. This paper uses Semi-Global Matching (SGM) with superpixel-based refinement to mitigate the inherent block artefacts and recover lost details. The approach balances the efficiency and accuracy of SGM and adds context-aware segmentation to support more coherent depth inference. The proposed method has been evaluated in three datasets with successful results: In a Mars analogue, the terrain maps obtained show improved structural consistency, particularly in sloped or occlusion-prone regions. Large gaps behind rocks, which are common in raw disparity outputs, are reduced, and surface details like small rocks and edges are captured more accurately. Another two datasets, evaluated to test the method's general robustness and adaptability, show more precise disparity maps and more consistent terrain models, better suited for the demands of autonomous navigation on Mars, and competitive accuracy across both non-occluded and full-image error metrics. This paper outlines the entire terrain modelling process, from finding corresponding features to generating the final 2D navigation maps, offering a complete pipeline suitable for integration in future planetary exploration missions.", "date": "2025-09-06", "categories": []}
{"id": "2509.06793", "title": "AIM 2025 Challenge on High FPS Motion Deblurring: Methods and Results", "abstract": "This paper presents a comprehensive review of the AIM 2025 High FPS Non-Uniform Motion Deblurring Challenge, highlighting the proposed solutions and final results. The objective of this challenge is to identify effective networks capable of producing clearer and visually compelling images in diverse and challenging conditions, by learning representative visual cues for complex aggregations of motion types. A total of 68 participants registered for the competition, and 9 teams ultimately submitted valid entries. This paper thoroughly evaluates the state-of-the-art advances in high-FPS single image motion deblurring, showcasing the significant progress in the field, while leveraging samples of the novel dataset, MIORe, that introduces challenging examples of movement patterns.", "date": "2025-09-08", "categories": []}
{"id": "2509.04848", "title": "Pose-Free 3D Quantitative Phase Imaging of Flowing Cellular Populations", "abstract": "High-throughput 3D quantitative phase imaging (QPI) in flow cytometry enables label-free, volumetric characterization of individual cells by reconstructing their refractive index (RI) distributions from multiple viewing angles during flow through microfluidic channels. However, current imaging methods assume that cells undergo uniform, single-axis rotation, which require their poses to be known at each frame. This assumption restricts applicability to near-spherical cells and prevents accurate imaging of irregularly shaped cells with complex rotations. As a result, only a subset of the cellular population can be analyzed, limiting the ability of flow-based assays to perform robust statistical analysis. We introduce OmniFHT, a pose-free 3D RI reconstruction framework that leverages the Fourier diffraction theorem and implicit neural representations (INRs) for high-throughput flow cytometry tomographic imaging. By jointly optimizing each cell's unknown rotational trajectory and volumetric structure under weak scattering assumptions, OmniFHT supports arbitrary cell geometries and multi-axis rotations. Its continuous representation also allows accurate reconstruction from sparsely sampled projections and restricted angular coverage, producing high-fidelity results with as few as 10 views or only 120 degrees of angular range. OmniFHT enables, for the first time, in situ, high-throughput tomographic imaging of entire flowing cell populations, providing a scalable and unbiased solution for label-free morphometric analysis in flow cytometry platforms.", "date": "2025-09-05", "categories": []}
{"id": "2309.16494", "title": "Accurate and lightweight dehazing via multi-receptive-field non-local network and novel contrastive regularization", "abstract": "Recently, deep learning-based methods have dominated image dehazing domain. A multi-receptive-field non-local network (MRFNLN) consisting of the multi-stream feature attention block (MSFAB) and the cross non-local block (CNLB) is presented in this paper to further enhance the performance. We start with extracting richer features for dehazing. Specifically, a multi-stream feature extraction (MSFE) sub-block, which contains three parallel convolutions with different receptive fields (i.e., $1\\times 1$, $3\\times 3$, $5\\times 5$), is designed for extracting multi-scale features. Following MSFE, an attention sub-block is employed to make the model adaptively focus on important channels/regions. These two sub-blocks constitute our MSFAB. Then, we design a cross non-local block (CNLB), which can capture long-range dependencies beyond the query. Instead of the same input source of query branch, the key and value branches are enhanced by fusing more preceding features. CNLB is computation-friendly by leveraging a spatial pyramid down-sampling (SPDS) strategy to reduce the computation and memory consumption without sacrificing the performance. Last but not least, a novel detail-focused contrastive regularization (DFCR) is presented by emphasizing the low-level details and ignoring the high-level semantic information in a representation space specially designed for dehazing. Comprehensive experimental results demonstrate that the proposed MRFNLN model outperforms recent state-of-the-art dehazing methods with less than 1.5 Million parameters.", "date": "2023-09-28", "categories": []}
{"id": "2505.20789", "title": "Integrating Intermediate Layer Optimization and Projected Gradient Descent for Solving Inverse Problems with Diffusion Models", "abstract": "Inverse problems (IPs) involve reconstructing signals from noisy observations. Recently, diffusion models (DMs) have emerged as a powerful framework for solving IPs, achieving remarkable reconstruction performance. However, existing DM-based methods frequently encounter issues such as heavy computational demands and suboptimal convergence. In this work, building upon the idea of the recent work DMPlug, we propose two novel methods, DMILO and DMILO-PGD, to address these challenges. Our first method, DMILO, employs intermediate layer optimization (ILO) to alleviate the memory burden inherent in DMPlug. Additionally, by introducing sparse deviations, we expand the range of DMs, enabling the exploration of underlying signals that may lie outside the range of the diffusion model. We further propose DMILO-PGD, which integrates ILO with projected gradient descent (PGD), thereby reducing the risk of suboptimal convergence. We provide an intuitive theoretical analysis of our approaches under appropriate conditions and validate their superiority through extensive experiments on diverse image datasets, encompassing both linear and nonlinear IPs. Our results demonstrate significant performance gains over state-of-the-art methods, highlighting the effectiveness of DMILO and DMILO-PGD in addressing common challenges in DM-based IP solvers.", "date": "2025-05-27", "categories": []}
{"id": "2509.04047", "title": "TensoIS: A Step Towards Feed-Forward Tensorial Inverse Subsurface Scattering for Perlin Distributed Heterogeneous Media", "abstract": "Estimating scattering parameters of heterogeneous media from images is a severely under-constrained and challenging problem. Most of the existing approaches model BSSRDF either through an analysis-by-synthesis approach, approximating complex path integrals, or using differentiable volume rendering techniques to account for heterogeneity. However, only a few studies have applied learning-based methods to estimate subsurface scattering parameters, but they assume homogeneous media. Interestingly, no specific distribution is known to us that can explicitly model the heterogeneous scattering parameters in the real world. Notably, procedural noise models such as Perlin and Fractal Perlin noise have been effective in representing intricate heterogeneities of natural, organic, and inorganic surfaces. Leveraging this, we first create HeteroSynth, a synthetic dataset comprising photorealistic images of heterogeneous media whose scattering parameters are modeled using Fractal Perlin noise. Furthermore, we propose Tensorial Inverse Scattering (TensoIS), a learning-based feed-forward framework to estimate these Perlin-distributed heterogeneous scattering parameters from sparse multi-view image observations. Instead of directly predicting the 3D scattering parameter volume, TensoIS uses learnable low-rank tensor components to represent the scattering volume. We evaluate TensoIS on unseen heterogeneous variations over shapes from the HeteroSynth test set, smoke and cloud geometries obtained from open-source realistic volumetric simulations, and some real-world samples to establish its effectiveness for inverse scattering. Overall, this study is an attempt to explore Perlin noise distribution, given the lack of any such well-defined distribution in literature, to potentially model real-world heterogeneous scattering in a feed-forward manner.", "date": "2025-09-04", "categories": []}
{"id": "2509.04402", "title": "Learning neural representations for X-ray ptychography reconstruction with unknown probes", "abstract": "X-ray ptychography provides exceptional nanoscale resolution and is widely applied in materials science, biology, and nanotechnology. However, its full potential is constrained by the critical challenge of accurately reconstructing images when the illuminating probe is unknown. Conventional iterative methods and deep learning approaches are often suboptimal, particularly under the low-signal conditions inherent to low-dose and high-speed experiments. These limitations compromise reconstruction fidelity and restrict the broader adoption of the technique. In this work, we introduce the Ptychographic Implicit Neural Representation (PtyINR), a self-supervised framework that simultaneously addresses the object and probe recovery problem. By parameterizing both as continuous neural representations, PtyINR performs end-to-end reconstruction directly from raw diffraction patterns without requiring any pre-characterization of the probe. Extensive evaluations demonstrate that PtyINR achieves superior reconstruction quality on both simulated and experimental data, with remarkable robustness under challenging low-signal conditions. Furthermore, PtyINR offers a generalizable, physics-informed framework for addressing probe-dependent inverse problems, making it applicable to a wide range of computational microscopy problems.", "date": "2025-09-04", "categories": []}
{"id": "2503.12615", "title": "LATINO-PRO: LAtent consisTency INverse sOlver with PRompt Optimization", "abstract": "Text-to-image latent diffusion models (LDMs) have recently emerged as powerful generative models with great potential for solving inverse problems in imaging. However, leveraging such models in a Plug & Play (PnP), zero-shot manner remains challenging because it requires identifying a suitable text prompt for the unknown image of interest. Also, existing text-to-image PnP approaches are highly computationally expensive. We herein address these challenges by proposing a novel PnP inference paradigm specifically designed for embedding generative models within stochastic inverse solvers, with special attention to Latent Consistency Models (LCMs), which distill LDMs into fast generators. We leverage our framework to propose LAtent consisTency INverse sOlver (LATINO), the first zero-shot PnP framework to solve inverse problems with priors encoded by LCMs. Our conditioning mechanism avoids automatic differentiation and reaches SOTA quality in as little as 8 neural function evaluations. As a result, LATINO delivers remarkably accurate solutions and is significantly more memory and computationally efficient than previous approaches. We then embed LATINO within an empirical Bayesian framework that automatically calibrates the text prompt from the observed measurements by marginal maximum likelihood estimation. Extensive experiments show that prompt self-calibration greatly improves estimation, allowing LATINO with PRompt Optimization to define new SOTAs in image reconstruction quality and computational efficiency. The code is available at https://latino-pro.github.io", "date": "2025-03-16", "categories": []}
{"id": "2507.03094", "title": "Neural Dynamic Modes: Computational Imaging of Dynamical Systems from Sparse Observations", "abstract": "Dynamical systems are ubiquitous within science and engineering, from turbulent flow across aircraft wings to structural variability of proteins. Although some systems are well understood and simulated, scientific imaging often confronts never-before-seen dynamics observed through indirect, noisy, and highly sparse measurements. We present NeuralDMD, a model-free framework that combines neural implicit representations with Dynamic Mode Decomposition (DMD) to reconstruct continuous spatio-temporal dynamics from such measurements. The expressiveness of neural representations enables capturing complex spatial structures, while the linear dynamical modes of DMD introduce an inductive bias that guides training and supports stable, low-dimensional representations and forecasting. We validate NeuralDMD on two real-world problems: reconstructing near-surface wind-speed fields over North America from sparse station observations, and recovering the evolution of plasma near the Galactic-center black hole, Sgr A*. In both cases, NeuralDMD outperforms established baselines, demonstrating its potential as a general tool for imaging dynamical systems across geoscience, astronomy, and beyond.", "date": "2025-07-03", "categories": []}
{"id": "2507.06075", "title": "Discontinuity-aware Normal Integration for Generic Central Camera Models", "abstract": "Recovering a 3D surface from its surface normal map, a problem known as normal integration, is a key component for photometric shape reconstruction techniques such as shape-from-shading and photometric stereo. The vast majority of existing approaches for normal integration handle only implicitly the presence of depth discontinuities and are limited to orthographic or ideal pinhole cameras. In this paper, we propose a novel formulation that allows modeling discontinuities explicitly and handling generic central cameras. Our key idea is based on a local planarity assumption, that we model through constraints between surface normals and ray directions. Compared to existing methods, our approach more accurately approximates the relation between depth and surface normals, achieves state-of-the-art results on the standard normal integration benchmark, and is the first to directly handle generic central camera models.", "date": "2025-07-08", "categories": []}
{"id": "2507.22459", "title": "Exploiting Diffusion Prior for Task-driven Image Restoration", "abstract": "Task-driven image restoration (TDIR) has recently emerged to address performance drops in high-level vision tasks caused by low-quality (LQ) inputs. Previous TDIR methods struggle to handle practical scenarios in which images are degraded by multiple complex factors, leaving minimal clues for restoration. This motivates us to leverage the diffusion prior, one of the most powerful natural image priors. However, while the diffusion prior can help generate visually plausible results, using it to restore task-relevant details remains challenging, even when combined with recent TDIR methods. To address this, we propose EDTR, which effectively harnesses the power of diffusion prior to restore task-relevant details. Specifically, we propose directly leveraging useful clues from LQ images in the diffusion process by generating from pixel-error-based pre-restored LQ images with mild noise added. Moreover, we employ a small number of denoising steps to prevent the generation of redundant details that dilute crucial task-related information. We demonstrate that our method effectively utilizes diffusion prior for TDIR, significantly enhancing task performance and visual quality across diverse tasks with multiple complex degradations.", "date": "2025-07-30", "categories": []}
{"id": "2509.00395", "title": "Double-Constraint Diffusion Model with Nuclear Regularization for Ultra-low-dose PET Reconstruction", "abstract": "Ultra-low-dose positron emission tomography (PET) reconstruction holds significant potential for reducing patient radiation exposure and shortening examination times. However, it may also lead to increased noise and reduced imaging detail, which could decrease the image quality. In this study, we present a Double-Constraint Diffusion Model (DCDM), which freezes the weights of a pre-trained diffusion model and injects a trainable double-constraint controller into the encoding architecture, greatly reducing the number of trainable parameters for ultra-low-dose PET reconstruction. Unlike full fine-tuning models, DCDM can adapt to different dose levels without retraining all model parameters, thereby improving reconstruction flexibility. Specifically, the two constraint modules, named the Nuclear Transformer Constraint (NTC) and the Encoding Nexus Constraint (ENC), serve to refine the pre-trained diffusion model. The NTC leverages the nuclear norm as an approximation for matrix rank minimization, integrates the low-rank property into the Transformer architecture, and enables efficient information extraction from low-dose images and conversion into compressed feature representations in the latent space. Subsequently, the ENC utilizes these compressed feature representations to encode and control the pre-trained diffusion model, ultimately obtaining reconstructed PET images in the pixel space. In clinical reconstruction, the compressed feature representations from NTC help select the most suitable ENC for efficient unknown low-dose PET reconstruction. Experiments conducted on the UDPET public dataset and the Clinical dataset demonstrated that DCDM outperforms state-of-the-art methods on known dose reduction factors (DRF) and generalizes well to unknown DRF scenarios, proving valuable even at ultra-low dose levels, such as 1% of the full dose.", "date": "2025-08-30", "categories": []}
{"id": "2509.00777", "title": "IntrinsicReal: Adapting IntrinsicAnything from Synthetic to Real Objects", "abstract": "Estimating albedo (a.k.a., intrinsic image decomposition) from single RGB images captured in real-world environments (e.g., the MVImgNet dataset) presents a significant challenge due to the absence of paired images and their ground truth albedos. Therefore, while recent methods (e.g., IntrinsicAnything) have achieved breakthroughs by harnessing powerful diffusion priors, they remain predominantly trained on large-scale synthetic datasets (e.g., Objaverse) and applied directly to real-world RGB images, which ignores the large domain gap between synthetic and real-world data and leads to suboptimal generalization performance. In this work, we address this gap by proposing IntrinsicReal, a novel domain adaptation framework that bridges the above-mentioned domain gap for real-world intrinsic image decomposition. Specifically, our IntrinsicReal adapts IntrinsicAnything to the real domain by fine-tuning it using its high-quality output albedos selected by a novel dual pseudo-labeling strategy: i) pseudo-labeling with an absolute confidence threshold on classifier predictions, and ii) pseudo-labeling using the relative preference ranking of classifier predictions for individual input objects. This strategy is inspired by human evaluation, where identifying the highest-quality outputs is straightforward, but absolute scores become less reliable for sub-optimal cases. In these situations, relative comparisons of outputs become more accurate. To implement this, we propose a novel two-phase pipeline that sequentially applies these pseudo-labeling techniques to effectively adapt IntrinsicAnything to the real domain. Experimental results show that our IntrinsicReal significantly outperforms existing methods, achieving state-of-the-art results for albedo estimation on both synthetic and real-world datasets.", "date": "2025-08-31", "categories": []}
{"id": "2509.00917", "title": "DarkVRAI: Capture-Condition Conditioning and Burst-Order Selective Scan for Low-light RAW Video Denoising", "abstract": "Low-light RAW video denoising is a fundamentally challenging task due to severe signal degradation caused by high sensor gain and short exposure times, which are inherently limited by video frame rate requirements. To address this, we propose DarkVRAI, a novel framework that achieved first place in the AIM 2025 Low-light RAW Video Denoising Challenge. Our method introduces two primary contributions: (1) a successful application of a conditioning scheme for image denoising, which explicitly leverages capture metadata, to video denoising to guide the alignment and denoising processes, and (2) a Burst-Order Selective Scan (BOSS) mechanism that effectively models long-range temporal dependencies within the noisy video sequence. By synergistically combining these components, DarkVRAI demonstrates state-of-the-art performance on a rigorous and realistic benchmark dataset, setting a new standard for low-light video denoising.", "date": "2025-08-31", "categories": []}
{"id": "2509.01572", "title": "User Manual for Model-based Imaging Inverse Problem", "abstract": "This user manual is intended to provide a detailed description on model-based optimization for imaging inverse problem. Theseproblems can be particularly complex and challenging, especially for individuals without prior exposure to convex optimization orinverse problem theory, like myself. In light of this, I am writing this manual to clarify and systematically organize the mathematicalrationale underlying imaging inverse problems. This manual might not be accurate in mathmatical notion but more focus on the logicalthinking on how to solve and proceed to solve the problems. If you want to think deep about something, try to raise questions! Thismanual is seaprated into four sections, aiming to answer the following four questions: (1) What is inverse imaging problem? (2) Why optimization is used to solve the inverse imaging problem? (3) How to solve the optimization problem? (4) How to implement the optimization algorithm in real imaging system?", "date": "2025-09-01", "categories": []}
{"id": "2311.17744", "title": "Variational Bayes image restoration with compressive autoencoders", "abstract": "Regularization of inverse problems is of paramount importance in computational imaging. The ability of neural networks to learn efficient image representations has been recently exploited to design powerful data-driven regularizers. While state-of-the-art plug-and-play (PnP) methods rely on an implicit regularization provided by neural denoisers, alternative Bayesian approaches consider Maximum A Posteriori (MAP) estimation in the latent space of a generative model, thus with an explicit regularization. However, state-of-the-art deep generative models require a huge amount of training data compared to denoisers. Besides, their complexity hampers the optimization involved in latent MAP derivation. In this work, we first propose to use compressive autoencoders instead. These networks, which can be seen as variational autoencoders with a flexible latent prior, are smaller and easier to train than state-of-the-art generative models. As a second contribution, we introduce the Variational Bayes Latent Estimation (VBLE) algorithm, which performs latent estimation within the framework of variational inference. Thanks to a simple yet efficient parameterization of the variational posterior, VBLE allows for fast and easy (approximate) posterior sampling. Experimental results on image datasets BSD and FFHQ demonstrate that VBLE reaches similar performance as state-of-the-art PnP methods, while being able to quantify uncertainties significantly faster than other existing posterior sampling techniques. The code associated to this paper is available in https://github.com/MaudBqrd/VBLE.", "date": "2023-11-29", "categories": []}
{"id": "2501.05828", "title": "UltraRay: Introducing Full-Path Ray Tracing in Physics-Based Ultrasound Simulation", "abstract": "Traditional ultrasound simulators solve the wave equation to model pressure distribution fields, achieving high accuracy but requiring significant computational time and resources. To address this, ray tracing approaches have been introduced, modeling wave propagation as rays interacting with boundaries and scatterers. However, existing models simplify ray propagation, generating echoes at interaction points without considering return paths to the sensor. This can result in unrealistic artifacts and necessitates careful scene tuning for plausible results. We propose a novel ultrasound simulation pipeline that utilizes a ray tracing algorithm to generate echo data, tracing each ray from the transducer through the scene and back to the sensor. To replicate advanced ultrasound imaging, we introduce a ray emission scheme optimized for plane wave imaging, incorporating delay and steering capabilities. Furthermore, we integrate a standard signal processing pipeline to simulate end-to-end ultrasound image formation. We showcase the efficacy of the proposed pipeline by modeling synthetic scenes featuring highly reflective objects, such as bones. In doing so, our proposed approach, UltraRay, not only enhances the overall visual quality but also improves the realism of the simulated images by accurately capturing secondary reflections and reducing unnatural artifacts. By building on top of a differentiable framework, the proposed pipeline lays the groundwork for a fast and differentiable ultrasound simulation tool necessary for gradient-based optimization, enabling advanced ultrasound beamforming strategies, neural network integration, and accurate inverse scene reconstruction.", "date": "2025-01-10", "categories": []}
{"id": "2508.19579", "title": "High-Speed FHD Full-Color Video Computer-Generated Holography", "abstract": "Computer-generated holography (CGH) is a promising technology for next-generation displays. However, generating high-speed, high-quality holographic video requires both high frame rate display and efficient computation, but is constrained by two key limitations: ($i$) Learning-based models often produce over-smoothed phases with narrow angular spectra, causing severe color crosstalk in high frame rate full-color displays such as depth-division multiplexing and thus resulting in a trade-off between frame rate and color fidelity. ($ii$) Existing frame-by-frame optimization methods typically optimize frames independently, neglecting spatial-temporal correlations between consecutive frames and leading to computationally inefficient solutions. To overcome these challenges, in this paper, we propose a novel high-speed full-color video CGH generation scheme. First, we introduce Spectrum-Guided Depth Division Multiplexing (SGDDM), which optimizes phase distributions via frequency modulation, enabling high-fidelity full-color display at high frame rates. Second, we present HoloMamba, a lightweight asymmetric Mamba-Unet architecture that explicitly models spatial-temporal correlations across video sequences to enhance reconstruction quality and computational efficiency. Extensive simulated and real-world experiments demonstrate that SGDDM achieves high-fidelity full-color display without compromise in frame rate, while HoloMamba generates FHD (1080p) full-color holographic video at over 260 FPS, more than 2.6$\\times$ faster than the prior state-of-the-art Divide-Conquer-and-Merge Strategy.", "date": "2025-08-27", "categories": []}
{"id": "2508.19713", "title": "Inferring geometry and material properties from Mueller matrices with machine learning", "abstract": "Mueller matrices (MMs) encode information on geometry and material properties, but recovering both simultaneously is an ill-posed problem. We explore whether MMs contain sufficient information to infer surface geometry and material properties with machine learning. We use a dataset of spheres of various isotropic materials, with MMs captured over the full angular domain at five visible wavelengths (450-650 nm). We train machine learning models to predict material properties and surface normals using only these MMs as input. We demonstrate that, even when the material type is unknown, surface normals can be predicted and object geometry reconstructed. Moreover, MMs allow models to identify material types correctly. Further analyses show that diagonal elements are key for material characterization, and off-diagonal elements are decisive for normal estimation.", "date": "2025-08-27", "categories": []}
{"id": "2508.03775", "title": "4D-PreNet: A Unified Preprocessing Framework for 4D-STEM Data Analysis", "abstract": "Automated experimentation with real time data analysis in scanning transmission electron microscopy (STEM) often require end-to-end framework. The four-dimensional scanning transmission electron microscopy (4D-STEM) with high-throughput data acquisition has been constrained by the critical bottleneck results from data preprocessing. Pervasive noise, beam center drift, and elliptical distortions during high-throughput acquisition inevitably corrupt diffraction patterns, systematically biasing quantitative measurements. Yet, conventional correction algorithms are often material-specific and fail to provide a robust, generalizable solution. In this work, we present 4D-PreNet, an end-to-end deep-learning pipeline that integrates attention-enhanced U-Net and ResNet architectures to simultaneously perform denoising, center correction, and elliptical distortion calibration. The network is trained on large, simulated datasets encompassing a wide range of noise levels, drift magnitudes, and distortion types, enabling it to generalize effectively to experimental data acquired under varying conditions. Quantitative evaluations demonstrate that our pipeline reduces mean squared error by up to 50% during denoising and achieves sub-pixel center localization in the center detection task, with average errors below 0.04 pixels. The outputs are bench-marked against traditional algorithms, highlighting improvements in both noise suppression and restoration of diffraction patterns, thereby facilitating high-throughput, reliable 4D-STEM real-time analysis for automated characterization.", "date": "2025-08-05", "categories": []}
{"id": "2508.16579", "title": "Towards High-Precision Depth Sensing via Monocular-Aided iToF and RGB Integration", "abstract": "This paper presents a novel iToF-RGB fusion framework designed to address the inherent limitations of indirect Time-of-Flight (iToF) depth sensing, such as low spatial resolution, limited field-of-view (FoV), and structural distortion in complex scenes. The proposed method first reprojects the narrow-FoV iToF depth map onto the wide-FoV RGB coordinate system through a precise geometric calibration and alignment module, ensuring pixel-level correspondence between modalities. A dual-encoder fusion network is then employed to jointly extract complementary features from the reprojected iToF depth and RGB image, guided by monocular depth priors to recover fine-grained structural details and perform depth super-resolution. By integrating cross-modal structural cues and depth consistency constraints, our approach achieves enhanced depth accuracy, improved edge sharpness, and seamless FoV expansion. Extensive experiments on both synthetic and real-world datasets demonstrate that the proposed framework significantly outperforms state-of-the-art methods in terms of accuracy, structural consistency, and visual quality.", "date": "2025-08-03", "categories": []}
{"id": "2508.16830", "title": "AIM 2025 Low-light RAW Video Denoising Challenge: Dataset, Methods and Results", "abstract": "This paper reviews the AIM 2025 (Advances in Image Manipulation) Low-Light RAW Video Denoising Challenge. The task is to develop methods that denoise low-light RAW video by exploiting temporal redundancy while operating under exposure-time limits imposed by frame rate and adapting to sensor-specific, signal-dependent noise. We introduce a new benchmark of 756 ten-frame sequences captured with 14 smartphone camera sensors across nine conditions (illumination: 1/5/10 lx; exposure: 1/24, 1/60, 1/120 s), with high-SNR references obtained via burst averaging. Participants process linear RAW sequences and output the denoised 10th frame while preserving the Bayer pattern. Submissions are evaluated on a private test set using full-reference PSNR and SSIM, with final ranking given by the mean of per-metric ranks. This report describes the dataset, challenge protocol, and submitted approaches.", "date": "2025-08-22", "categories": []}
{"id": "2508.17816", "title": "UniSino: Physics-Driven Foundational Model for Universal CT Sinogram Standardization", "abstract": "During raw-data acquisition in CT imaging, diverse factors can degrade the collected sinograms, with undersampling and noise leading to severe artifacts and noise in reconstructed images and compromising diagnostic accuracy. Conventional correction methods rely on manually designed algorithms or fixed empirical parameters, but these approaches often lack generalizability across heterogeneous artifact types. To address these limitations, we propose UniSino, a foundation model for universal CT sinogram standardization. Unlike existing foundational models that operate in image domain, UniSino directly standardizes data in the projection domain, which enables stronger generalization across diverse undersampling scenarios. Its training framework incorporates the physical characteristics of sinograms, enhancing generalization and enabling robust performance across multiple subtasks spanning four benchmark datasets. Experimental results demonstrate thatUniSino achieves superior reconstruction quality both single and mixed undersampling case, demonstrating exceptional robustness and generalization in sinogram enhancement for CT imaging. The code is available at: https://github.com/yqx7150/UniSino.", "date": "2025-08-25", "categories": []}
{"id": "2304.12294", "title": "Explicit Correspondence Matching for Generalizable Neural Radiance Fields", "abstract": "We present a new generalizable NeRF method that is able to directly generalize to new unseen scenarios and perform novel view synthesis with as few as two source views. The key to our approach lies in the explicitly modeled correspondence matching information, so as to provide the geometry prior to the prediction of NeRF color and density for volume rendering. The explicit correspondence matching is quantified with the cosine similarity between image features sampled at the 2D projections of a 3D point on different views, which is able to provide reliable cues about the surface geometry. Unlike previous methods where image features are extracted independently for each view, we consider modeling the cross-view interactions via Transformer cross-attention, which greatly improves the feature matching quality. Our method achieves state-of-the-art results on different evaluation settings, with the experiments showing a strong correlation between our learned cosine feature similarity and volume density, demonstrating the effectiveness and superiority of our proposed method. The code and model are on our project page: https://donydchen.github.io/matchnerf", "date": "2023-04-24", "categories": []}
{"id": "2505.05054", "title": "Direct Image Classification from Fourier Ptychographic Microscopy Measurements without Reconstruction", "abstract": "The computational imaging technique of Fourier Ptychographic Microscopy (FPM) enables high-resolution imaging with a wide field of view and can serve as an extremely valuable tool, e.g. in the classification of cells in medical applications. However, reconstructing a high-resolution image from tens or even hundreds of measurements is computationally expensive, particularly for a wide field of view. Therefore, in this paper, we investigate the idea of classifying the image content in the FPM measurements directly without performing a reconstruction step first. We show that Convolutional Neural Networks (CNN) can extract meaningful information from measurement sequences, significantly outperforming the classification on a single band-limited image (up to 12 %) while being significantly more efficient than a reconstruction of a high-resolution image. Furthermore, we demonstrate that a learned multiplexing of several raw measurements allows maintaining the classification accuracy while reducing the amount of data (and consequently also the acquisition time) significantly.", "date": "2025-05-08", "categories": []}
{"id": "2508.15530", "title": "Self-supervised physics-informed generative networks for phase retrieval from a single X-ray hologram", "abstract": "X-ray phase contrast imaging significantly improves the visualization of structures with weak or uniform absorption, broadening its applications across a wide range of scientific disciplines. Propagation-based phase contrast is particularly suitable for time- or dose-critical in vivo/in situ/operando (tomography) experiments because it requires only a single intensity measurement. However, the phase information of the wave field is lost during the measurement and must be recovered. Conventional algebraic and iterative methods often rely on specific approximations or boundary conditions that may not be met by many samples or experimental setups. In addition, they require manual tuning of reconstruction parameters by experts, making them less adaptable for complex or variable conditions. Here we present a self-learning approach for solving the inverse problem of phase retrieval in the near-field regime of Fresnel theory using a single intensity measurement (hologram). A physics-informed generative adversarial network is employed to reconstruct both the phase and absorbance of the unpropagated wave field in the sample plane from a single hologram. Unlike most deep learning approaches for phase retrieval, our approach does not require paired, unpaired, or simulated training data. This significantly broadens the applicability of our approach, as acquiring or generating suitable training data remains a major challenge due to the wide variability in sample types and experimental configurations. The algorithm demonstrates robust and consistent performance across diverse imaging conditions and sample types, delivering quantitative, high-quality reconstructions for both simulated data and experimental datasets acquired at beamline P05 at PETRA III (DESY, Hamburg), operated by Helmholtz-Zentrum Hereon. Furthermore, it enables the simultaneous retrieval of both phase and absorption information.", "date": "2025-08-21", "categories": []}
{"id": "2503.04171", "title": "DuCos: Duality Constrained Depth Super-Resolution via Foundation Model", "abstract": "We introduce DuCos, a novel depth super-resolution framework grounded in Lagrangian duality theory, offering a flexible integration of multiple constraints and reconstruction objectives to enhance accuracy and robustness. Our DuCos is the first to significantly improve generalization across diverse scenarios with foundation models as prompts. The prompt design consists of two key components: Correlative Fusion (CF) and Gradient Regulation (GR). CF facilitates precise geometric alignment and effective fusion between prompt and depth features, while GR refines depth predictions by enforcing consistency with sharp-edged depth maps derived from foundation models. Crucially, these prompts are seamlessly embedded into the Lagrangian constraint term, forming a synergistic and principled framework. Extensive experiments demonstrate that DuCos outperforms existing state-of-the-art methods, achieving superior accuracy, robustness, and generalization.", "date": "2025-03-06", "categories": []}
{"id": "2508.14563", "title": "GOGS: High-Fidelity Geometry and Relighting for Glossy Objects via Gaussian Surfels", "abstract": "Inverse rendering of glossy objects from RGB imagery remains fundamentally limited by inherent ambiguity. Although NeRF-based methods achieve high-fidelity reconstruction via dense-ray sampling, their computational cost is prohibitive. Recent 3D Gaussian Splatting achieves high reconstruction efficiency but exhibits limitations under specular reflections. Multi-view inconsistencies introduce high-frequency surface noise and structural artifacts, while simplified rendering equations obscure material properties, leading to implausible relighting results. To address these issues, we propose GOGS, a novel two-stage framework based on 2D Gaussian surfels. First, we establish robust surface reconstruction through physics-based rendering with split-sum approximation, enhanced by geometric priors from foundation models. Second, we perform material decomposition by leveraging Monte Carlo importance sampling of the full rendering equation, modeling indirect illumination via differentiable 2D Gaussian ray tracing and refining high-frequency specular details through spherical mipmap-based directional encoding that captures anisotropic highlights. Extensive experiments demonstrate state-of-the-art performance in geometry reconstruction, material separation, and photorealistic relighting under novel illuminations, outperforming existing inverse rendering approaches.", "date": "2025-08-20", "categories": []}
{"id": "2508.14730", "title": "Improved Mapping Between Illuminations and Sensors for RAW Images", "abstract": "RAW images are unprocessed camera sensor output with sensor-specific RGB values based on the sensor's color filter spectral sensitivities. RAW images also incur strong color casts due to the sensor's response to the spectral properties of scene illumination. The sensor- and illumination-specific nature of RAW images makes it challenging to capture RAW datasets for deep learning methods, as scenes need to be captured for each sensor and under a wide range of illumination. Methods for illumination augmentation for a given sensor and the ability to map RAW images between sensors are important for reducing the burden of data capture. To explore this problem, we introduce the first-of-its-kind dataset comprising carefully captured scenes under a wide range of illumination. Specifically, we use a customized lightbox with tunable illumination spectra to capture several scenes with different cameras. Our illumination and sensor mapping dataset has 390 illuminations, four cameras, and 18 scenes. Using this dataset, we introduce a lightweight neural network approach for illumination and sensor mapping that outperforms competing methods. We demonstrate the utility of our approach on the downstream task of training a neural ISP. Link to project page: https://github.com/SamsungLabs/illum-sensor-mapping.", "date": "2025-08-20", "categories": []}
{"id": "2404.16647", "title": "Application of RESNET50 Convolution Neural Network for the Extraction of Optical Parameters in Scattering Media", "abstract": "Estimation of the optical properties of scattering media such as tissue is important in diagnostics as well as in the development of techniques to image deeper. As light penetrates the sample scattering events occur that alter the propagation direction of the photons in a random manner leading degradation of image quality. The distribution of the scattered light does, however, give a measure of the optical properties such as the reduced scattering coefficient and the absorption coefficient. Unfortunately, inverting scattering patterns to recover the optical properties is not simple especially in the regime where the light is partially randomized. Machine learning has been proposed by several authors as a means of recovering these properties from either the back scattered or the transmitted light. In the present paper we train a general purpose convolutional neural network RESNET 50 with simulated data based on Monte Carlo simulations. We show that compared with previous work our approach gives comparable or better reconstruction accuracy with training on a much smaller dataset. Moreover, by training on multiple parameters such as the intensity distribution at multiple planes or the exit angle and spatial distribution one achieves improved performance compared to training on a single input such as the intensity distribution captured at the sample surface. While our approach gives good parameter reconstruction, we identify factors that limit accuracy of the recovered properties, particularly the absorption coefficient. In the light of these limitations, we suggest how the present approach may be enhanced for even better performance.", "date": "2024-04-25", "categories": []}
{"id": "2508.13503", "title": "AdaptiveAE: An Adaptive Exposure Strategy for HDR Capturing in Dynamic Scenes", "abstract": "Mainstream high dynamic range imaging techniques typically rely on fusing multiple images captured with different exposure setups (shutter speed and ISO). A good balance between shutter speed and ISO is crucial for achieving high-quality HDR, as high ISO values introduce significant noise, while long shutter speeds can lead to noticeable motion blur. However, existing methods often overlook the complex interaction between shutter speed and ISO and fail to account for motion blur effects in dynamic scenes.\n  In this work, we propose AdaptiveAE, a reinforcement learning-based method that optimizes the selection of shutter speed and ISO combinations to maximize HDR reconstruction quality in dynamic environments. AdaptiveAE integrates an image synthesis pipeline that incorporates motion blur and noise simulation into our training procedure, leveraging semantic information and exposure histograms. It can adaptively select optimal ISO and shutter speed sequences based on a user-defined exposure time budget, and find a better exposure schedule than traditional solutions. Experimental results across multiple datasets demonstrate that it achieves the state-of-the-art performance.", "date": "2025-08-19", "categories": []}
{"id": "2508.13907", "title": "Learning to See Through Flare", "abstract": "Machine vision systems are susceptible to laser flare, where unwanted intense laser illumination blinds and distorts its perception of the environment through oversaturation or permanent damage to sensor pixels. We introduce NeuSee, the first computational imaging framework for high-fidelity sensor protection across the full visible spectrum. It jointly learns a neural representation of a diffractive optical element (DOE) and a frequency-space Mamba-GAN network for image restoration. NeuSee system is adversarially trained end-to-end on 100K unique images to suppress the peak laser irradiance as high as $10^6$ times the sensor saturation threshold $I_{\\textrm{sat}}$, the point at which camera sensors may experience damage without the DOE. Our system leverages heterogeneous data and model parallelism for distributed computing, integrating hyperspectral information and multiple neural networks for realistic simulation and image restoration. NeuSee takes into account open-world scenes with dynamically varying laser wavelengths, intensities, and positions, as well as lens flare effects, unknown ambient lighting conditions, and sensor noises. It outperforms other learned DOEs, achieving full-spectrum imaging and laser suppression for the first time, with a 10.1\\% improvement in restored image quality.", "date": "2025-08-19", "categories": []}
{"id": "2508.11654", "title": "Data-driven RF Tomography via Cross-modal Sensing and Continual Learning", "abstract": "Data-driven radio frequency (RF) tomography has demonstrated significant potential for underground target detection, due to the penetrative nature of RF signals through soil. However, it is still challenging to achieve accurate and robust performance in dynamic environments. In this work, we propose a data-driven radio frequency tomography (DRIFT) framework with the following key components to reconstruct cross section images of underground root tubers, even with significant changes in RF signals. First, we design a cross-modal sensing system with RF and visual sensors, and propose to train an RF tomography deep neural network (DNN) model following the cross-modal learning approach. Then we propose to apply continual learning to automatically update the DNN model, once environment changes are detected in a dynamic environment. Experimental results show that our approach achieves an average equivalent diameter error of 2.29 cm, 23.2% improvement upon the state-of-the-art approach. Our DRIFT code and dataset are publicly available on https://github.com/Data-driven-RTI/DRIFT.", "date": "2025-08-04", "categories": []}
{"id": "2508.12147", "title": "KP-INR: A Dual-Branch Implicit Neural Representation Model for Cardiac Cine MRI Reconstruction", "abstract": "Cardiac Magnetic Resonance (CMR) imaging is a non-invasive method for assessing cardiac structure, function, and blood flow. Cine MRI extends this by capturing heart motion, providing detailed insights into cardiac mechanics. To reduce scan time and breath-hold discomfort, fast acquisition techniques have been utilized at the cost of lowering image quality. Recently, Implicit Neural Representation (INR) methods have shown promise in unsupervised reconstruction by learning coordinate-to-value mappings from undersampled data, enabling high-quality image recovery. However, current existing INR methods primarily focus on using coordinate-based positional embeddings to learn the mapping, while overlooking the feature representations of the target point and its neighboring context. In this work, we propose KP-INR, a dual-branch INR method operating in k-space for cardiac cine MRI reconstruction: one branch processes the positional embedding of k-space coordinates, while the other learns from local multi-scale k-space feature representations at those coordinates. By enabling cross-branch interaction and approximating the target k-space values from both branches, KP-INR can achieve strong performance on challenging Cartesian k-space data. Experiments on the CMRxRecon2024 dataset confirms its improved performance over baseline models and highlights its potential in this field.", "date": "2025-08-16", "categories": []}
{"id": "2404.18394", "title": "Reconstructing Satellites in 3D from Amateur Telescope Images", "abstract": "Monitoring space objects is crucial for space situational awareness, yet reconstructing 3D satellite models from ground-based telescope images is challenging due to atmospheric turbulence, long observation distances, limited viewpoints, and low signal-to-noise ratios. In this paper, we propose a novel computational imaging framework that overcomes these obstacles by integrating a hybrid image pre-processing pipeline with a joint pose estimation and 3D reconstruction module based on controlled Gaussian Splatting (GS) and Branch-and-Bound (BnB) search. We validate our approach on both synthetic satellite datasets and on-sky observations of China's Tiangong Space Station and the International Space Station, achieving robust 3D reconstructions of low-Earth orbit satellites from ground-based data. Quantitative evaluations using SSIM, PSNR, LPIPS, and Chamfer Distance demonstrate that our method outperforms state-of-the-art NeRF-based approaches, and ablation studies confirm the critical role of each component. Our framework enables high-fidelity 3D satellite monitoring from Earth, offering a cost-effective alternative for space situational awareness. Project page: https://ai4scientificimaging.org/ReconstructingSatellites", "date": "2024-04-29", "categories": []}
{"id": "2508.10359", "title": "AtomDiffuser: Time-Aware Degradation Modeling for Drift and Beam Damage in STEM Imaging", "abstract": "Scanning transmission electron microscopy (STEM) plays a critical role in modern materials science, enabling direct imaging of atomic structures and their evolution under external interferences. However, interpreting time-resolved STEM data remains challenging due to two entangled degradation effects: spatial drift caused by mechanical and thermal instabilities, and beam-induced signal loss resulting from radiation damage. These factors distort both geometry and intensity in complex, temporally correlated ways, making it difficult for existing methods to explicitly separate their effects or model material dynamics at atomic resolution. In this work, we present AtomDiffuser, a time-aware degradation modeling framework that disentangles sample drift and radiometric attenuation by predicting an affine transformation and a spatially varying decay map between any two STEM frames. Unlike traditional denoising or registration pipelines, our method leverages degradation as a physically heuristic, temporally conditioned process, enabling interpretable structural evolutions across time. Trained on synthetic degradation processes, AtomDiffuser also generalizes well to real-world cryo-STEM data. It further supports high-resolution degradation inference and drift alignment, offering tools for visualizing and quantifying degradation patterns that correlate with radiation-induced atomic instabilities.", "date": "2025-08-14", "categories": []}
{"id": "2508.10582", "title": "EvTurb: Event Camera Guided Turbulence Removal", "abstract": "Atmospheric turbulence degrades image quality by introducing blur and geometric tilt distortions, posing significant challenges to downstream computer vision tasks. Existing single-image and multi-frame methods struggle with the highly ill-posed nature of this problem due to the compositional complexity of turbulence-induced distortions. To address this, we propose EvTurb, an event guided turbulence removal framework that leverages high-speed event streams to decouple blur and tilt effects. EvTurb decouples blur and tilt effects by modeling event-based turbulence formation, specifically through a novel two-step event-guided network: event integrals are first employed to reduce blur in the coarse outputs. This is followed by employing a variance map, derived from raw event streams, to eliminate the tilt distortion for the refined outputs. Additionally, we present TurbEvent, the first real-captured dataset featuring diverse turbulence scenarios. Experimental results demonstrate that EvTurb surpasses state-of-the-art methods while maintaining computational efficiency.", "date": "2025-08-14", "categories": []}
{"id": "2508.08291", "title": "Probabilistic Emissivity Retrieval from Hyperspectral Data via Physics-Guided Variational Inference", "abstract": "Recent research has proven neural networks to be a powerful tool for performing hyperspectral imaging (HSI) target identification. However, many deep learning frameworks deliver a single material class prediction and operate on a per-pixel basis; such approaches are limited in their interpretability and restricted to predicting materials that are accessible in available training libraries. In this work, we present an inverse modeling approach in the form of a physics-conditioned generative model.A probabilistic latent-variable model learns the underlying distribution of HSI radiance measurements and produces the conditional distribution of the emissivity spectrum. Moreover, estimates of the HSI scene's atmosphere and background are used as a physically relevant conditioning mechanism to contextualize a given radiance measurement during the encoding and decoding processes. Furthermore, we employ an in-the-loop augmentation scheme and physics-based loss criteria to avoid bias towards a predefined training material set and to encourage the model to learn physically consistent inverse mappings. Monte-Carlo sampling of the model's conditioned posterior delivers a sought emissivity distribution and allows for interpretable uncertainty quantification. Moreover, a distribution-based material matching scheme is presented to return a set of likely material matches for an inferred emissivity distribution. Hence, we present a strategy to incorporate contextual information about a given HSI scene, capture the possible variation of underlying material spectra, and provide interpretable probability measures of a candidate material accounting for given remotely-sensed radiance measurement.", "date": "2025-08-05", "categories": []}
{"id": "2508.09528", "title": "Physics-guided Deep Unfolding Network for Enhanced Kronecker Compressive sensing", "abstract": "Deep networks have achieved remarkable success in image compressed sensing (CS) task, namely reconstructing a high-fidelity image from its compressed measurement. However, existing works are deficient inincoherent compressed measurement at sensing phase and implicit measurement representations at reconstruction phase, limiting the overall performance. In this work, we answer two questions: 1) how to improve the measurement incoherence for decreasing the ill-posedness; 2) how to learn informative representations from measurements. To this end, we propose a novel asymmetric Kronecker CS (AKCS) model and theoretically present its better incoherence than previous Kronecker CS with minimal complexity increase. Moreover, we reveal that the unfolding networks' superiority over non-unfolding ones result from sufficient gradient descents, called explicit measurement representations. We propose a measurement-aware cross attention (MACA) mechanism to learn implicit measurement representations. We integrate AKCS and MACA into widely-used unfolding architecture to get a measurement-enhanced unfolding network (MEUNet). Extensive experiences demonstrate that our MEUNet achieves state-of-the-art performance in reconstruction accuracy and inference speed.", "date": "2025-08-13", "categories": []}
{"id": "2310.03546", "title": "Plug-and-Play Posterior Sampling under Mismatched Measurement and Prior Models", "abstract": "Posterior sampling has been shown to be a powerful Bayesian approach for solving imaging inverse problems. The recent plug-and-play unadjusted Langevin algorithm (PnP-ULA) has emerged as a promising method for Monte Carlo sampling and minimum mean squared error (MMSE) estimation by combining physical measurement models with deep-learning priors specified using image denoisers. However, the intricate relationship between the sampling distribution of PnP-ULA and the mismatched data-fidelity and denoiser has not been theoretically analyzed. We address this gap by proposing a posterior-L2 pseudometric and using it to quantify an explicit error bound for PnP-ULA under mismatched posterior distribution. We numerically validate our theory on several inverse problems such as sampling from Gaussian mixture models and image deblurring. Our results suggest that the sensitivity of the sampling distribution of PnP-ULA to a mismatch in the measurement model and the denoiser can be precisely characterized.", "date": "2023-10-05", "categories": []}
{"id": "2508.08309", "title": "Variational volume reconstruction with the Deep Ritz Method", "abstract": "We present a novel approach to variational volume reconstruction from sparse, noisy slice data using the Deep Ritz method. Motivated by biomedical imaging applications such as MRI-based slice-to-volume reconstruction (SVR), our approach addresses three key challenges: (i) the reliance on image segmentation to extract boundaries from noisy grayscale slice images, (ii) the need to reconstruct volumes from a limited number of slice planes, and (iii) the computational expense of traditional mesh-based methods. We formulate a variational objective that combines a regression loss designed to avoid image segmentation by operating on noisy slice data directly with a modified Cahn-Hilliard energy incorporating anisotropic diffusion to regularize the reconstructed geometry. We discretize the phase field with a neural network, approximate the objective at each optimization step with Monte Carlo integration, and use ADAM to find the minimum of the approximated variational objective. While the stochastic integration may not yield the true solution to the variational problem, we demonstrate that our method reliably produces high-quality reconstructed volumes in a matter of seconds, even when the slice data is sparse and noisy.", "date": "2025-08-08", "categories": []}
{"id": "2408.10679", "title": "Alignment-free Raw Video Demoireing", "abstract": "Video demoireing aims to remove undesirable interference patterns that arise during the capture of screen content, restoring artifact-free frames while maintaining temporal consistency. Existing video demoireing methods typically utilize carefully designed alignment modules to estimate inter-frame motion for leveraging temporal information; however, these modules are often complex and computationally demanding. Meanwhile, recent works indicate that using raw data as input significantly enhances demoireing performance. Building on this insight, this paper introduces a novel alignment-free raw video demoireing network with frequency-assisted spatio-temporal Mamba (DemMamba). It incorporates sequentially arranged Spatial Mamba Blocks (SMB) and Temporal Mamba Blocks (TMB) to effectively model the inter- and intra-relationships in raw video demoireing. The SMB employs a multi-directional scanning mechanism coupled with a learnable frequency compressor to effectively differentiate interference patterns across various orientations and frequencies, resulting in reduced artifacts, sharper edges, and faithful texture reconstruction. Concurrently, the TMB enhances temporal consistency by performing bidirectional scanning across the temporal sequences and integrating channel attention techniques, facilitating improved temporal information fusion. Extensive experiments demonstrate that DemMamba surpasses state-of-the-art methods by 1.6 dB in PSNR, and also delivers a satisfactory visual experience.", "date": "2024-08-20", "categories": []}
{"id": "2410.13592", "title": "OAH-Net: A Deep Neural Network for Hologram Reconstruction of Off-axis Digital Holographic Microscope", "abstract": "Off-axis digital holographic microscopy is a high-throughput, label-free imaging technology that provides three-dimensional, high-resolution information about samples, particularly useful in large-scale cellular imaging. However, the hologram reconstruction process poses a significant bottleneck for timely data analysis. To address this challenge, we propose a novel reconstruction approach that integrates deep learning with the physical principles of off-axis holography. We initialized part of the network weights based on the physical principle and then fine-tuned them via weakly supersized learning. Our off-axis hologram network (OAH-Net) retrieves phase and amplitude images with errors that fall within the measurement error range attributable to hardware, and its reconstruction speed significantly surpasses the microscope's acquisition rate. Crucially, OAH-Net demonstrates remarkable external generalization capabilities on unseen samples with distinct patterns and can be seamlessly integrated with other models for downstream tasks to achieve end-to-end real-time hologram analysis. This capability further expands off-axis holography's applications in both biological and medical studies.", "date": "2024-10-17", "categories": []}
{"id": "2508.06703", "title": "Fourier Optics and Deep Learning Methods for Fast 3D Reconstruction in Digital Holography", "abstract": "Computer-generated holography (CGH) is a promising method that modulates user-defined waveforms with digital holograms. An efficient and fast pipeline framework is proposed to synthesize CGH using initial point cloud and MRI data. This input data is reconstructed into volumetric objects that are then input into non-convex Fourier optics optimization algorithms for phase-only hologram (POH) and complex-hologram (CH) generation using alternating projection, SGD, and quasi-Netwton methods. Comparison of reconstruction performance of these algorithms as measured by MSE, RMSE, and PSNR is analyzed as well as to HoloNet deep learning CGH. Performance metrics are shown to be improved by using 2D median filtering to remove artifacts and speckled noise during optimization.", "date": "2025-08-08", "categories": []}
{"id": "2508.06768", "title": "DiffUS: Differentiable Ultrasound Rendering from Volumetric Imaging", "abstract": "Intraoperative ultrasound imaging provides real-time guidance during numerous surgical procedures, but its interpretation is complicated by noise, artifacts, and poor alignment with high-resolution preoperative MRI/CT scans. To bridge the gap between reoperative planning and intraoperative guidance, we present DiffUS, a physics-based, differentiable ultrasound renderer that synthesizes realistic B-mode images from volumetric imaging. DiffUS first converts MRI 3D scans into acoustic impedance volumes using a machine learning approach. Next, we simulate ultrasound beam propagation using ray tracing with coupled reflection-transmission equations. DiffUS formulates wave propagation as a sparse linear system that captures multiple internal reflections. Finally, we reconstruct B-mode images via depth-resolved echo extraction across fan-shaped acquisition geometry, incorporating realistic artifacts including speckle noise and depth-dependent degradation. DiffUS is entirely implemented as differentiable tensor operations in PyTorch, enabling gradient-based optimization for downstream applications such as slice-to-volume registration and volumetric reconstruction. Evaluation on the ReMIND dataset demonstrates DiffUS's ability to generate anatomically accurate ultrasound images from brain MRI data.", "date": "2025-08-09", "categories": []}
{"id": "2508.07214", "title": "Unsupervised Real-World Super-Resolution via Rectified Flow Degradation Modelling", "abstract": "Unsupervised real-world super-resolution (SR) faces critical challenges due to the complex, unknown degradation distributions in practical scenarios. Existing methods struggle to generalize from synthetic low-resolution (LR) and high-resolution (HR) image pairs to real-world data due to a significant domain gap. In this paper, we propose an unsupervised real-world SR method based on rectified flow to effectively capture and model real-world degradation, synthesizing LR-HR training pairs with realistic degradation. Specifically, given unpaired LR and HR images, we propose a novel Rectified Flow Degradation Module (RFDM) that introduces degradation-transformed LR (DT-LR) images as intermediaries. By modeling the degradation trajectory in a continuous and invertible manner, RFDM better captures real-world degradation and enhances the realism of generated LR images. Additionally, we propose a Fourier Prior Guided Degradation Module (FGDM) that leverages structural information embedded in Fourier phase components to ensure more precise modeling of real-world degradation. Finally, the LR images are processed by both FGDM and RFDM, producing final synthetic LR images with real-world degradation. The synthetic LR images are paired with the given HR images to train the off-the-shelf SR networks. Extensive experiments on real-world datasets demonstrate that our method significantly enhances the performance of existing SR approaches in real-world scenarios.", "date": "2025-08-10", "categories": []}
{"id": "2508.07618", "title": "An Iterative Reconstruction Method for Dental Cone-Beam Computed Tomography with a Truncated Field of View", "abstract": "In dental cone-beam computed tomography (CBCT), compact and cost-effective system designs often use small detectors, resulting in a truncated field of view (FOV) that does not fully encompass the patient's head. In iterative reconstruction approaches, the discrepancy between the actual projection and the forward projection within the truncated FOV accumulates over iterations, leading to significant degradation in the reconstructed image quality. In this study, we propose a two-stage approach to mitigate truncation artifacts in dental CBCT. In the first stage, we employ Implicit Neural Representation (INR), leveraging its superior representation power, to generate a prior image over an extended region so that its forward projection fully covers the patient's head. To reduce computational and memory burdens, INR reconstruction is performed with a coarse voxel size. The forward projection of this prior image is then used to estimate the discrepancy due to truncated FOV in the measured projection data. In the second stage, the discrepancy-corrected projection data is utilized in a conventional iterative reconstruction process within the truncated region. Our numerical results demonstrate that the proposed two-grid approach effectively suppresses truncation artifacts, leading to improved CBCT image quality.", "date": "2025-08-11", "categories": []}
{"id": "2508.08183", "title": "THAT: Token-wise High-frequency Augmentation Transformer for Hyperspectral Pansharpening", "abstract": "Transformer-based methods have demonstrated strong potential in hyperspectral pansharpening by modeling long-range dependencies. However, their effectiveness is often limited by redundant token representations and a lack of multi-scale feature modeling. Hyperspectral images exhibit intrinsic spectral priors (e.g., abundance sparsity) and spatial priors (e.g., non-local similarity), which are critical for accurate reconstruction. From a spectral-spatial perspective, Vision Transformers (ViTs) face two major limitations: they struggle to preserve high-frequency components--such as material edges and texture transitions--and suffer from attention dispersion across redundant tokens. These issues stem from the global self-attention mechanism, which tends to dilute high-frequency signals and overlook localized details. To address these challenges, we propose the Token-wise High-frequency Augmentation Transformer (THAT), a novel framework designed to enhance hyperspectral pansharpening through improved high-frequency feature representation and token selection. Specifically, THAT introduces: (1) Pivotal Token Selective Attention (PTSA) to prioritize informative tokens and suppress redundancy; (2) a Multi-level Variance-aware Feed-forward Network (MVFN) to enhance high-frequency detail learning. Experiments on standard benchmarks show that THAT achieves state-of-the-art performance with improved reconstruction quality and efficiency. The source code is available at https://github.com/kailuo93/THAT.", "date": "2025-08-11", "categories": []}
{"id": "2508.05907", "title": "Neural Field Representations of Mobile Computational Photography", "abstract": "Over the past two decades, mobile imaging has experienced a profound transformation, with cell phones rapidly eclipsing all other forms of digital photography in popularity. Today's cell phones are equipped with a diverse range of imaging technologies - laser depth ranging, multi-focal camera arrays, and split-pixel sensors - alongside non-visual sensors such as gyroscopes, accelerometers, and magnetometers. This, combined with on-board integrated chips for image and signal processing, makes the cell phone a versatile pocket-sized computational imaging platform. Parallel to this, we have seen in recent years how neural fields - small neural networks trained to map continuous spatial input coordinates to output signals - enable the reconstruction of complex scenes without explicit data representations such as pixel arrays or point clouds. In this thesis, I demonstrate how carefully designed neural field models can compactly represent complex geometry and lighting effects. Enabling applications such as depth estimation, layer separation, and image stitching directly from collected in-the-wild mobile photography data. These methods outperform state-of-the-art approaches without relying on complex pre-processing steps, labeled ground truth data, or machine learning priors. Instead, they leverage well-constructed, self-regularized models that tackle challenging inverse problems through stochastic gradient descent, fitting directly to raw measurements from a smartphone.", "date": "2025-08-08", "categories": []}
{"id": "2508.05908", "title": "Hybrid Physics-Machine Learning Models for Quantitative Electron Diffraction Refinements", "abstract": "High-fidelity electron microscopy simulations required for quantitative crystal structure refinements face a fundamental challenge: while physical interactions are well-described theoretically, real-world experimental effects are challenging to model analytically. To address this gap, we present a novel hybrid physics-machine learning framework that integrates differentiable physical simulations with neural networks. By leveraging automatic differentiation throughout the simulation pipeline, our method enables gradient-based joint optimization of physical parameters and neural network components representing experimental variables, offering superior scalability compared to traditional second-order methods. We demonstrate this framework through application to three-dimensional electron diffraction (3D-ED) structure refinement, where our approach learns complex thickness distributions directly from diffraction data rather than relying on simplified geometric models. This method achieves state-of-the-art refinement performance across synthetic and experimental datasets, recovering atomic positions, thermal displacements, and thickness profiles with high fidelity. The modular architecture proposed can naturally be extended to accommodate additional physical phenomena and extended to other electron microscopy techniques. This establishes differentiable hybrid modeling as a powerful new paradigm for quantitative electron microscopy, where experimental complexities have historically limited analysis.", "date": "2025-08-08", "categories": []}
{"id": "2403.11350", "title": "Robustness of data-driven approaches in limited angle tomography", "abstract": "The limited angle Radon transform is notoriously difficult to invert due to its ill-posedness. In this work, we give a mathematical explanation that data-driven approaches can stably reconstruct more information compared to traditional methods like filtered backprojection. In addition, we use experiments based on the U-Net neural network to validate our theory.", "date": "2024-03-17", "categories": []}
{"id": "2412.09774", "title": "A Differentiable Wave Optics Model for End-to-End Computational Imaging System Optimization", "abstract": "End-to-end optimization, which simultaneously optimizes optics and algorithms, has emerged as a powerful data-driven method for computational imaging system design. This method achieves joint optimization through backpropagation by incorporating differentiable optics simulators to generate measurements and algorithms to extract information from measurements. However, due to high computational costs, it is challenging to model both aberration and diffraction in light transport for end-to-end optimization of compound optics. Therefore, most existing methods compromise physical accuracy by neglecting wave optics effects or off-axis aberrations, which raises concerns about the robustness of the resulting designs. In this paper, we propose a differentiable optics simulator that efficiently models both aberration and diffraction for compound optics. Using the simulator, we conduct end-to-end optimization on scene reconstruction and classification. Experimental results demonstrate that both lenses and algorithms adopt different configurations depending on whether wave optics is modeled. We also show that systems optimized without wave optics suffer from performance degradation when wave optics effects are introduced during testing. These findings underscore the importance of accurate wave optics modeling in optimizing imaging systems for robust, high-performance applications.", "date": "2024-12-13", "categories": []}
{"id": "2410.21111", "title": "LAMA: Stable Dual-Domain Deep Reconstruction For Sparse-View CT", "abstract": "Inverse problems arise in many applications, especially tomographic imaging. We develop a Learned Alternating Minimization Algorithm (LAMA) to solve such problems via two-block optimization by synergizing data-driven and classical techniques with proven convergence. LAMA is naturally induced by a variational model with learnable regularizers in both data and image domains, parameterized as composite functions of neural networks trained with domain-specific data. We allow these regularizers to be nonconvex and nonsmooth to extract features from data effectively. We minimize the overall objective function using Nesterov's smoothing technique and residual learning architecture. It is demonstrated that LAMA reduces network complexity, improves memory efficiency, and enhances reconstruction accuracy, stability, and interpretability. Extensive experiments show that LAMA significantly outperforms state-of-the-art methods on popular benchmark datasets for Computed Tomography.", "date": "2024-10-28", "categories": []}
{"id": "2508.04051", "title": "Towards Globally Predictable k-Space Interpolation: A White-box Transformer Approach", "abstract": "Interpolating missing data in k-space is essential for accelerating imaging. However, existing methods, including convolutional neural network-based deep learning, primarily exploit local predictability while overlooking the inherent global dependencies in k-space. Recently, Transformers have demonstrated remarkable success in natural language processing and image analysis due to their ability to capture long-range dependencies. This inspires the use of Transformers for k-space interpolation to better exploit its global structure. However, their lack of interpretability raises concerns regarding the reliability of interpolated data. To address this limitation, we propose GPI-WT, a white-box Transformer framework based on Globally Predictable Interpolation (GPI) for k-space. Specifically, we formulate GPI from the perspective of annihilation as a novel k-space structured low-rank (SLR) model. The global annihilation filters in the SLR model are treated as learnable parameters, and the subgradients of the SLR model naturally induce a learnable attention mechanism. By unfolding the subgradient-based optimization algorithm of SLR into a cascaded network, we construct the first white-box Transformer specifically designed for accelerated MRI. Experimental results demonstrate that the proposed method significantly outperforms state-of-the-art approaches in k-space interpolation accuracy while providing superior interpretability.", "date": "2025-08-06", "categories": []}
{"id": "2508.04366", "title": "RotatedMVPS: Multi-view Photometric Stereo with Rotated Natural Light", "abstract": "Multiview photometric stereo (MVPS) seeks to recover high-fidelity surface shapes and reflectances from images captured under varying views and illuminations. However, existing MVPS methods often require controlled darkroom settings for varying illuminations or overlook the recovery of reflectances and illuminations properties, limiting their applicability in natural illumination scenarios and downstream inverse rendering tasks. In this paper, we propose RotatedMVPS to solve shape and reflectance recovery under rotated natural light, achievable with a practical rotation stage. By ensuring light consistency across different camera and object poses, our method reduces the unknowns associated with complex environment light. Furthermore, we integrate data priors from off-the-shelf learning-based single-view photometric stereo methods into our MVPS framework, significantly enhancing the accuracy of shape and reflectance recovery. Experimental results on both synthetic and real-world datasets demonstrate the effectiveness of our approach.", "date": "2025-08-06", "categories": []}
{"id": "2508.04648", "title": "Super Resolved Imaging with Adaptive Optics", "abstract": "Astronomical telescopes suffer from a tradeoff between field of view (FoV) and image resolution: increasing the FoV leads to an optical field that is under-sampled by the science camera. This work presents a novel computational imaging approach to overcome this tradeoff by leveraging the existing adaptive optics (AO) systems in modern ground-based telescopes. Our key idea is to use the AO system's deformable mirror to apply a series of learned, precisely controlled distortions to the optical wavefront, producing a sequence of images that exhibit distinct, high-frequency, sub-pixel shifts. These images can then be jointly upsampled to yield the final super-resolved image. Crucially, we show this can be done while simultaneously maintaining the core AO operation--correcting for the unknown and rapidly changing wavefront distortions caused by Earth's atmosphere. To achieve this, we incorporate end-to-end optimization of both the induced mirror distortions and the upsampling algorithm, such that telescope-specific optics and temporal statistics of atmospheric wavefront distortions are accounted for. Our experimental results with a hardware prototype, as well as simulations, demonstrate significant SNR improvements of up to 12 dB over non-AO super-resolution baselines, using only existing telescope optics and no hardware modifications. Moreover, by using a precise bench-top replica of a complete telescope and AO system, we show that our methodology can be readily transferred to an operational telescope. Project webpage: https://www.cs.toronto.edu/~robin/aosr/", "date": "2025-08-06", "categories": []}
{"id": "2501.14287", "title": "Snapshot multi-spectral imaging through defocusing and a Fourier imager network", "abstract": "Multi-spectral imaging, which simultaneously captures the spatial and spectral information of a scene, is widely used across diverse fields, including remote sensing, biomedical imaging, and agricultural monitoring. Here, we introduce a snapshot multi-spectral imaging approach employing a standard monochrome image sensor with no additional spectral filters or customized components. Our system leverages the inherent chromatic aberration of wavelength-dependent defocusing as a natural source of physical encoding of multi-spectral information; this encoded image information is rapidly decoded via a deep learning-based multi-spectral Fourier Imager Network (mFIN). We experimentally tested our method with six illumination bands and demonstrated an overall accuracy of 92.98% for predicting the illumination channels at the input and achieved a robust multi-spectral image reconstruction on various test objects. This deep learning-powered framework achieves high-quality multi-spectral image reconstruction using snapshot image acquisition with a monochrome image sensor and could be useful for applications in biomedicine, industrial quality control, and agriculture, among others.", "date": "2025-01-24", "categories": []}
{"id": "2508.00539", "title": "Phase-Locked SNR Band Selection for Weak Mineral Signal Detection in Hyperspectral Imagery", "abstract": "Hyperspectral imaging offers detailed spectral information for mineral mapping; however, weak mineral signatures are often masked by noisy and redundant bands, limiting detection performance. To address this, we propose a two-stage integrated framework for enhanced mineral detection in the Cuprite mining district. In the first stage, we compute the signal-to-noise ratio (SNR) for each spectral band and apply a phase-locked thresholding technique to discard low-SNR bands, effectively removing redundancy and suppressing background noise. Savitzky-Golay filtering is then employed for spectral smoothing, serving a dual role first to stabilize trends during band selection, and second to preserve fine-grained spectral features during preprocessing. In the second stage, the refined HSI data is reintroduced into the model, where KMeans clustering is used to extract 12 endmember spectra (W1 custom), followed by non negative least squares (NNLS) for abundance unmixing. The resulting endmembers are quantitatively compared with laboratory spectra (W1 raw) using cosine similarity and RMSE metrics. Experimental results confirm that our proposed pipeline improves unmixing accuracy and enhances the detection of weak mineral zones. This two-pass strategy demonstrates a practical and reproducible solution for spectral dimensionality reduction and unmixing in geological HSI applications.", "date": "2025-08-01", "categories": []}
{"id": "2508.02168", "title": "After the Party: Navigating the Mapping From Color to Ambient Lighting", "abstract": "Illumination in practical scenarios is inherently complex, involving colored light sources, occlusions, and diverse material interactions that produce intricate reflectance and shading effects. However, existing methods often oversimplify this challenge by assuming a single light source or uniform, white-balanced lighting, leaving many of these complexities unaddressed. In this paper, we introduce CL3AN, the first large-scale, high-resolution dataset of its kind designed to facilitate the restoration of images captured under multiple Colored Light sources to their Ambient-Normalized counterparts. Through benchmarking, we find that leading approaches often produce artifacts, such as illumination inconsistencies, texture leakage, and color distortion, primarily due to their limited ability to precisely disentangle illumination from reflectance. Motivated by this insight, we achieve such a desired decomposition through a novel learning framework that leverages explicit chromaticity-luminance components guidance, drawing inspiration from the principles of the Retinex model. Extensive evaluations on existing benchmarks and our dataset demonstrate the effectiveness of our approach, showcasing enhanced robustness under non-homogeneous color lighting and material-specific reflectance variations, all while maintaining a highly competitive computational cost. The benchmark, codes, and models are available at www.github.com/fvasluianu97/RLN2.", "date": "2025-08-04", "categories": []}
{"id": "2508.03403", "title": "Sparsity and Total Variation Constrained Multilayer Linear Unmixing for Hyperspectral Imagery", "abstract": "Hyperspectral unmixing aims at estimating material signatures (known as endmembers) and the corresponding proportions (referred to abundances), which is a critical preprocessing step in various hyperspectral imagery applications. This study develops a novel approach called sparsity and total variation (TV) constrained multilayer linear unmixing (STVMLU) for hyperspectral imagery. Specifically, based on a multilayer matrix factorization model, to improve the accuracy of unmixing, a TV constraint is incorporated to consider adjacent spatial similarity. Additionally, a L1/2-norm sparse constraint is adopted to effectively characterize the sparsity of the abundance matrix. For optimizing the STVMLU model, the method of alternating direction method of multipliers (ADMM) is employed, which allows for the simultaneous extraction of endmembers and their corresponding abundance matrix. Experimental results illustrate the enhanced performance of the proposed STVMLU when compared to other algorithms.", "date": "2025-08-05", "categories": []}
{"id": "2508.03596", "title": "MetaScope: Optics-Driven Neural Network for Ultra-Micro Metalens Endoscopy", "abstract": "Miniaturized endoscopy has advanced accurate visual perception within the human body. Prevailing research remains limited to conventional cameras employing convex lenses, where the physical constraints with millimetre-scale thickness impose serious impediments on the micro-level clinical. Recently, with the emergence of meta-optics, ultra-micro imaging based on metalenses (micron-scale) has garnered great attention, serving as a promising solution. However, due to the physical difference of metalens, there is a large gap in data acquisition and algorithm research. In light of this, we aim to bridge this unexplored gap, advancing the novel metalens endoscopy. First, we establish datasets for metalens endoscopy and conduct preliminary optical simulation, identifying two derived optical issues that physically adhere to strong optical priors. Second, we propose MetaScope, a novel optics-driven neural network tailored for metalens endoscopy driven by physical optics. MetaScope comprises two novel designs: Optics-informed Intensity Adjustment (OIA), rectifying intensity decay by learning optical embeddings, and Optics-informed Chromatic Correction (OCC), mitigating chromatic aberration by learning spatial deformations informed by learned Point Spread Function (PSF) distributions. To enhance joint learning, we further deploy a gradient-guided distillation to transfer knowledge from the foundational model adaptively. Extensive experiments demonstrate that MetaScope not only outperforms state-of-the-art methods in both metalens segmentation and restoration but also achieves impressive generalized ability in real biomedical scenes.", "date": "2025-08-05", "categories": []}
{"id": "2206.09736", "title": "Geo-NI: Geometry-aware Neural Interpolation for Light Field Rendering", "abstract": "In this paper, we present a Geometry-aware Neural Interpolation (Geo-NI) framework for light field rendering. Previous learning-based approaches either rely on the capability of neural networks to perform direct interpolation, which we dubbed Neural Interpolation (NI), or explore scene geometry for novel view synthesis, also known as Depth Image-Based Rendering (DIBR). Instead, we incorporate the ideas behind these two kinds of approaches by launching the NI with a novel DIBR pipeline. Specifically, the proposed Geo-NI first performs NI using input light field sheared by a set of depth hypotheses. Then the DIBR is implemented by assigning the sheared light fields with a novel reconstruction cost volume according to the reconstruction quality under different depth hypotheses. The reconstruction cost is interpreted as a blending weight to render the final output light field by blending the reconstructed light fields along the dimension of depth hypothesis. By combining the superiorities of NI and DIBR, the proposed Geo-NI is able to render views with large disparity with the help of scene geometry while also reconstruct non-Lambertian effect when depth is prone to be ambiguous. Extensive experiments on various datasets demonstrate the superior performance of the proposed geometry-aware light field rendering framework.", "date": "2022-06-20", "categories": []}
{"id": "2407.15708", "title": "SwinSF: Image Reconstruction from Spatial-Temporal Spike Streams", "abstract": "The spike camera, with its high temporal resolution, low latency, and high dynamic range, addresses high-speed imaging challenges like motion blur. It captures photons at each pixel independently, creating binary spike streams rich in temporal information but challenging for image reconstruction. Current algorithms, both traditional and deep learning-based, still need to be improved in the utilization of the rich temporal detail and the restoration of the details of the reconstructed image. To overcome this, we introduce Swin Spikeformer (SwinSF), a novel model for dynamic scene reconstruction from spike streams. SwinSF is composed of Spike Feature Extraction, Spatial-Temporal Feature Extraction, and Final Reconstruction Module. It combines shifted window self-attention and proposed temporal spike attention, ensuring a comprehensive feature extraction that encapsulates both spatial and temporal dynamics, leading to a more robust and accurate reconstruction of spike streams. Furthermore, we build a new synthesized dataset for spike image reconstruction which matches the resolution of the latest spike camera, ensuring its relevance and applicability to the latest developments in spike camera imaging. Experimental results demonstrate that the proposed network SwinSF sets a new benchmark, achieving state-of-the-art performance across a series of datasets, including both real-world and synthesized data across various resolutions. Our codes and proposed dataset will be available soon.", "date": "2024-07-22", "categories": []}
{"id": "2409.10876", "title": "Coordinate-based Speed of Sound Recovery for Aberration-Corrected Photoacoustic Computed Tomography", "abstract": "Photoacoustic computed tomography (PACT) is a non-invasive imaging modality, similar to ultrasound, with wide-ranging medical applications. Conventional PACT images are degraded by wavefront distortion caused by the heterogeneous speed of sound (SOS) in tissue. Accounting for these effects can improve image quality and provide medically useful information, but measuring the SOS directly is burdensome and the existing joint reconstruction method is computationally expensive. Traditional supervised learning techniques are currently inaccessible in this data-starved domain. In this work, we introduce an efficient, self-supervised joint reconstruction method that recovers SOS and high-quality images for ring array PACT systems. To solve this semi-blind inverse problem, we parametrize the SOS using either a pixel grid or a neural field (NF) and update it directly by backpropagating the gradients through a differentiable imaging forward model. Our method removes SOS aberrations more accurately and 35x faster than the current SOTA. We demonstrate the success of our method quantitatively in simulation and qualitatively on experimentally-collected and in vivo data. Our code and synthetic numerical phantoms are available on our project page: https://lukeli0425.github.io/Coord-SoS-PACT/.", "date": "2024-09-17", "categories": []}
{"id": "2505.07839", "title": "Deep Learning Empowered Sub-Diffraction Terahertz Backpropagation Single-Pixel Imaging", "abstract": "Terahertz single-pixel imaging (THz SPI) has garnered widespread attention for its potential to overcome challenges associated with THz focal plane arrays. However, the inherently long wavelength of THz waves limits imaging resolution, while achieving subwavelength resolution requires harsh experimental conditions and time-consuming processes. Here, we propose a sub-diffraction THz backpropagation SPI technique. We illuminate the object with continuous-wave 0.36-THz radiation ({\\lambda}0 = 833.3 {\\mu}m). The transmitted THz wave is modulated by prearranged patterns generated on a 500-{\\mu}m-thick silicon wafer and subsequently recorded by a far-field single-pixel detector. An untrained neural network constrained with the physical SPI process iteratively reconstructs the THz images with an ultralow sampling ratio of 1.5625%, significantly reducing the long sampling times. To further suppress the THz diffraction-field effects, a backpropagation SPI from near field to far field is implemented by integrating with a THz physical propagation model into the output layer of the network. Notably, using the thick wafer where THz evanescent field cannot be fully recorded, we achieve a spatial resolution of 118 {\\mu}m (~{\\lambda}0/7) through backpropagation SPI, thus eliminating the need for ultrathin photomodulators. This approach provides an efficient solution for advancing THz microscopic imaging and addressing other inverse imaging challenges.", "date": "2025-05-05", "categories": []}
{"id": "2506.08163", "title": "SpINRv2: Implicit Neural Representation for Passband FMCW Radars", "abstract": "We present SpINRv2, a neural framework for high-fidelity volumetric reconstruction using Frequency-Modulated Continuous-Wave (FMCW) radar. Extending our prior work (SpINR), this version introduces enhancements that allow accurate learning under high start frequencies-where phase aliasing and sub-bin ambiguity become prominent. Our core contribution is a fully differentiable frequency-domain forward model that captures the complex radar response using closed-form synthesis, paired with an implicit neural representation (INR) for continuous volumetric scene modeling. Unlike time-domain baselines, SpINRv2 directly supervises the complex frequency spectrum, preserving spectral fidelity while drastically reducing computational overhead. Additionally, we introduce sparsity and smoothness regularization to disambiguate sub-bin ambiguities that arise at fine range resolutions. Experimental results show that SpINRv2 significantly outperforms both classical and learning-based baselines, especially under high-frequency regimes, establishing a new benchmark for neural radar-based 3D imaging.", "date": "2025-06-09", "categories": []}
{"id": "2507.23685", "title": "UniLDiff: Unlocking the Power of Diffusion Priors for All-in-One Image Restoration", "abstract": "All-in-One Image Restoration (AiOIR) has emerged as a promising yet challenging research direction. To address the core challenges of diverse degradation modeling and detail preservation, we propose UniLDiff, a unified framework enhanced with degradation- and detail-aware mechanisms, unlocking the power of diffusion priors for robust image restoration. Specifically, we introduce a Degradation-Aware Feature Fusion (DAFF) to dynamically inject low-quality features into each denoising step via decoupled fusion and adaptive modulation, enabling implicit modeling of diverse and compound degradations. Furthermore, we design a Detail-Aware Expert Module (DAEM) in the decoder to enhance texture and fine-structure recovery through expert routing. Extensive experiments across multi-task and mixed degradation settings demonstrate that our method consistently achieves state-of-the-art performance, highlighting the practical potential of diffusion priors for unified image restoration. Our code will be released.", "date": "2025-07-31", "categories": []}
{"id": "2508.01435", "title": "Hyperspectral Image Recovery Constrained by Multi-Granularity Non-Local Self-Similarity Priors", "abstract": "Hyperspectral image (HSI) recovery, as an upstream image processing task,\n  holds significant importance for downstream tasks such as classification,\n  segmentation, and detection. In recent years, HSI recovery methods based on\n  non-local prior representations have demonstrated outstanding performance. However,\n  these methods employ a fixed-format factor to represent the non-local self-similarity\n  tensor groups, making them unable to adapt to diverse missing scenarios. To address\n  this issue, we introduce the concept of granularity in tensor decomposition for the first\n  time and propose an HSI recovery model constrained by multi-granularity non-local\n  self-similarity priors. Specifically, the proposed model alternately performs\n  coarse-grained decomposition and fine-grained decomposition on the non-local\n  self-similarity tensor groups. Among them, the coarse-grained decomposition builds\n  upon Tucker tensor decomposition, which extracts global structural information of the\n  image by performing singular value shrinkage on the mode-unfolded matrices. The\n  fine-grained decomposition employs the FCTN decomposition, capturing local detail\n  information through modeling pairwise correlations among factor tensors. This\n  architectural approach achieves a unified representation of global, local, and non-local\n  priors for HSIs. Experimental results demonstrate that the model has strong\n  applicability and exhibits outstanding recovery effects in various types of missing\n  scenes such as pixels and stripes.", "date": "2025-08-02", "categories": []}
{"id": "2508.02003", "title": "Fast and Memory-efficient Non-line-of-sight Imaging with Quasi-Fresnel Transform", "abstract": "Non-line-of-sight (NLOS) imaging seeks to reconstruct hidden objects by analyzing reflections from intermediary surfaces. Existing methods typically model both the measurement data and the hidden scene in three dimensions, overlooking the inherently two-dimensional nature of most hidden objects. This oversight leads to high computational costs and substantial memory consumption, limiting practical applications and making real-time, high-resolution NLOS imaging on lightweight devices challenging. In this paper, we introduce a novel approach that represents the hidden scene using two-dimensional functions and employs a Quasi-Fresnel transform to establish a direct inversion formula between the measurement data and the hidden scene. This transformation leverages the two-dimensional characteristics of the problem to significantly reduce computational complexity and memory requirements. Our algorithm efficiently performs fast transformations between these two-dimensional aggregated data, enabling rapid reconstruction of hidden objects with minimal memory usage. Compared to existing methods, our approach reduces runtime and memory demands by several orders of magnitude while maintaining imaging quality. The substantial reduction in memory usage not only enhances computational efficiency but also enables NLOS imaging on lightweight devices such as mobile and embedded systems. We anticipate that this method will facilitate real-time, high-resolution NLOS imaging and broaden its applicability across a wider range of platforms.", "date": "2025-08-04", "categories": []}
{"id": "2508.02113", "title": "DeflareMamba: Hierarchical Vision Mamba for Contextually Consistent Lens Flare Removal", "abstract": "Lens flare removal remains an information confusion challenge in the underlying image background and the optical flares, due to the complex optical interactions between light sources and camera lens. While recent solutions have shown promise in decoupling the flare corruption from image, they often fail to maintain contextual consistency, leading to incomplete and inconsistent flare removal. To eliminate this limitation, we propose DeflareMamba, which leverages the efficient sequence modeling capabilities of state space models while maintains the ability to capture local-global dependencies. Particularly, we design a hierarchical framework that establishes long-range pixel correlations through varied stride sampling patterns, and utilize local-enhanced state space models that simultaneously preserves local details. To the best of our knowledge, this is the first work that introduces state space models to the flare removal task. Extensive experiments demonstrate that our method effectively removes various types of flare artifacts, including scattering and reflective flares, while maintaining the natural appearance of non-flare regions. Further downstream applications demonstrate the capacity of our method to improve visual object recognition and cross-modal semantic understanding. Code is available at https://github.com/BNU-ERC-ITEA/DeflareMamba.", "date": "2025-08-04", "categories": []}
{"id": "2407.04230", "title": "A Physical Model-Guided Framework for Underwater Image Enhancement and Depth Estimation", "abstract": "Due to the selective absorption and scattering of light by diverse aquatic media, underwater images usually suffer from various visual degradations. Existing underwater image enhancement (UIE) approaches that combine underwater physical imaging models with neural networks often fail to accurately estimate imaging model parameters such as depth and veiling light, resulting in poor performance in certain scenarios. To address this issue, we propose a physical model-guided framework for jointly training a Deep Degradation Model (DDM) with any advanced UIE model. DDM includes three well-designed sub-networks to accurately estimate various imaging parameters: a veiling light estimation sub-network, a factors estimation sub-network, and a depth estimation sub-network. Based on the estimated parameters and the underwater physical imaging model, we impose physical constraints on the enhancement process by modeling the relationship between underwater images and desired clean images, i.e., outputs of the UIE model. Moreover, while our framework is compatible with any UIE model, we design a simple yet effective fully convolutional UIE model, termed UIEConv. UIEConv utilizes both global and local features for image enhancement through a dual-branch structure. UIEConv trained within our framework achieves remarkable enhancement results across diverse underwater scenes. Furthermore, as a byproduct of UIE, the trained depth estimation sub-network enables accurate underwater scene depth estimation. Extensive experiments conducted in various real underwater imaging scenarios, including deep-sea environments with artificial light sources, validate the effectiveness of our framework and the UIEConv model.", "date": "2024-07-05", "categories": []}
{"id": "2508.00330", "title": "Spectral Sensitivity Estimation with an Uncalibrated Diffraction Grating", "abstract": "This paper introduces a practical and accurate calibration method for camera spectral sensitivity using a diffraction grating. Accurate calibration of camera spectral sensitivity is crucial for various computer vision tasks, including color correction, illumination estimation, and material analysis. Unlike existing approaches that require specialized narrow-band filters or reference targets with known spectral reflectances, our method only requires an uncalibrated diffraction grating sheet, readily available off-the-shelf. By capturing images of the direct illumination and its diffracted pattern through the grating sheet, our method estimates both the camera spectral sensitivity and the diffraction grating parameters in a closed-form manner. Experiments on synthetic and real-world data demonstrate that our method outperforms conventional reference target-based methods, underscoring its effectiveness and practicality.", "date": "2025-08-01", "categories": []}
{"id": "2508.00406", "title": "PMR: Physical Model-Driven Multi-Stage Restoration of Turbulent Dynamic Videos", "abstract": "Geometric distortions and blurring caused by atmospheric turbulence degrade the quality of long-range dynamic scene videos. Existing methods struggle with restoring edge details and eliminating mixed distortions, especially under conditions of strong turbulence and complex dynamics. To address these challenges, we introduce a Dynamic Efficiency Index ($DEI$), which combines turbulence intensity, optical flow, and proportions of dynamic regions to accurately quantify video dynamic intensity under varying turbulence conditions and provide a high-dynamic turbulence training dataset. Additionally, we propose a Physical Model-Driven Multi-Stage Video Restoration ($PMR$) framework that consists of three stages: \\textbf{de-tilting} for geometric stabilization, \\textbf{motion segmentation enhancement} for dynamic region refinement, and \\textbf{de-blurring} for quality restoration. $PMR$ employs lightweight backbones and stage-wise joint training to ensure both efficiency and high restoration quality. Experimental results demonstrate that the proposed method effectively suppresses motion trailing artifacts, restores edge details and exhibits strong generalization capability, especially in real-world scenarios characterized by high-turbulence and complex dynamics. We will make the code and datasets openly available.", "date": "2025-08-01", "categories": []}
{"id": "2404.17484", "title": "Sparse Reconstruction of Optical Doppler Tomography with Alternative State Space Model and Attention", "abstract": "Optical coherence Doppler tomography (ODT) is an emerging blood flow imaging technique. The fundamental unit of ODT is the 1D depth-resolved trace named raw A-scans (or A-line). A 2D ODT image (B-scan) is formed by reconstructing a cross-sectional flow image via Doppler phase-subtraction of raw A-scans along B-line. To obtain a high-fidelity B-scan, densely sampled A-scans are required currently, leading to prolonged scanning time and increased storage demands. Addressing this issue, we propose a novel sparse ODT reconstruction framework with an Alternative State Space Attention Network (ASSAN) that effectively reduces raw A-scans needed. Inspired by the distinct distributions of information along A-line and B-line, ASSAN applies 1D State Space Model (SSM) to each A-line to learn the intra-A-scan representation, while using 1D gated self-attention along B-line to capture the inter-A-scan features. In addition, an effective feedforward network based on sequential 1D convolutions along different axes is employed to enhance the local feature. In validation experiments on real animal data, ASSAN shows clear effectiveness in the reconstruction in comparison with state-of-the-art reconstruction methods.", "date": "2024-04-26", "categories": []}
{"id": "2507.23129", "title": "MRpro - open PyTorch-based MR reconstruction and processing package", "abstract": "We introduce MRpro, an open-source image reconstruction package built upon PyTorch and open data formats. The framework comprises three main areas. First, it provides unified data structures for the consistent manipulation of MR datasets and their associated metadata (e.g., k-space trajectories). Second, it offers a library of composable operators, proximable functionals, and optimization algorithms, including a unified Fourier operator for all common trajectories and an extended phase graph simulation for quantitative MR. These components are used to create ready-to-use implementations of key reconstruction algorithms. Third, for deep learning, MRpro includes essential building blocks such as data consistency layers, differentiable optimization layers, and state-of-the-art backbone networks and integrates public datasets to facilitate reproducibility. MRpro is developed as a collaborative project supported by automated quality control. We demonstrate the versatility of MRpro across multiple applications, including Cartesian, radial, and spiral acquisitions; motion-corrected reconstruction; cardiac MR fingerprinting; learned spatially adaptive regularization weights; model-based learned image reconstruction and quantitative parameter estimation. MRpro offers an extensible framework for MR image reconstruction. With reproducibility and maintainability at its core, it facilitates collaborative development and provides a foundation for future MR imaging research.", "date": "2025-07-30", "categories": []}
{"id": "2507.23162", "title": "Neural Multi-View Self-Calibrated Photometric Stereo without Photometric Stereo Cues", "abstract": "We propose a neural inverse rendering approach that jointly reconstructs geometry, spatially varying reflectance, and lighting conditions from multi-view images captured under varying directional lighting. Unlike prior multi-view photometric stereo methods that require light calibration or intermediate cues such as per-view normal maps, our method jointly optimizes all scene parameters from raw images in a single stage. We represent both geometry and reflectance as neural implicit fields and apply shadow-aware volume rendering. A spatial network first predicts the signed distance and a reflectance latent code for each scene point. A reflectance network then estimates reflectance values conditioned on the latent code and angularly encoded surface normal, view, and light directions. The proposed method outperforms state-of-the-art normal-guided approaches in shape and lighting estimation accuracy, generalizes to view-unaligned multi-light images, and handles objects with challenging geometry and reflectance.", "date": "2025-07-30", "categories": []}
{"id": "2507.22316", "title": "LAMA-Net: A Convergent Network Architecture for Dual-Domain Reconstruction", "abstract": "We propose a learnable variational model that learns the features and leverages complementary information from both image and measurement domains for image reconstruction. In particular, we introduce a learned alternating minimization algorithm (LAMA) from our prior work, which tackles two-block nonconvex and nonsmooth optimization problems by incorporating a residual learning architecture in a proximal alternating framework. In this work, our goal is to provide a complete and rigorous convergence proof of LAMA and show that all accumulation points of a specified subsequence of LAMA must be Clarke stationary points of the problem. LAMA directly yields a highly interpretable neural network architecture called LAMA-Net. Notably, in addition to the results shown in our prior work, we demonstrate that the convergence property of LAMA yields outstanding stability and robustness of LAMA-Net in this work. We also show that the performance of LAMA-Net can be further improved by integrating a properly designed network that generates suitable initials, which we call iLAMA-Net. To evaluate LAMA-Net/iLAMA-Net, we conduct several experiments and compare them with several state-of-the-art methods on popular benchmark datasets for Sparse-View Computed Tomography.", "date": "2025-07-30", "categories": []}
{"id": "2507.22523", "title": "Learned Off-aperture Encoding for Wide Field-of-view RGBD Imaging", "abstract": "End-to-end (E2E) designed imaging systems integrate coded optical designs with decoding algorithms to enhance imaging fidelity for diverse visual tasks. However, existing E2E designs encounter significant challenges in maintaining high image fidelity at wide fields of view, due to high computational complexity, as well as difficulties in modeling off-axis wave propagation while accounting for off-axis aberrations. In particular, the common approach of placing the encoding element into the aperture or pupil plane results in only a global control of the wavefront. To overcome these limitations, this work explores an additional design choice by positioning a DOE off-aperture, enabling a spatial unmixing of the degrees of freedom and providing local control over the wavefront over the image plane. Our approach further leverages hybrid refractive-diffractive optical systems by linking differentiable ray and wave optics modeling, thereby optimizing depth imaging quality and demonstrating system versatility. Experimental results reveal that the off-aperture DOE enhances the imaging quality by over 5 dB in PSNR at a FoV of approximately $45^\\circ$ when paired with a simple thin lens, outperforming traditional on-aperture systems. Furthermore, we successfully recover color and depth information at nearly $28^\\circ$ FoV using off-aperture DOE configurations with compound optics. Physical prototypes for both applications validate the effectiveness and versatility of the proposed method.", "date": "2025-07-30", "categories": []}
{"id": "2403.01142", "title": "Edge-guided Low-light Image Enhancement with Inertial Bregman Alternating Linearized Minimization", "abstract": "Prior-based methods for low-light image enhancement often face challenges in extracting available prior information from dim images. To overcome this limitation, we introduce a simple yet effective Retinex model with the proposed edge extraction prior. More specifically, we design an edge extraction network to capture the fine edge features from the low-light image directly. Building upon the Retinex theory, we decompose the low-light image into its illumination and reflectance components and introduce an edge-guided Retinex model for enhancing low-light images. To solve the proposed model, we propose a novel inertial Bregman alternating linearized minimization algorithm. This algorithm addresses the optimization problem associated with the edge-guided Retinex model, enabling effective enhancement of low-light images. Through rigorous theoretical analysis, we establish the convergence properties of the algorithm. Besides, we prove that the proposed algorithm converges to a stationary point of the problem through nonconvex optimization theory. Furthermore, extensive experiments are conducted on multiple real-world low-light image datasets to demonstrate the efficiency and superiority of the proposed scheme.", "date": "2024-03-02", "categories": []}
{"id": "2507.20575", "title": "Implicit Spatiotemporal Bandwidth Enhancement Filter by Sine-activated Deep Learning Model for Fast 3D Photoacoustic Tomography", "abstract": "3D photoacoustic tomography (3D-PAT) using high-frequency hemispherical transducers offers near-omnidirectional reception and enhanced sensitivity to the finer structural details encoded in the high-frequency components of the broadband photoacoustic (PA) signal. However, practical constraints such as limited number of channels with bandlimited sampling rate often result in sparse and bandlimited sensors that degrade image quality. To address this, we revisit the 2D deep learning (DL) approach applied directly to sensor-wise PA radio-frequency (PARF) data. Specifically, we introduce sine activation into the DL model to restore the broadband nature of PARF signals given the observed band-limited and high-frequency PARF data. Given the scarcity of 3D training data, we employ simplified training strategies by simulating random spherical absorbers. This combination of sine-activated model and randomized training is designed to emphasize bandwidth learning over dataset memorization. Our model was evaluated on a leaf skeleton phantom, a micro-CT-verified 3D spiral phantom and in-vivo human palm vasculature. The results showed that the proposed training mechanism on sine-activated model was well-generalized across the different tests by effectively increasing the sensor density and recovering the spatiotemporal bandwidth. Qualitatively, the sine-activated model uniquely enhanced high-frequency content that produces clearer vascular structure with fewer artefacts. Quantitatively, the sine-activated model exhibits full bandwidth at -12 dB spectrum and significantly higher contrast-to-noise ratio with minimal loss of structural similarity index. Lastly, we optimized our approach to enable fast enhanced 3D-PAT at 2 volumes-per-second for better practical imaging of a free-moving targets.", "date": "2025-07-28", "categories": []}
{"id": "2503.22026", "title": "Multispectral Demosaicing via Dual Cameras", "abstract": "Multispectral (MS) images capture detailed scene information across a wide range of spectral bands, making them invaluable for applications requiring rich spectral data. Integrating MS imaging into multi camera devices, such as smartphones, has the potential to enhance both spectral applications and RGB image quality. A critical step in processing MS data is demosaicing, which reconstructs color information from the mosaic MS images captured by the camera. This paper proposes a method for MS image demosaicing specifically designed for dual-camera setups where both RGB and MS cameras capture the same scene. Our approach leverages co-captured RGB images, which typically have higher spatial fidelity, to guide the demosaicing of lower-fidelity MS images. We introduce the Dual-camera RGB-MS Dataset - a large collection of paired RGB and MS mosaiced images with ground-truth demosaiced outputs - that enables training and evaluation of our method. Experimental results demonstrate that our method achieves state-of-the-art accuracy compared to existing techniques.", "date": "2025-03-27", "categories": []}
{"id": "2507.17800", "title": "Improving Multislice Electron Ptychography with a Generative Prior", "abstract": "Multislice electron ptychography (MEP) is an inverse imaging technique that computationally reconstructs the highest-resolution images of atomic crystal structures from diffraction patterns. Available algorithms often solve this inverse problem iteratively but are both time consuming and produce suboptimal solutions due to their ill-posed nature. We develop MEP-Diffusion, a diffusion model trained on a large database of crystal structures specifically for MEP to augment existing iterative solvers. MEP-Diffusion is easily integrated as a generative prior into existing reconstruction methods via Diffusion Posterior Sampling (DPS). We find that this hybrid approach greatly enhances the quality of the reconstructed 3D volumes, achieving a 90.50% improvement in SSIM over existing methods.", "date": "2025-07-23", "categories": []}
{"id": "2507.18740", "title": "Learned Single-Pixel Fluorescence Microscopy", "abstract": "Single-pixel imaging has emerged as a key technique in fluorescence microscopy, where fast acquisition and reconstruction are crucial. In this context, images are reconstructed from linearly compressed measurements. In practice, total variation minimisation is still used to reconstruct the image from noisy measurements of the inner product between orthogonal sampling pattern vectors and the original image data. However, data can be leveraged to learn the measurement vectors and the reconstruction process, thereby enhancing compression, reconstruction quality, and speed. We train an autoencoder through self-supervision to learn an encoder (or measurement matrix) and a decoder. We then test it on physically acquired multispectral and intensity data. During acquisition, the learned encoder becomes part of the physical device. Our approach can enhance single-pixel imaging in fluorescence microscopy by reducing reconstruction time by two orders of magnitude, achieving superior image quality, and enabling multispectral reconstructions. Ultimately, learned single-pixel fluorescence microscopy could advance diagnosis and biological research, providing multispectral imaging at a fraction of the cost.", "date": "2025-07-24", "categories": []}
{"id": "2407.03605", "title": "Orthogonal Constrained Minimization with Tensor $\\ell_{2,p}$ Regularization for HSI Denoising and Destriping", "abstract": "Hyperspectral images~(HSIs) are often contaminated by a mixture of noise such as Gaussian noise, dead lines, stripes, and so on. In this paper, we propose a multi-scale low-rank tensor regularized $\\ell_{2,p}$ (MLTL2p) approach for HSI denoising and destriping, which consists of an orthogonal constrained minimization model and an iterative algorithm with convergence guarantees. The model of the proposed MLTL2p approach is built based on a new sparsity-enhanced Multi-scale Low-rank Tensor regularization and a tensor $\\ell_{2,p}$ norm with \\(p\\in (0,1)\\). The multi-scale low-rank regularization for HSI denoising utilizes the global and local spectral correlation as well as the spatial nonlocal self-similarity priors of HSIs. The corresponding low-rank constraints are formulated based on independent higher-order singular value decomposition with sparsity enhancement on its core tensor to prompt more low-rankness. The tensor $\\ell_{2,p}$ norm for HSI destriping is extended from the matrix $\\ell_{2,p}$ norm. A proximal block coordinate descent algorithm is proposed in the MLTL2p approach to solve the resulting nonconvex nonsmooth minimization with orthogonal constraints. We show any accumulation point of the sequence generated by the proposed algorithm converges to a first-order stationary point, which is defined using three equalities of substationarity, symmetry, and feasibility for orthogonal constraints. In the numerical experiments, we compare the proposed method with state-of-the-art methods including a deep learning based method, and test the methods on both simulated and real HSI datasets. Our proposed MLTL2p method demonstrates outperformance in terms of metrics such as mean peak signal-to-noise ratio as well as visual quality.", "date": "2024-07-04", "categories": []}
{"id": "2507.17268", "title": "PolarAnything: Diffusion-based Polarimetric Image Synthesis", "abstract": "Polarization images facilitate image enhancement and 3D reconstruction tasks, but the limited accessibility of polarization cameras hinders their broader application. This gap drives the need for synthesizing photorealistic polarization images. The existing polarization simulator Mitsuba relies on a parametric polarization image formation model and requires extensive 3D assets covering shape and PBR materials, preventing it from generating large-scale photorealistic images. To address this problem, we propose PolarAnything, capable of synthesizing polarization images from a single RGB input with both photorealism and physical accuracy, eliminating the dependency on 3D asset collections. Drawing inspiration from the zero-shot performance of pretrained diffusion models, we introduce a diffusion-based generative framework with an effective representation strategy that preserves the fidelity of polarization properties. Experiments show that our model generates high-quality polarization images and supports downstream tasks like shape from polarization.", "date": "2025-07-23", "categories": []}
{"id": "2507.17764", "title": "Diffusion-Assisted Frequency Attention Model for Whole-body Low-field MRI Reconstruction", "abstract": "By integrating the generative strengths of diffusion models with the representation capabilities of frequency-domain attention, DFAM effectively enhances reconstruction performance under low-SNR condi-tions. Experimental results demonstrate that DFAM consistently outperforms both conventional reconstruction algorithms and recent learning-based approaches. These findings highlight the potential of DFAM as a promising solution to advance low-field MRI reconstruction, particularly in resource-constrained or underdeveloped clinical settings.", "date": "2025-07-09", "categories": []}
{"id": "2407.21517", "title": "A Simple Low-bit Quantization Framework for Video Snapshot Compressive Imaging", "abstract": "Video Snapshot Compressive Imaging (SCI) aims to use a low-speed 2D camera to capture high-speed scene as snapshot compressed measurements, followed by a reconstruction algorithm to reconstruct the high-speed video frames. State-of-the-art (SOTA) deep learning-based algorithms have achieved impressive performance, yet with heavy computational workload. Network quantization is a promising way to reduce computational cost. However, a direct low-bit quantization will bring large performance drop. To address this challenge, in this paper, we propose a simple low-bit quantization framework (dubbed Q-SCI) for the end-to-end deep learning-based video SCI reconstruction methods which usually consist of a feature extraction, feature enhancement, and video reconstruction module. Specifically, we first design a high-quality feature extraction module and a precise video reconstruction module to extract and propagate high-quality features in the low-bit quantized model. In addition, to alleviate the information distortion of the Transformer branch in the quantized feature enhancement module, we introduce a shift operation on the query and key distributions to further bridge the performance gap. Comprehensive experimental results manifest that our Q-SCI framework can achieve superior performance, e.g., 4-bit quantized EfficientSCI-S derived by our Q-SCI framework can theoretically accelerate the real-valued EfficientSCI-S by 7.8X with only 2.3% performance gap on the simulation testing datasets. Code is available at https://github.com/mcao92/QuantizedSCI.", "date": "2024-07-31", "categories": []}
{"id": "2504.13682", "title": "AnyTSR: Any-Scale Thermal Super-Resolution for UAV", "abstract": "Thermal imaging can greatly enhance the application of intelligent unmanned aerial vehicles (UAV) in challenging environments. However, the inherent low resolution of thermal sensors leads to insufficient details and blurred boundaries. Super-resolution (SR) offers a promising solution to address this issue, while most existing SR methods are designed for fixed-scale SR. They are computationally expensive and inflexible in practical applications. To address above issues, this work proposes a novel any-scale thermal SR method (AnyTSR) for UAV within a single model. Specifically, a new image encoder is proposed to explicitly assign specific feature code to enable more accurate and flexible representation. Additionally, by effectively embedding coordinate offset information into the local feature ensemble, an innovative any-scale upsampler is proposed to better understand spatial relationships and reduce artifacts. Moreover, a novel dataset (UAV-TSR), covering both land and water scenes, is constructed for thermal SR tasks. Experimental results demonstrate that the proposed method consistently outperforms state-of-the-art methods across all scaling factors as well as generates more accurate and detailed high-resolution images. The code is located at https://github.com/vision4robotics/AnyTSR.", "date": "2025-04-18", "categories": []}
{"id": "2507.03733", "title": "Inverse Synthetic Aperture Fourier Ptychography", "abstract": "Fourier ptychography (FP) is a powerful light-based synthetic aperture imaging technique that allows one to reconstruct a high-resolution, wide field-of-view image by computationally integrating a diverse collection of low-resolution, far-field measurements. Typically, FP measurement diversity is introduced by changing the angle of the illumination or the position of the camera; either approach results in sampling different portions of the target's spatial frequency content, but both approaches introduce substantial costs and complexity to the acquisition process. In this work, we introduce Inverse Synthetic Aperture Fourier Ptychography, a novel approach to FP that foregoes changing the illumination angle or camera position and instead generates measurement diversity through target motion. Critically, we also introduce a novel learning-based method for estimating k-space coordinates from dual plane intensity measurements, thereby enabling synthetic aperture imaging without knowing the rotation of the target. We experimentally validate our method in simulation and on a tabletop optical system.", "date": "2025-07-04", "categories": []}
{"id": "2507.14031", "title": "QuantEIT: Ultra-Lightweight Quantum-Assisted Inference for Chest Electrical Impedance Tomography", "abstract": "Electrical Impedance Tomography (EIT) is a non-invasive, low-cost bedside imaging modality with high temporal resolution, making it suitable for bedside monitoring. However, its inherently ill-posed inverse problem poses significant challenges for accurate image reconstruction. Deep learning (DL)-based approaches have shown promise but often rely on complex network architectures with a large number of parameters, limiting efficiency and scalability. Here, we propose an Ultra-Lightweight Quantum-Assisted Inference (QuantEIT) framework for EIT image reconstruction. QuantEIT leverages a Quantum-Assisted Network (QA-Net), combining parallel 2-qubit quantum circuits to generate expressive latent representations that serve as implicit nonlinear priors, followed by a single linear layer for conductivity reconstruction. This design drastically reduces model complexity and parameter number. Uniquely, QuantEIT operates in an unsupervised, training-data-free manner and represents the first integration of quantum circuits into EIT image reconstruction. Extensive experiments on simulated and real-world 2D and 3D EIT lung imaging data demonstrate that QuantEIT outperforms conventional methods, achieving comparable or superior reconstruction accuracy using only 0.2% of the parameters, with enhanced robustness to noise.", "date": "2025-07-18", "categories": []}
{"id": "2310.11535", "title": "Learning Lens Blur Fields", "abstract": "Optical blur is an inherent property of any lens system and is challenging to model in modern cameras because of their complex optical elements. To tackle this challenge, we introduce a high-dimensional neural representation of blur$-$$\\textit{the lens blur field}$$-$and a practical method for acquiring it. The lens blur field is a multilayer perceptron (MLP) designed to (1) accurately capture variations of the lens 2D point spread function over image plane location, focus setting and, optionally, depth and (2) represent these variations parametrically as a single, sensor-specific function. The representation models the combined effects of defocus, diffraction, aberration, and accounts for sensor features such as pixel color filters and pixel-specific micro-lenses. To learn the real-world blur field of a given device, we formulate a generalized non-blind deconvolution problem that directly optimizes the MLP weights using a small set of focal stacks as the only input. We also provide a first-of-its-kind dataset of 5D blur fields$-$for smartphone cameras, camera bodies equipped with a variety of lenses, etc. Lastly, we show that acquired 5D blur fields are expressive and accurate enough to reveal, for the first time, differences in optical behavior of smartphone devices of the same make and model. Code and data can be found at blur-fields.github.io.", "date": "2023-10-17", "categories": []}
{"id": "2507.12967", "title": "RGB Pre-Training Enhanced Unobservable Feature Latent Diffusion Model for Spectral Reconstruction", "abstract": "Spectral reconstruction (SR) is a crucial problem in image processing that requires reconstructing hyperspectral images (HSIs) from the corresponding RGB images. A key difficulty in SR is estimating the unobservable feature, which encapsulates significant spectral information not captured by RGB imaging sensors. The solution lies in effectively constructing the spectral-spatial joint distribution conditioned on the RGB image to complement the unobservable feature. Since HSIs share a similar spatial structure with the corresponding RGB images, it is rational to capitalize on the rich spatial knowledge in RGB pre-trained models for spectral-spatial joint distribution learning. To this end, we extend the RGB pre-trained latent diffusion model (RGB-LDM) to an unobservable feature LDM (ULDM) for SR. As the RGB-LDM and its corresponding spatial autoencoder (SpaAE) already excel in spatial knowledge, the ULDM can focus on modeling spectral structure. Moreover, separating the unobservable feature from the HSI reduces the redundant spectral information and empowers the ULDM to learn the joint distribution in a compact latent space. Specifically, we propose a two-stage pipeline consisting of spectral structure representation learning and spectral-spatial joint distribution learning to transform the RGB-LDM into the ULDM. In the first stage, a spectral unobservable feature autoencoder (SpeUAE) is trained to extract and compress the unobservable feature into a 3D manifold aligned with RGB space. In the second stage, the spectral and spatial structures are sequentially encoded by the SpeUAE and the SpaAE, respectively. The ULDM is then acquired to model the distribution of the coded unobservable feature with guidance from the corresponding RGB images. Experimental results on SR and downstream relighting tasks demonstrate that our proposed method achieves state-of-the-art performance.", "date": "2025-07-17", "categories": []}
{"id": "2507.11853", "title": "A Spatial-Physics Informed Model for 3D Spiral Sample Scanned by SQUID Microscopy", "abstract": "The development of advanced packaging is essential in the semiconductor manufacturing industry. However, non-destructive testing (NDT) of advanced packaging becomes increasingly challenging due to the depth and complexity of the layers involved. In such a scenario, Magnetic field imaging (MFI) enables the imaging of magnetic fields generated by currents. For MFI to be effective in NDT, the magnetic fields must be converted into current density. This conversion has typically relied solely on a Fast Fourier Transform (FFT) for magnetic field inversion; however, the existing approach does not consider eddy current effects or image misalignment in the test setup. In this paper, we present a spatial-physics informed model (SPIM) designed for a 3D spiral sample scanned using Superconducting QUantum Interference Device (SQUID) microscopy. The SPIM encompasses three key components: i) magnetic image enhancement by aligning all the \"sharp\" wire field signals to mitigate the eddy current effect using both in-phase (I-channel) and quadrature-phase (Q-channel) images; (ii) magnetic image alignment that addresses skew effects caused by any misalignment of the scanning SQUID microscope relative to the wire segments; and (iii) an inversion method for converting magnetic fields to magnetic currents by integrating the Biot-Savart Law with FFT. The results show that the SPIM improves I-channel sharpness by 0.3% and reduces Q-channel sharpness by 25%. Also, we were able to remove rotational and skew misalignments of 0.30 in a real image. Overall, SPIM highlights the potential of combining spatial analysis with physics-driven models in practical applications.", "date": "2025-07-16", "categories": []}
{"id": "2409.07040", "title": "Retinex-RAWMamba: Bridging Demosaicing and Denoising for Low-Light RAW Image Enhancement", "abstract": "Low-light image enhancement, particularly in cross-domain tasks such as mapping from the raw domain to the sRGB domain, remains a significant challenge. Many deep learning-based methods have been developed to address this issue and have shown promising results in recent years. However, single-stage methods, which attempt to unify the complex mapping across both domains, leading to limited denoising performance. In contrast, existing two-stage approaches typically overlook the characteristic of demosaicing within the Image Signal Processing (ISP) pipeline, leading to color distortions under varying lighting conditions, especially in low-light scenarios. To address these issues, we propose a novel Mamba-based method customized for low light RAW images, called RAWMamba, to effectively handle raw images with different CFAs. Furthermore, we introduce a Retinex Decomposition Module (RDM) grounded in Retinex prior, which decouples illumination from reflectance to facilitate more effective denoising and automatic non-linear exposure correction, reducing the effect of manual linear illumination enhancement. By bridging demosaicing and denoising, better enhancement for low light RAW images is achieved. Experimental evaluations conducted on public datasets SID and MCR demonstrate that our proposed RAWMamba achieves state-of-the-art performance on cross-domain mapping. The code is available at https://github.com/Cynicarlos/RetinexRawMamba.", "date": "2024-09-11", "categories": []}
{"id": "2409.16921", "title": "Moner: Motion Correction in Undersampled Radial MRI with Unsupervised Neural Representation", "abstract": "Motion correction (MoCo) in radial MRI is a particularly challenging problem due to the unpredictability of subject movement. Current state-of-the-art (SOTA) MoCo algorithms often rely on extensive high-quality MR images to pre-train neural networks, which constrains the solution space and leads to outstanding image reconstruction results. However, the need for large-scale datasets significantly increases costs and limits model generalization. In this work, we propose Moner, an unsupervised MoCo method that jointly reconstructs artifact-free MR images and estimates accurate motion from undersampled, rigid motion-corrupted k-space data, without requiring any training data. Our core idea is to leverage the continuous prior of implicit neural representation (INR) to constrain this ill-posed inverse problem, facilitating optimal solutions. Specifically, we integrate a quasi-static motion model into the INR, granting its ability to correct subject's motion. To stabilize model optimization, we reformulate radial MRI reconstruction as a back-projection problem using the Fourier-slice theorem. Additionally, we propose a novel coarse-to-fine hash encoding strategy, significantly enhancing MoCo accuracy. Experiments on multiple MRI datasets show our Moner achieves performance comparable to SOTA MoCo techniques on in-domain data, while demonstrating significant improvements on out-of-domain data. The code is available at: https://github.com/iwuqing/Moner", "date": "2024-09-25", "categories": []}
{"id": "2410.20203", "title": "Physics-informed Shadowgraph Network: An End-to-end Density Field Reconstruction Method", "abstract": "This study presents a novel approach for quantificationally reconstructing density fields from shadowgraph images using physics-informed neural networks", "date": "2024-10-26", "categories": []}
{"id": "2505.16658", "title": "Zero-Shot Hyperspectral Pansharpening Using Hysteresis-Based Tuning for Spectral Quality Control", "abstract": "Hyperspectral pansharpening has received much attention in recent years due to technological and methodological advances that open the door to new application scenarios. However, research on this topic is only now gaining momentum. The most popular methods are still borrowed from the more mature field of multispectral pansharpening and often overlook the unique challenges posed by hyperspectral data fusion, such as i) the very large number of bands, ii) the overwhelming noise in selected spectral ranges, iii) the significant spectral mismatch between panchromatic and hyperspectral components, iv) a typically high resolution ratio. Imprecise data modeling especially affects spectral fidelity. Even state-of-the-art methods perform well in certain spectral ranges and much worse in others, failing to ensure consistent quality across all bands, with the risk of generating unreliable results. Here, we propose a hyperspectral pansharpening method that explicitly addresses this problem and ensures uniform spectral quality. To this end, a single lightweight neural network is used, with weights that adapt on the fly to each band. During fine-tuning, the spatial loss is turned on and off to ensure a fast convergence of the spectral loss to the desired level, according to a hysteresis-like dynamic. Furthermore, the spatial loss itself is appropriately redefined to account for nonlinear dependencies between panchromatic and spectral bands. Overall, the proposed method is fully unsupervised, with no prior training on external data, flexible, and low-complexity. Experiments on a recently published benchmarking toolbox show that it ensures excellent sharpening quality, competitive with the state-of-the-art, consistently across all bands. The software code and the full set of results are shared online on https://github.com/giu-guarino/rho-PNN.", "date": "2025-05-22", "categories": []}
{"id": "2507.11293", "title": "3D Magnetic Inverse Routine for Single-Segment Magnetic Field Images", "abstract": "In semiconductor packaging, accurately recovering 3D information is crucial for non-destructive testing (NDT) to localize circuit defects. This paper presents a novel approach called the 3D Magnetic Inverse Routine (3D MIR), which leverages Magnetic Field Images (MFI) to retrieve the parameters for the 3D current flow of a single-segment. The 3D MIR integrates a deep learning (DL)-based Convolutional Neural Network (CNN), spatial-physics-based constraints, and optimization techniques. The method operates in three stages: i) The CNN model processes the MFI data to predict ($\\ell/z_o$), where $\\ell$ is the wire length and $z_o$ is the wire's vertical depth beneath the magnetic sensors and classify segment type ($c$). ii) By leveraging spatial-physics-based constraints, the routine provides initial estimates for the position ($x_o$, $y_o$, $z_o$), length ($\\ell$), current ($I$), and current flow direction (positive or negative) of the current segment. iii) An optimizer then adjusts these five parameters ($x_o$, $y_o$, $z_o$, $\\ell$, $I$) to minimize the difference between the reconstructed MFI and the actual MFI. The results demonstrate that the 3D MIR method accurately recovers 3D information with high precision, setting a new benchmark for magnetic image reconstruction in semiconductor packaging. This method highlights the potential of combining DL and physics-driven optimization in practical applications.", "date": "2025-07-15", "categories": []}
{"id": "2210.14231", "title": "Neural Architecture Search generated Phase Retrieval Net for Real-time Off-axis Quantitative Phase Imaging", "abstract": "In off-axis Quantitative Phase Imaging (QPI), artificial neural networks have been recently applied for phase retrieval with aberration compensation and phase unwrapping. However, the involved neural network architectures are largely unoptimized and inefficient with low inference speed, which hinders the realization of real-time imaging. Here, we propose a Neural Architecture Search (NAS) generated Phase Retrieval Net (NAS-PRNet) for accurate and fast phase retrieval. NAS-PRNet is an encoder-decoder style neural network, automatically found from a large neural network architecture search space through NAS. By modifying the differentiable NAS scheme from SparseMask, we learn the optimized skip connections through gradient descent. Specifically, we implement MobileNet-v2 as the encoder and define a synthesized loss that incorporates phase reconstruction loss and network sparsity loss. NAS-PRNet has achieved high-fidelity phase retrieval by achieving a peak Signal-to-Noise Ratio (PSNR) of 36.7 dB and a Structural SIMilarity (SSIM) of 86.6% as tested on interferograms of biological cells. Notably, NAS-PRNet achieves phase retrieval in only 31 ms, representing 15x speedup over the most recent Mamba-UNet with only a slightly lower phase retrieval accuracy.", "date": "2022-10-25", "categories": []}
{"id": "2312.05357", "title": "Unmixing Optical Signals from Undersampled Volumetric Measurements by Filtering the Pixel Latent Variables", "abstract": "The development of signal unmixing algorithms is essential for leveraging multimodal datasets acquired through a wide array of scientific imaging technologies, including hyperspectral or time-resolved acquisitions. In experimental physics, enhancing the spatio-temporal resolution or expanding the number of detection channels often leads to diminished sampling rate and signal-to-noise ratio, significantly affecting the efficacy of signal unmixing algorithms. We propose Latent Unmixing, a new approach which applies bandpass filters to the latent space of a multidimensional convolutional neural network to disentangle overlapping signal components. It enables better isolation and quantification of individual signal contributions, especially in the context of undersampled distributions. Using multidimensional convolution kernels to process all dimensions simultaneously enhances the network's ability to extract information from adjacent pixels, and time or spectral bins. This approach enables more effective separation of components in cases where individual pixels do not provide clear, well-resolved information. We showcase the method's practical use in experimental physics through two test cases that highlight the versatility of our approach: fluorescence lifetime microscopy and mode decomposition in optical fibers. The latent unmixing method extracts valuable information from complex signals that cannot be resolved by standard methods. It opens up new possibilities in optics and photonics for multichannel separation at an increased sampling rate.", "date": "2023-12-08", "categories": []}
{"id": "2408.17339", "title": "Enhancing Underwater Imaging with 4-D Light Fields: Dataset and Method", "abstract": "In this paper, we delve into the realm of 4-D light fields (LFs) to enhance underwater imaging plagued by light absorption, scattering, and other challenges. Contrasting with conventional 2-D RGB imaging, 4-D LF imaging excels in capturing scenes from multiple perspectives, thereby indirectly embedding geometric information. This intrinsic property is anticipated to effectively address the challenges associated with underwater imaging. By leveraging both explicit and implicit depth cues present in 4-D LF images, we propose a progressive, mutually reinforcing framework for underwater 4-D LF image enhancement and depth estimation. Specifically, our framework explicitly utilizes estimated depth information alongside implicit depth-related dynamic convolutional kernels to modulate output features. The entire framework decomposes this complex task, iteratively optimizing the enhanced image and depth information to progressively achieve optimal enhancement results. More importantly, we construct the first 4-D LF-based underwater image dataset for quantitative evaluation and supervised training of learning-based methods, comprising 75 underwater scenes and 3675 high-resolution 2K pairs. To craft vibrant and varied underwater scenes, we build underwater environments with various objects and adopt several types of degradation. Through extensive experimentation, we showcase the potential and superiority of 4-D LF-based underwater imaging vis-a-vis traditional 2-D RGB-based approaches. Moreover, our method effectively corrects color bias and achieves state-of-the-art performance. The dataset and code will be publicly available at https://github.com/linlos1234/LFUIE.", "date": "2024-08-30", "categories": []}
{"id": "2507.09524", "title": "When Schr\\\"odinger Bridge Meets Real-World Image Dehazing with Unpaired Training", "abstract": "Recent advancements in unpaired dehazing, particularly those using GANs, show promising performance in processing real-world hazy images. However, these methods tend to face limitations due to the generator's limited transport mapping capability, which hinders the full exploitation of their effectiveness in unpaired training paradigms. To address these challenges, we propose DehazeSB, a novel unpaired dehazing framework based on the Schr\\\"odinger Bridge. By leveraging optimal transport (OT) theory, DehazeSB directly bridges the distributions between hazy and clear images. This enables optimal transport mappings from hazy to clear images in fewer steps, thereby generating high-quality results. To ensure the consistency of structural information and details in the restored images, we introduce detail-preserving regularization, which enforces pixel-level alignment between hazy inputs and dehazed outputs. Furthermore, we propose a novel prompt learning to leverage pre-trained CLIP models in distinguishing hazy images and clear ones, by learning a haze-aware vision-language alignment. Extensive experiments on multiple real-world datasets demonstrate our method's superiority. Code: https://github.com/ywxjm/DehazeSB.", "date": "2025-07-13", "categories": []}
{"id": "2504.10012", "title": "EBAD-Gaussian: Event-driven Bundle Adjusted Deblur Gaussian Splatting", "abstract": "While 3D Gaussian Splatting (3D-GS) achieves photorealistic novel view synthesis, its performance degrades with motion blur. In scenarios with rapid motion or low-light conditions, existing RGB-based deblurring methods struggle to model camera pose and radiance changes during exposure, reducing reconstruction accuracy. Event cameras, capturing continuous brightness changes during exposure, can effectively assist in modeling motion blur and improving reconstruction quality. Therefore, we propose Event-driven Bundle Adjusted Deblur Gaussian Splatting (EBAD-Gaussian), which reconstructs sharp 3D Gaussians from event streams and severely blurred images. This method jointly learns the parameters of these Gaussians while recovering camera motion trajectories during exposure time. Specifically, we first construct a blur loss function by synthesizing multiple latent sharp images during the exposure time, minimizing the difference between real and synthesized blurred images. Then we use event stream to supervise the light intensity changes between latent sharp images at any time within the exposure period, supplementing the light intensity dynamic changes lost in RGB images. Furthermore, we optimize the latent sharp images at intermediate exposure times based on the event-based double integral (EDI) prior, applying consistency constraints to enhance the details and texture information of the reconstructed images. Extensive experiments on synthetic and real-world datasets show that EBAD-Gaussian can achieve high-quality 3D scene reconstruction under the condition of blurred images and event stream inputs.", "date": "2025-04-14", "categories": []}
{"id": "2507.07789", "title": "Computationally Efficient Information-Driven Optical Design with Interchanging Optimization", "abstract": "Recent work has demonstrated that imaging systems can be evaluated through the information content of their measurements alone, enabling application-agnostic optical design that avoids computational decoding challenges. Information-Driven Encoder Analysis Learning (IDEAL) was proposed to automate this process through gradient-based optimization. In this work, we study IDEAL across diverse imaging systems and find that it suffers from high memory usage, long runtimes, and a potentially mismatched objective function due to end-to-end differentiability requirements. We introduce IDEAL with Interchanging Optimization (IDEAL-IO), a method that decouples density estimation from optical parameter optimization by alternating between fitting models to current measurements and updating optical parameters using fixed models for information estimation. This approach reduces runtime and memory usage by up to 6x while enabling more expressive density models that guide optimization toward superior designs. We validate our method on diffractive optics, lensless imaging, and snapshot 3D microscopy applications, establishing information-theoretic optimization as a practical, scalable strategy for real-world imaging system design.", "date": "2025-07-10", "categories": []}
{"id": "2412.04639", "title": "Multi-dynamic deep image prior for cardiac MRI", "abstract": "Cardiovascular magnetic resonance imaging is a powerful diagnostic tool for assessing cardiac structure and function. However, traditional breath-held imaging protocols pose challenges for patients with arrhythmias or limited breath-holding capacity. This work aims to overcome these limitations by developing a reconstruction framework that enables high-quality imaging in free-breathing conditions for various dynamic cardiac MRI protocols. Multi-Dynamic Deep Image Prior (M-DIP), a novel unsupervised reconstruction framework for accelerated real-time cardiac MRI, is introduced. To capture contrast or content variation, M-DIP first employs a spatial dictionary to synthesize a time-dependent intermediate image. Then, this intermediate image is further refined using time-dependent deformation fields that model cardiac and respiratory motion. Unlike prior DIP-based methods, M-DIP simultaneously captures physiological motion and frame-to-frame content variations, making it applicable to a wide range of dynamic applications. We validate M-DIP using simulated MRXCAT cine phantom data as well as free-breathing real-time cine, single-shot late gadolinium enhancement (LGE), and first-pass perfusion data from clinical patients. Comparative analyses against state-of-the-art supervised and unsupervised approaches demonstrate M-DIP's performance and versatility. M-DIP achieved better image quality metrics on phantom data, higher reader scores on in-vivo cine and LGE data, and comparable scores on in-vivo perfusion data relative to another DIP-based approach. M-DIP enables high-quality reconstructions of real-time free-breathing cardiac MRI without requiring external training data. Its ability to model physiological motion and content variations makes it a promising approach for various dynamic imaging applications.", "date": "2024-12-05", "categories": []}
{"id": "2507.07707", "title": "Compressive Imaging Reconstruction via Tensor Decomposed Multi-Resolution Grid Encoding", "abstract": "Compressive imaging (CI) reconstruction, such as snapshot compressive imaging (SCI) and compressive sensing magnetic resonance imaging (MRI), aims to recover high-dimensional images from low-dimensional compressed measurements. This process critically relies on learning an accurate representation of the underlying high-dimensional image. However, existing unsupervised representations may struggle to achieve a desired balance between representation ability and efficiency. To overcome this limitation, we propose Tensor Decomposed multi-resolution Grid encoding (GridTD), an unsupervised continuous representation framework for CI reconstruction. GridTD optimizes a lightweight neural network and the input tensor decomposition model whose parameters are learned via multi-resolution hash grid encoding. It inherently enjoys the hierarchical modeling ability of multi-resolution grid encoding and the compactness of tensor decomposition, enabling effective and efficient reconstruction of high-dimensional images. Theoretical analyses for the algorithm's Lipschitz property, generalization error bound, and fixed-point convergence reveal the intrinsic superiority of GridTD as compared with existing continuous representation models. Extensive experiments across diverse CI tasks, including video SCI, spectral SCI, and compressive dynamic MRI reconstruction, consistently demonstrate the superiority of GridTD over existing methods, positioning GridTD as a versatile and state-of-the-art CI reconstruction method.", "date": "2025-07-10", "categories": []}
{"id": "2507.06684", "title": "Photometric Stereo using Gaussian Splatting and inverse rendering", "abstract": "Recent state-of-the-art algorithms in photometric stereo rely on neural networks and operate either through prior learning or inverse rendering optimization. Here, we revisit the problem of calibrated photometric stereo by leveraging recent advances in 3D inverse rendering using the Gaussian Splatting formalism. This allows us to parameterize the 3D scene to be reconstructed and optimize it in a more interpretable manner. Our approach incorporates a simplified model for light representation and demonstrates the potential of the Gaussian Splatting rendering engine for the photometric stereo problem.", "date": "2025-07-09", "categories": []}
{"id": "2307.08106", "title": "Polarization Multi-Image Synthesis with Birefringent Metasurfaces", "abstract": "Optical metasurfaces composed of precisely engineered nanostructures have gained significant attention for their ability to manipulate light and implement distinct functionalities based on the properties of the incident field. Computational imaging systems have started harnessing this capability to produce sets of coded measurements that benefit certain tasks when paired with digital post-processing. Inspired by these works, we introduce a new system that uses a birefringent metasurface with a polarizer-mosaicked photosensor to capture four optically-coded measurements in a single exposure. We apply this system to the task of incoherent opto-electronic filtering, where digital spatial-filtering operations are replaced by simpler, per-pixel sums across the four polarization channels, independent of the spatial filter size. In contrast to previous work on incoherent opto-electronic filtering that can realize only one spatial filter, our approach can realize a continuous family of filters from a single capture, with filters being selected from the family by adjusting the post-capture digital summation weights. To find a metasurface that can realize a set of user-specified spatial filters, we introduce a form of gradient descent with a novel regularizer that encourages light efficiency and a high signal-to-noise ratio. We demonstrate several examples in simulation and with fabricated prototypes, including some with spatial filters that have prescribed variations with respect to depth and wavelength.\n  Visit the Project Page at https://deanhazineh.github.io/publications/Multi_Image_Synthesis/MIS_Home.html", "date": "2023-07-16", "categories": []}
{"id": "2504.11512", "title": "Learned enclosure method for experimental EIT data", "abstract": "Electrical impedance tomography (EIT) is a non-invasive imaging method with diverse applications, including medical imaging and non-destructive testing. The inverse problem of reconstructing internal electrical conductivity from boundary measurements is nonlinear and highly ill-posed, making it difficult to solve accurately. In recent years, there has been growing interest in combining analytical methods with machine learning to solve inverse problems. In this paper, we propose a method for estimating the convex hull of inclusions from boundary measurements by combining the enclosure method proposed by Ikehata with neural networks. We demonstrate its performance using experimental data. Compared to the classical enclosure method with least squares fitting, the learned convex hull achieves superior performance on both simulated and experimental data.", "date": "2025-04-15", "categories": []}
{"id": "2507.03369", "title": "Adaptive Gate-Aware Mamba Networks for Magnetic Resonance Fingerprinting", "abstract": "Magnetic Resonance Fingerprinting (MRF) enables fast quantitative imaging by matching signal evolutions to a predefined dictionary. However, conventional dictionary matching suffers from exponential growth in computational cost and memory usage as the number of parameters increases, limiting its scalability to multi-parametric mapping. To address this, recent work has explored deep learning-based approaches as alternatives to DM. We propose GAST-Mamba, an end-to-end framework that combines a dual Mamba-based encoder with a Gate-Aware Spatial-Temporal (GAST) processor. Built on structured state-space models, our architecture efficiently captures long-range spatial dependencies with linear complexity. On 5 times accelerated simulated MRF data (200 frames), GAST-Mamba achieved a T1 PSNR of 33.12~dB, outperforming SCQ (31.69~dB). For T2 mapping, it reached a PSNR of 30.62~dB and SSIM of 0.9124. In vivo experiments further demonstrated improved anatomical detail and reduced artifacts. Ablation studies confirmed that each component contributes to performance, with the GAST module being particularly important under strong undersampling. These results demonstrate the effectiveness of GAST-Mamba for accurate and robust reconstruction from highly undersampled MRF acquisitions, offering a scalable alternative to traditional DM-based methods.", "date": "2025-07-04", "categories": []}
{"id": "2507.03893", "title": "Hierarchical Semantic-Visual Fusion of Visible and Near-infrared Images for Long-range Haze Removal", "abstract": "While image dehazing has advanced substantially in the past decade, most efforts have focused on short-range scenarios, leaving long-range haze removal under-explored. As distance increases, intensified scattering leads to severe haze and signal loss, making it impractical to recover distant details solely from visible images. Near-infrared, with superior fog penetration, offers critical complementary cues through multimodal fusion. However, existing methods focus on content integration while often neglecting haze embedded in visible images, leading to results with residual haze. In this work, we argue that the infrared and visible modalities not only provide complementary low-level visual features, but also share high-level semantic consistency. Motivated by this, we propose a Hierarchical Semantic-Visual Fusion (HSVF) framework, comprising a semantic stream to reconstruct haze-free scenes and a visual stream to incorporate structural details from the near-infrared modality. The semantic stream first acquires haze-robust semantic prediction by aligning modality-invariant intrinsic representations. Then the shared semantics act as strong priors to restore clear and high-contrast distant scenes under severe haze degradation. In parallel, the visual stream focuses on recovering lost structural details from near-infrared by fusing complementary cues from both visible and near-infrared images. Through the cooperation of dual streams, HSVF produces results that exhibit both high-contrast scenes and rich texture details. Moreover, we introduce a novel pixel-aligned visible-infrared haze dataset with semantic labels to facilitate benchmarking. Extensive experiments demonstrate the superiority of our method over state-of-the-art approaches in real-world long-range haze removal.", "date": "2025-07-05", "categories": []}
{"id": "2507.04118", "title": "PromptSR: Cascade Prompting for Lightweight Image Super-Resolution", "abstract": "Although the lightweight Vision Transformer has significantly advanced image super-resolution (SR), it faces the inherent challenge of a limited receptive field due to the window-based self-attention modeling. The quadratic computational complexity relative to window size restricts its ability to use a large window size for expanding the receptive field while maintaining low computational costs. To address this challenge, we propose PromptSR, a novel prompt-empowered lightweight image SR method. The core component is the proposed cascade prompting block (CPB), which enhances global information access and local refinement via three cascaded prompting layers: a global anchor prompting layer (GAPL) and two local prompting layers (LPLs). The GAPL leverages downscaled features as anchors to construct low-dimensional anchor prompts (APs) through cross-scale attention, significantly reducing computational costs. These APs, with enhanced global perception, are then used to provide global prompts, efficiently facilitating long-range token connections. The two LPLs subsequently combine category-based self-attention and window-based self-attention to refine the representation in a coarse-to-fine manner. They leverage attention maps from the GAPL as additional global prompts, enabling them to perceive features globally at different granularities for adaptive local refinement. In this way, the proposed CPB effectively combines global priors and local details, significantly enlarging the receptive field while maintaining the low computational costs of our PromptSR. The experimental results demonstrate the superiority of our method, which outperforms state-of-the-art lightweight SR methods in quantitative, qualitative, and complexity evaluations. Our code will be released at https://github.com/wenyang001/PromptSR.", "date": "2025-07-05", "categories": []}
{"id": "2507.04190", "title": "Towards Spatially-Varying Gain and Binning", "abstract": "Pixels in image sensors have progressively become smaller, driven by the goal of producing higher-resolution imagery. However, ceteris paribus, a smaller pixel accumulates less light, making image quality worse. This interplay of resolution, noise, and the dynamic range of the sensor and their impact on the eventual quality of acquired imagery is a fundamental concept in photography. In this paper, we propose spatially-varying gain and binning to enhance the noise performance and dynamic range of image sensors. First, we show that by varying gain spatially to local scene brightness, the read noise can be made negligible, and the dynamic range of a sensor is expanded by an order of magnitude. Second, we propose a simple analysis to find a binning size that best balances resolution and noise for a given light level; this analysis predicts a spatially-varying binning strategy, again based on local scene brightness, to effectively increase the overall signal-to-noise ratio. % without sacrificing resolution. We discuss analog and digital binning modes and, perhaps surprisingly, show that digital binning outperforms its analog counterparts when a larger gain is allowed. Finally, we demonstrate that combining spatially-varying gain and binning in various applications, including high dynamic range imaging, vignetting, and lens distortion.", "date": "2025-07-05", "categories": []}
{"id": "2507.04617", "title": "Comprehensive Modeling of Camera Spectral and Color Behavior", "abstract": "The spectral response of a digital camera defines the mapping between scene radiance and pixel intensity. Despite its critical importance, there is currently no comprehensive model that considers the end-to-end interaction between light input and pixel intensity output. This paper introduces a novel technique to model the spectral response of an RGB digital camera, addressing this gap. Such models are indispensable for applications requiring accurate color and spectral data interpretation. The proposed model is tested across diverse imaging scenarios by varying illumination conditions and is validated against experimental data. Results demonstrate its effectiveness in improving color fidelity and spectral accuracy, with significant implications for applications in machine vision, remote sensing, and spectral imaging. This approach offers a powerful tool for optimizing camera systems in scientific, industrial, and creative domains where spectral precision is paramount.", "date": "2025-07-07", "categories": []}
{"id": "2507.04622", "title": "A Deep Unfolding Framework for Diffractive Snapshot Spectral Imaging", "abstract": "Snapshot hyperspectral imaging systems acquire spectral data cubes through compressed sensing. Recently, diffractive snapshot spectral imaging (DSSI) methods have attracted significant attention. While various optical designs and improvements continue to emerge, research on reconstruction algorithms remains limited. Although numerous networks and deep unfolding methods have been applied on similar tasks, they are not fully compatible with DSSI systems because of their distinct optical encoding mechanism. In this paper, we propose an efficient deep unfolding framework for diffractive systems, termed diffractive deep unfolding (DDU). Specifically, we derive an analytical solution for the data fidelity term in DSSI, ensuring both the efficiency and the effectiveness during the iterative reconstruction process. Given the severely ill-posed nature of the problem, we employ a network-based initialization strategy rather than non-learning-based methods or linear layers, leading to enhanced stability and performance. Our framework demonstrates strong compatibility with existing state-of-the-art (SOTA) models, which effectively address the initialization and prior subproblem. Extensive experiments validate the superiority of the proposed DDU framework, showcasing improved performance while maintaining comparable parameter counts and computational complexity. These results suggest that DDU provides a solid foundation for future unfolding-based methods in DSSI.", "date": "2025-07-07", "categories": []}
{"id": "2507.04684", "title": "SPIDER: Structure-Preferential Implicit Deep Network for Biplanar X-ray Reconstruction", "abstract": "Biplanar X-ray imaging is widely used in health screening, postoperative rehabilitation evaluation of orthopedic diseases, and injury surgery due to its rapid acquisition, low radiation dose, and straightforward setup. However, 3D volume reconstruction from only two orthogonal projections represents a profoundly ill-posed inverse problem, owing to the intrinsic lack of depth information and irreducible ambiguities in soft-tissue visualization. Some existing methods can reconstruct skeletal structures and Computed Tomography (CT) volumes, they often yield incomplete bone geometry, imprecise tissue boundaries, and a lack of anatomical realism, thereby limiting their clinical utility in scenarios such as surgical planning and postoperative assessment. In this study, we introduce SPIDER, a novel supervised framework designed to reconstruct CT volumes from biplanar X-ray images. SPIDER incorporates tissue structure as prior (e.g., anatomical segmentation) into an implicit neural representation decoder in the form of joint supervision through a unified encoder-decoder architecture. This design enables the model to jointly learn image intensities and anatomical structures in a pixel-aligned fashion. To address the challenges posed by sparse input and structural ambiguity, SPIDER directly embeds anatomical constraints into the reconstruction process, thereby enhancing structural continuity and reducing soft-tissue artifacts. We conduct comprehensive experiments on clinical head CT datasets and show that SPIDER generates anatomically accurate reconstructions from only two projections. Furthermore, our approach demonstrates strong potential in downstream segmentation tasks, underscoring its utility in personalized treatment planning and image-guided surgical navigation.", "date": "2025-07-07", "categories": []}
{"id": "2409.16200", "title": "Upper-body free-breathing Magnetic Resonance Fingerprinting applied to the quantification of water T1 and fat fraction", "abstract": "Over the past decade, Magnetic Resonance Fingerprinting (MRF) has emerged as an efficient paradigm for the rapid and simultaneous quantification of multiple MRI parameters, including fat fraction (FF), water T1 ($T1_{H2O}$), water T2 ($T2_{H2O}$), and fat T1 ($T1_{fat}$). These parameters serve as promising imaging biomarkers in various anatomical targets such as the heart, liver, and skeletal muscles. However, measuring these parameters in the upper body poses challenges due to physiological motion, particularly respiratory motion. In this work, we propose a novel approach, motion-corrected (MoCo) MRF T1-FF, which estimates the motion field using an optimized preliminary motion scan and uses it to correct the MRF acquisition data before dictionary search for reconstructing motion-corrected FF and $T1_{H2O}$ parametric maps of the upper-body region. We validated this framework using an $\\textit{in vivo}$ dataset comprising ten healthy volunteers and a 10-year-old boy with Duchenne muscular dystrophy. At the ROI level, in regions minimally affected by motion, no significant bias was observed between the uncorrected and MoCo reconstructions for FF (mean difference of -0.7%) and $T1_{H2O}$ (-4.9 ms) values. Moreover, MoCo MRF T1-FF significantly reduced the standard deviations of distributions assessed in these regions, indicating improved precision. Notably, in regions heavily affected by motion, such as respiratory muscles, liver, and kidneys, the MRF parametric maps exhibited a marked reduction in motion blurring and streaking artifacts after motion correction. Furthermore, the diaphragm was consistently discernible on parametric maps after motion correction. This approach lays the groundwork for the joint 3D quantification of FF and $T1_{H2O}$ in regions that are rarely studied, such as the respiratory muscles, particularly the intercostal muscles and diaphragm.", "date": "2024-09-24", "categories": []}
{"id": "2505.22128", "title": "Real-Time Blind Defocus Deblurring for Earth Observation: The IMAGIN-e Mission Approach", "abstract": "This work addresses mechanical defocus in Earth observation images from the IMAGIN-e mission aboard the ISS, proposing a blind deblurring approach adapted to space-based edge computing constraints. Leveraging Sentinel-2 data, our method estimates the defocus kernel and trains a restoration model within a GAN framework, effectively operating without reference images.\n  On Sentinel-2 images with synthetic degradation, SSIM improved by 72.47% and PSNR by 25.00%, confirming the model's ability to recover lost details when the original clean image is known. On IMAGIN-e, where no reference images exist, perceptual quality metrics indicate a substantial enhancement, with NIQE improving by 60.66% and BRISQUE by 48.38%, validating real-world onboard restoration. The approach is currently deployed aboard the IMAGIN-e mission, demonstrating its practical application in an operational space environment.\n  By efficiently handling high-resolution images under edge computing constraints, the method enables applications such as water body segmentation and contour detection while maintaining processing viability despite resource limitations.", "date": "2025-05-28", "categories": []}
{"id": "2507.00482", "title": "Physics-Aware Style Transfer for Adaptive Holographic Reconstruction", "abstract": "Inline holographic imaging presents an ill-posed inverse problem of reconstructing objects' complex amplitude from recorded diffraction patterns. Although recent deep learning approaches have shown promise over classical phase retrieval algorithms, they often require high-quality ground truth datasets of complex amplitude maps to achieve a statistical inverse mapping operation between the two domains. Here, we present a physics-aware style transfer approach that interprets the object-to-sensor distance as an implicit style within diffraction patterns. Using the style domain as the intermediate domain to construct cyclic image translation, we show that the inverse mapping operation can be learned in an adaptive manner only with datasets composed of intensity measurements. We further demonstrate its biomedical applicability by reconstructing the morphology of dynamically flowing red blood cells, highlighting its potential for real-time, label-free imaging. As a framework that leverages physical cues inherently embedded in measurements, the presented method offers a practical learning strategy for imaging applications where ground truth is difficult or impossible to obtain.", "date": "2025-07-01", "categories": []}
{"id": "2507.00501", "title": "Laplace-Mamba: Laplace Frequency Prior-Guided Mamba-CNN Fusion Network for Image Dehazing", "abstract": "Recent progress in image restoration has underscored Spatial State Models (SSMs) as powerful tools for modeling long-range dependencies, owing to their appealing linear complexity and computational efficiency. However, SSM-based approaches exhibit limitations in reconstructing localized structures and tend to be less effective when handling high-dimensional data, frequently resulting in suboptimal recovery of fine image features. To tackle these challenges, we introduce Laplace-Mamba, a novel framework that integrates Laplace frequency prior with a hybrid Mamba-CNN architecture for efficient image dehazing. Leveraging the Laplace decomposition, the image is disentangled into low-frequency components capturing global texture and high-frequency components representing edges and fine details. This decomposition enables specialized processing via dual parallel pathways: the low-frequency branch employs SSMs for global context modeling, while the high-frequency branch utilizes CNNs to refine local structural details, effectively addressing diverse haze scenarios. Notably, the Laplace transformation facilitates information-preserving downsampling of low-frequency components in accordance with the Nyquist theory, thereby significantly improving computational efficiency. Extensive evaluations across multiple benchmarks demonstrate that our method outperforms state-of-the-art approaches in both restoration quality and efficiency. The source code and pretrained models are available at https://github.com/yz-wang/Laplace-Mamba.", "date": "2025-07-01", "categories": []}
{"id": "2502.20532", "title": "Finer Disentanglement of Aleatoric Uncertainty Can Accelerate Chemical Histopathology Imaging", "abstract": "Label-free chemical imaging holds significant promise for improving digital pathology workflows, but data acquisition speed remains a limiting factor. To address this gap, we propose an adaptive strategy-initially scan the low information (LI) content of the entire tissue quickly, identify regions with high aleatoric uncertainty (AU), and selectively re-image them at better quality to capture higher information (HI) details. The primary challenge lies in distinguishing between high-AU regions mitigable through HI imaging and those that are not. However, since existing uncertainty frameworks cannot separate such AU subcategories, we propose a fine-grained disentanglement method based on post-hoc latent space analysis to unmix resolvable from irresolvable high-AU regions. We apply our approach to streamline infrared spectroscopic imaging of breast tissues, achieving superior downstream segmentation performance. This marks the first study focused on fine-grained AU disentanglement within dynamic image spaces (LI-to-HI), with novel application to streamline histopathology.", "date": "2025-02-27", "categories": []}
{"id": "2506.22753", "title": "Degradation-Modeled Multipath Diffusion for Tunable Metalens Photography", "abstract": "Metalenses offer significant potential for ultra-compact computational imaging but face challenges from complex optical degradation and computational restoration difficulties. Existing methods typically rely on precise optical calibration or massive paired datasets, which are non-trivial for real-world imaging systems. Furthermore, a lack of control over the inference process often results in undesirable hallucinated artifacts. We introduce Degradation-Modeled Multipath Diffusion for tunable metalens photography, leveraging powerful natural image priors from pretrained models instead of large datasets. Our framework uses positive, neutral, and negative-prompt paths to balance high-frequency detail generation, structural fidelity, and suppression of metalens-specific degradation, alongside \\textit{pseudo} data augmentation. A tunable decoder enables controlled trade-offs between fidelity and perceptual quality. Additionally, a spatially varying degradation-aware attention (SVDA) module adaptively models complex optical and sensor-induced degradation. Finally, we design and build a millimeter-scale MetaCamera for real-world validation. Extensive results show that our approach outperforms state-of-the-art methods, achieving high-fidelity and sharp image reconstruction. More materials: https://dmdiff.github.io/.", "date": "2025-06-28", "categories": []}
{"id": "2506.23466", "title": "FD-DiT: Frequency Domain-Directed Diffusion Transformer for Low-Dose CT Reconstruction", "abstract": "Low-dose computed tomography (LDCT) reduces radiation exposure but suffers from image artifacts and loss of detail due to quantum and electronic noise, potentially impacting diagnostic accuracy. Transformer combined with diffusion models has been a promising approach for image generation. Nevertheless, existing methods exhibit limitations in preserving finegrained image details. To address this issue, frequency domain-directed diffusion transformer (FD-DiT) is proposed for LDCT reconstruction. FD-DiT centers on a diffusion strategy that progressively introduces noise until the distribution statistically aligns with that of LDCT data, followed by denoising processing. Furthermore, we employ a frequency decoupling technique to concentrate noise primarily in high-frequency domain, thereby facilitating effective capture of essential anatomical structures and fine details. A hybrid denoising network is then utilized to optimize the overall data reconstruction process. To enhance the capability in recognizing high-frequency noise, we incorporate sliding sparse local attention to leverage the sparsity and locality of shallow-layer information, propagating them via skip connections for improving feature representation. Finally, we propose a learnable dynamic fusion strategy for optimal component integration. Experimental results demonstrate that at identical dose levels, LDCT images reconstructed by FD-DiT exhibit superior noise and artifact suppression compared to state-of-the-art methods.", "date": "2025-06-30", "categories": []}
{"id": "2506.23542", "title": "Consistent Time-of-Flight Depth Denoising via Graph-Informed Geometric Attention", "abstract": "Depth images captured by Time-of-Flight (ToF) sensors are prone to noise, requiring denoising for reliable downstream applications. Previous works either focus on single-frame processing, or perform multi-frame processing without considering depth variations at corresponding pixels across frames, leading to undesirable temporal inconsistency and spatial ambiguity. In this paper, we propose a novel ToF depth denoising network leveraging motion-invariant graph fusion to simultaneously enhance temporal stability and spatial sharpness. Specifically, despite depth shifts across frames, graph structures exhibit temporal self-similarity, enabling cross-frame geometric attention for graph fusion. Then, by incorporating an image smoothness prior on the fused graph and data fidelity term derived from ToF noise distribution, we formulate a maximum a posterior problem for ToF denoising. Finally, the solution is unrolled into iterative filters whose weights are adaptively learned from the graph-informed geometric attention, producing a high-performance yet interpretable network. Experimental results demonstrate that the proposed scheme achieves state-of-the-art performance in terms of accuracy and consistency on synthetic DVToF dataset and exhibits robust generalization on the real Kinectv2 dataset. Source code will be released at \\href{https://github.com/davidweidawang/GIGA-ToF}{https://github.com/davidweidawang/GIGA-ToF}.", "date": "2025-06-30", "categories": []}
{"id": "2506.21680", "title": "PhotonSplat: 3D Scene Reconstruction and Colorization from SPAD Sensors", "abstract": "Advances in 3D reconstruction using neural rendering have enabled high-quality 3D capture. However, they often fail when the input imagery is corrupted by motion blur, due to fast motion of the camera or the objects in the scene. This work advances neural rendering techniques in such scenarios by using single-photon avalanche diode (SPAD) arrays, an emerging sensing technology capable of sensing images at extremely high speeds. However, the use of SPADs presents its own set of unique challenges in the form of binary images, that are driven by stochastic photon arrivals. To address this, we introduce PhotonSplat, a framework designed to reconstruct 3D scenes directly from SPAD binary images, effectively navigating the noise vs. blur trade-off. Our approach incorporates a novel 3D spatial filtering technique to reduce noise in the renderings. The framework also supports both no-reference using generative priors and reference-based colorization from a single blurry image, enabling downstream applications such as segmentation, object detection and appearance editing tasks. Additionally, we extend our method to incorporate dynamic scene representations, making it suitable for scenes with moving objects. We further contribute PhotonScenes, a real-world multi-view dataset captured with the SPAD sensors.", "date": "2025-06-26", "categories": []}
{"id": "2506.22246", "title": "EAMamba: Efficient All-Around Vision State Space Model for Image Restoration", "abstract": "Image restoration is a key task in low-level computer vision that aims to reconstruct high-quality images from degraded inputs. The emergence of Vision Mamba, which draws inspiration from the advanced state space model Mamba, marks a significant advancement in this field. Vision Mamba demonstrates excellence in modeling long-range dependencies with linear complexity, a crucial advantage for image restoration tasks. Despite its strengths, Vision Mamba encounters challenges in low-level vision tasks, including computational complexity that scales with the number of scanning sequences and local pixel forgetting. To address these limitations, this study introduces Efficient All-Around Mamba (EAMamba), an enhanced framework that incorporates a Multi-Head Selective Scan Module (MHSSM) with an all-around scanning mechanism. MHSSM efficiently aggregates multiple scanning sequences, which avoids increases in computational complexity and parameter count. The all-around scanning strategy implements multiple patterns to capture holistic information and resolves the local pixel forgetting issue. Our experimental evaluations validate these innovations across several restoration tasks, including super resolution, denoising, deblurring, and dehazing. The results validate that EAMamba achieves a significant 31-89% reduction in FLOPs while maintaining favorable performance compared to existing low-level Vision Mamba methods.", "date": "2025-06-27", "categories": []}
{"id": "2506.20877", "title": "THIRDEYE: Cue-Aware Monocular Depth Estimation via Brain-Inspired Multi-Stage Fusion", "abstract": "Monocular depth estimation methods traditionally train deep models to infer depth directly from RGB pixels. This implicit learning often overlooks explicit monocular cues that the human visual system relies on, such as occlusion boundaries, shading, and perspective. Rather than expecting a network to discover these cues unaided, we present ThirdEye, a cue-aware pipeline that deliberately supplies each cue through specialised, pre-trained, and frozen networks. These cues are fused in a three-stage cortical hierarchy (V1->V2->V3) equipped with a key-value working-memory module that weights them by reliability. An adaptive-bins transformer head then produces a high-resolution disparity map. Because the cue experts are frozen, ThirdEye inherits large amounts of external supervision while requiring only modest fine-tuning. This extended version provides additional architectural detail, neuroscientific motivation, and an expanded experimental protocol; quantitative results will appear in a future revision.", "date": "2025-06-25", "categories": []}
{"id": "2506.19845", "title": "A Comparative Study of NAFNet Baselines for Image Restoration", "abstract": "We study NAFNet (Nonlinear Activation Free Network), a simple and efficient deep learning baseline for image restoration. By using CIFAR10 images corrupted with noise and blur, we conduct an ablation study of NAFNet's core components. Our baseline model implements SimpleGate activation, Simplified Channel Activation (SCA), and LayerNormalization. We compare this baseline to different variants that replace or remove components. Quantitative results (PSNR, SSIM) and examples illustrate how each modification affects restoration performance. Our findings support the NAFNet design: the SimpleGate and simplified attention mechanisms yield better results than conventional activations and attention, while LayerNorm proves to be important for stable training. We conclude with recommendations for model design, discuss potential improvements, and future work.", "date": "2025-06-24", "categories": []}
{"id": "2410.13295", "title": "PiLocNet: Physics-informed neural network on 3D localization with rotating point spread function", "abstract": "For the 3D localization problem using point spread function (PSF) engineering, we propose a novel enhancement of our previously introduced localization neural network, LocNet. The improved network is a physics-informed neural network (PINN) that we call PiLocNet. Previous works on the localization problem may be categorized separately into model-based optimization and neural network approaches. Our PiLocNet combines the unique strengths of both approaches by incorporating forward-model-based information into the network via a data-fitting loss term that constrains the neural network to yield results that are physically sensible. We additionally incorporate certain regularization terms from the variational method, which further improves the robustness of the network in the presence of image noise, as we show for the Poisson and Gaussian noise models. This framework accords interpretability to the neural network, and the results we obtain show its superiority. Although the paper focuses on the use of single-lobe rotating PSF to encode the full 3D source location, we expect the method to be widely applicable to other PSFs and imaging problems that are constrained by known forward processes.", "date": "2024-10-17", "categories": []}
{"id": "2412.10338", "title": "XYScanNet: A State Space Model for Single Image Deblurring", "abstract": "Deep state-space models (SSMs), like recent Mamba architectures, are emerging as a promising alternative to CNN and Transformer networks. Existing Mamba-based restoration methods process visual data by leveraging a flatten-and-scan strategy that converts image patches into a 1D sequence before scanning. However, this scanning paradigm ignores local pixel dependencies and introduces spatial misalignment by positioning distant pixels incorrectly adjacent, which reduces local noise-awareness and degrades image sharpness in low-level vision tasks. To overcome these issues, we propose a novel slice-and-scan strategy that alternates scanning along intra- and inter-slices. We further design a new Vision State Space Module (VSSM) for image deblurring, and tackle the inefficiency challenges of the current Mamba-based vision module. Building upon this, we develop XYScanNet, an SSM architecture integrated with a lightweight feature fusion module for enhanced image deblurring. XYScanNet, maintains competitive distortion metrics and significantly improves perceptual performance. Experimental results show that XYScanNet enhances KID by $17\\%$ compared to the nearest competitor.", "date": "2024-12-13", "categories": []}
{"id": "2506.18279", "title": "Optimal spectral initializers impact on phase retrieval phase transitions -- an RDT view", "abstract": "We analyze the relation between spectral initializers and theoretical limits of \\emph{descending} phase retrieval algorithms (dPR). In companion paper [104], for any sample complexity ratio, $\\alpha$, \\emph{parametric manifold}, ${\\mathcal {PM}}(\\alpha)$, is recognized as a critically important structure that generically determines dPRs abilities to solve phase retrieval (PR). Moreover, overlap between the algorithmic solution and the true signal is positioned as a key ${\\mathcal {PM}}$'s component. We here consider the so-called \\emph{overlap optimal} spectral initializers (OptSpins) as dPR's starting points and develop a generic \\emph{Random duality theory} (RDT) based program to statistically characterize them. In particular, we determine the functional structure of OptSpins and evaluate the starting overlaps that they provide for the dPRs. Since ${\\mathcal {PM}}$'s so-called \\emph{flat regions} are highly susceptible to \\emph{local jitteriness} and as such are key obstacles on dPR's path towards PR's global optimum, a precise characterization of the starting overlap allows to determine if such regions can be successfully circumvented. Through the presented theoretical analysis we observe two key points in that regard: \\textbf{\\emph{(i)}} dPR's theoretical phase transition (critical $\\alpha$ above which they solve PR) might be difficult to practically achieve as the ${\\mathcal {PM}}$'s flat regions are large causing the associated OptSpins to fall exactly within them; and \\textbf{\\emph{(ii)}} Opting for so-called ``\\emph{safer compression}'' and slightly increasing $\\alpha$ (by say $15\\%$) shrinks flat regions and allows OptSpins to fall outside them and dPRs to ultimately solve PR. Numerical simulations are conducted as well and shown to be in an excellent agreement with theoretical predictions.", "date": "2025-06-23", "categories": []}
{"id": "2404.11511", "title": "Event Cameras Meet SPADs for High-Speed, Low-Bandwidth Imaging", "abstract": "Traditional cameras face a trade-off between low-light performance and high-speed imaging: longer exposure times to capture sufficient light results in motion blur, whereas shorter exposures result in Poisson-corrupted noisy images. While burst photography techniques help mitigate this tradeoff, conventional cameras are fundamentally limited in their sensor noise characteristics. Event cameras and single-photon avalanche diode (SPAD) sensors have emerged as promising alternatives to conventional cameras due to their desirable properties. SPADs are capable of single-photon sensitivity with microsecond temporal resolution, and event cameras can measure brightness changes up to 1 MHz with low bandwidth requirements. We show that these properties are complementary, and can help achieve low-light, high-speed image reconstruction with low bandwidth requirements. We introduce a sensor fusion framework to combine SPADs with event cameras to improves the reconstruction of high-speed, low-light scenes while reducing the high bandwidth cost associated with using every SPAD frame. Our evaluation, on both synthetic and real sensor data, demonstrates significant enhancements ( > 5 dB PSNR) in reconstructing low-light scenes at high temporal resolution (100 kHz) compared to conventional cameras. Event-SPAD fusion shows great promise for real-world applications, such as robotics or medical imaging.", "date": "2024-04-17", "categories": []}
{"id": "2506.16803", "title": "Temperature calibration of surface emissivities with an improved thermal image enhancement network", "abstract": "Infrared thermography faces persistent challenges in temperature accuracy due to material emissivity variations, where existing methods often neglect the joint optimization of radiometric calibration and image degradation. This study introduces a physically guided neural framework that unifies temperature correction and image enhancement through a symmetric skip-CNN architecture and an emissivity-aware attention module. The pre-processing stage segments the ROIs of the image and and initially corrected the firing rate. A novel dual-constrained loss function strengthens the statistical consistency between the target and reference regions through mean-variance alignment and histogram matching based on Kullback-Leibler dispersion. The method works by dynamically fusing thermal radiation features and spatial context, and the model suppresses emissivity artifacts while recovering structural details. After validating the industrial blower system under different conditions, the improved network realizes the dynamic fusion of thermal radiation characteristics and spatial background, with accurate calibration results in various industrial conditions.", "date": "2025-06-20", "categories": []}
{"id": "2506.14995", "title": "Improved Image Reconstruction and Diffusion Parameter Estimation Using a Temporal Convolutional Network Model of Gradient Trajectory Errors", "abstract": "Summary: Errors in gradient trajectories introduce significant artifacts and distortions in magnetic resonance images, particularly in non-Cartesian imaging sequences, where imperfect gradient waveforms can greatly reduce image quality. Purpose: Our objective is to develop a general, nonlinear gradient system model that can accurately predict gradient distortions using convolutional networks. Methods: A set of training gradient waveforms were measured on a small animal imaging system, and used to train a temporal convolutional network to predict the gradient waveforms produced by the imaging system. Results: The trained network was able to accurately predict nonlinear distortions produced by the gradient system. Network prediction of gradient waveforms was incorporated into the image reconstruction pipeline and provided improvements in image quality and diffusion parameter mapping compared to both the nominal gradient waveform and the gradient impulse response function. Conclusion: Temporal convolutional networks can more accurately model gradient system behavior than existing linear methods and may be used to retrospectively correct gradient errors.", "date": "2025-06-17", "categories": []}
{"id": "2506.15346", "title": "Acoustic Waveform Inversion with Image-to-Image Schr\\\"odinger Bridges", "abstract": "Recent developments in application of deep learning models to acoustic Full Waveform Inversion (FWI) are marked by the use of diffusion models as prior distributions for Bayesian-like inference procedures. The advantage of these methods is the ability to generate high-resolution samples, which are otherwise unattainable with classical inversion methods or other deep learning-based solutions. However, the iterative and stochastic nature of sampling from diffusion models along with heuristic nature of output control remain limiting factors for their applicability. For instance, an optimal way to include the approximate velocity model into diffusion-based inversion scheme remains unclear, even though it is considered an essential part of FWI pipeline. We address the issue by employing a Schr\\\"odinger Bridge that interpolates between the distributions of ground truth and smoothed velocity models. To facilitate the learning of nonlinear drifts that transfer samples between distributions we extend the concept of Image-to-Image Schr\\\"odinger Bridge ($\\text{I}^2\\text{SB}$) to conditional sampling, resulting in a conditional Image-to-Image Schr\\\"odinger Bridge (c$\\text{I}^2\\text{SB}$) framework. To validate our method, we assess its effectiveness in reconstructing the reference velocity model from its smoothed approximation, coupled with the observed seismic signal of fixed shape. Our experiments demonstrate that the proposed solution outperforms our reimplementation of conditional diffusion model suggested in earlier works, while requiring only a few neural function evaluations (NFEs) to achieve sample fidelity superior to that attained with supervised learning-based approach. The supplementary code implementing the algorithms described in this paper can be found in the repository https://github.com/stankevich-mipt/seismic_inversion_via_I2SB.", "date": "2025-06-18", "categories": []}
{"id": "2506.14605", "title": "Unsupervised Imaging Inverse Problems with Diffusion Distribution Matching", "abstract": "This work addresses image restoration tasks through the lens of inverse problems using unpaired datasets. In contrast to traditional approaches -- which typically assume full knowledge of the forward model or access to paired degraded and ground-truth images -- the proposed method operates under minimal assumptions and relies only on small, unpaired datasets. This makes it particularly well-suited for real-world scenarios, where the forward model is often unknown or misspecified, and collecting paired data is costly or infeasible. The method leverages conditional flow matching to model the distribution of degraded observations, while simultaneously learning the forward model via a distribution-matching loss that arises naturally from the framework. Empirically, it outperforms both single-image blind and unsupervised approaches on deblurring and non-uniform point spread function (PSF) calibration tasks. It also matches state-of-the-art performance on blind super-resolution. We also showcase the effectiveness of our method with a proof of concept for lens calibration: a real-world application traditionally requiring time-consuming experiments and specialized equipment. In contrast, our approach achieves this with minimal data acquisition effort.", "date": "2025-06-17", "categories": []}
{"id": "2506.14719", "title": "Plug-and-Play with 2.5D Artifact Reduction Prior for Fast and Accurate Industrial Computed Tomography Reconstruction", "abstract": "Cone-beam X-ray computed tomography (XCT) is an essential imaging technique for generating 3D reconstructions of internal structures, with applications ranging from medical to industrial imaging. Producing high-quality reconstructions typically requires many X-ray measurements; this process can be slow and expensive, especially for dense materials. Recent work incorporating artifact reduction priors within a plug-and-play (PnP) reconstruction framework has shown promising results in improving image quality from sparse-view XCT scans while enhancing the generalizability of deep learning-based solutions. However, this method uses a 2D convolutional neural network (CNN) for artifact reduction, which captures only slice-independent information from the 3D reconstruction, limiting performance. In this paper, we propose a PnP reconstruction method that uses a 2.5D artifact reduction CNN as the prior. This approach leverages inter-slice information from adjacent slices, capturing richer spatial context while remaining computationally efficient. We show that this 2.5D prior not only improves the quality of reconstructions but also enables the model to directly suppress commonly occurring XCT artifacts (such as beam hardening), eliminating the need for artifact correction pre-processing. Experiments on both experimental and synthetic cone-beam XCT data demonstrate that the proposed method better preserves fine structural details, such as pore size and shape, leading to more accurate defect detection compared to 2D priors. In particular, we demonstrate strong performance on experimental XCT data using a 2.5D artifact reduction prior trained entirely on simulated scans, highlighting the proposed method's ability to generalize across domains.", "date": "2025-06-17", "categories": []}
{"id": "2206.01397", "title": "Dynamic Structured Illumination Microscopy with a Neural Space-time Model", "abstract": "Structured illumination microscopy (SIM) reconstructs a super-resolved image from multiple raw images captured with different illumination patterns; hence, acquisition speed is limited, making it unsuitable for dynamic scenes. We propose a new method, Speckle Flow SIM, that uses static patterned illumination with moving samples and models the sample motion during data capture in order to reconstruct the dynamic scene with super-resolution. Speckle Flow SIM relies on sample motion to capture a sequence of raw images. The spatio-temporal relationship of the dynamic scene is modeled using a neural space-time model with coordinate-based multi-layer perceptrons (MLPs), and the motion dynamics and the super-resolved scene are jointly recovered. We validate Speckle Flow SIM for coherent imaging in simulation and build a simple, inexpensive experimental setup with off-the-shelf components. We demonstrate that Speckle Flow SIM can reconstruct a dynamic scene with deformable motion and 1.88x the diffraction-limited resolution in experiment.", "date": "2022-06-03", "categories": []}
{"id": "2403.07786", "title": "Physics-informed generative real-time lens-free imaging", "abstract": "Advancements in high-throughput biomedical applications require real-time, large field-of-view (FOV) imaging. While current 2D lens-free imaging (LFI) systems improve FOV, they are often hindered by time-consuming multi-position measurements, extensive data pre-processing, and strict optical parameterization, limiting their application to static, thin samples. To overcome these limitations, we introduce GenLFI, combining a generative unsupervised physics-informed neural network (PINN) with a large FOV LFI setup for straightforward holographic image reconstruction, without multi-measurement. GenLFI enables real-time 2D imaging for 3D samples, such as droplet-based microfluidics and 3D cell models, in dynamic complex optical fields. Unlike previous methods, our approach decouples the reconstruction algorithm from optical setup parameters, enabling a large FOV limited only by hardware. We demonstrate a real-time FOV exceeding 550 mm$^2$, over 20 times larger than current real-time LFI systems. This framework unlocks the potential of LFI systems, providing a robust tool for advancing automated high-throughput biomedical applications.", "date": "2024-03-12", "categories": []}
{"id": "2501.18500", "title": "HSRMamba: Contextual Spatial-Spectral State Space Model for Single Image Hyperspectral Super-Resolution", "abstract": "Mamba has demonstrated exceptional performance in visual tasks due to its powerful global modeling capabilities and linear computational complexity, offering considerable potential in hyperspectral image super-resolution (HSISR). However, in HSISR, Mamba faces challenges as transforming images into 1D sequences neglects the spatial-spectral structural relationships between locally adjacent pixels, and its performance is highly sensitive to input order, which affects the restoration of both spatial and spectral details. In this paper, we propose HSRMamba, a contextual spatial-spectral modeling state space model for HSISR, to address these issues both locally and globally. Specifically, a local spatial-spectral partitioning mechanism is designed to establish patch-wise causal relationships among adjacent pixels in 3D features, mitigating the local forgetting issue. Furthermore, a global spectral reordering strategy based on spectral similarity is employed to enhance the causal representation of similar pixels across both spatial and spectral dimensions. Finally, experimental results demonstrate our HSRMamba outperforms the state-of-the-art methods in quantitative quality and visual results. Code is available at: https://github.com/Tomchenshi/HSRMamba.", "date": "2025-01-30", "categories": []}
{"id": "2506.12471", "title": "Adaptive Multi-resolution Hash-Encoding Framework for INR-based Dental CBCT Reconstruction with Truncated FOV", "abstract": "Implicit neural representation (INR), particularly in combination with hash encoding, has recently emerged as a promising approach for computed tomography (CT) image reconstruction. However, directly applying INR techniques to 3D dental cone-beam CT (CBCT) with a truncated field of view (FOV) is challenging. During the training process, if the FOV does not fully encompass the patient's head, a discrepancy arises between the measured projections and the forward projections computed within the truncated domain. This mismatch leads the network to estimate attenuation values inaccurately, producing severe artifacts in the reconstructed images. In this study, we propose a computationally efficient INR-based reconstruction framework that leverages multi-resolution hash encoding for 3D dental CBCT with a truncated FOV. To mitigate truncation artifacts, we train the network over an expanded reconstruction domain that fully encompasses the patient's head. For computational efficiency, we adopt an adaptive training strategy that uses a multi-resolution grid: finer resolution levels and denser sampling inside the truncated FOV, and coarser resolution levels with sparser sampling outside. To maintain consistent input dimensionality of the network across spatially varying resolutions, we introduce an adaptive hash encoder that selectively activates the lower-level features of the hash hierarchy for points outside the truncated FOV. The proposed method with an extended FOV effectively mitigates truncation artifacts. Compared with a naive domain extension using fixed resolution levels and a fixed sampling rate, the adaptive strategy reduces computational time by over 60% for an image volume of 800x800x600, while preserving the PSNR within the truncated FOV.", "date": "2025-06-14", "categories": []}
{"id": "2506.12824", "title": "Learning Unpaired Image Dehazing with Physics-based Rehazy Generation", "abstract": "Overfitting to synthetic training pairs remains a critical challenge in image dehazing, leading to poor generalization capability to real-world scenarios. To address this issue, existing approaches utilize unpaired realistic data for training, employing CycleGAN or contrastive learning frameworks. Despite their progress, these methods often suffer from training instability, resulting in limited dehazing performance. In this paper, we propose a novel training strategy for unpaired image dehazing, termed Rehazy, to improve both dehazing performance and training stability. This strategy explores the consistency of the underlying clean images across hazy images and utilizes hazy-rehazy pairs for effective learning of real haze characteristics. To favorably construct hazy-rehazy pairs, we develop a physics-based rehazy generation pipeline, which is theoretically validated to reliably produce high-quality rehazy images. Additionally, leveraging the rehazy strategy, we introduce a dual-branch framework for dehazing network training, where a clean branch provides a basic dehazing capability in a synthetic manner, and a hazy branch enhances the generalization ability with hazy-rehazy pairs. Moreover, we design a new dehazing network within these branches to improve the efficiency, which progressively restores clean scenes from coarse to fine. Extensive experiments on four benchmarks demonstrate the superior performance of our approach, exceeding the previous state-of-the-art methods by 3.58 dB on the SOTS-Indoor dataset and by 1.85 dB on the SOTS-Outdoor dataset in PSNR. Our code will be publicly available.", "date": "2025-06-15", "categories": []}
{"id": "2506.13391", "title": "Zero-Shot Solving of Imaging Inverse Problems via Noise-Refined Likelihood Guided Diffusion Models", "abstract": "Diffusion models have achieved remarkable success in imaging inverse problems owing to their powerful generative capabilities. However, existing approaches typically rely on models trained for specific degradation types, limiting their generalizability to various degradation scenarios. To address this limitation, we propose a zero-shot framework capable of handling various imaging inverse problems without model retraining. We introduce a likelihood-guided noise refinement mechanism that derives a closed-form approximation of the likelihood score, simplifying score estimation and avoiding expensive gradient computations. This estimated score is subsequently utilized to refine the model-predicted noise, thereby better aligning the restoration process with the generative framework of diffusion models. In addition, we integrate the Denoising Diffusion Implicit Models (DDIM) sampling strategy to further improve inference efficiency. The proposed mechanism can be applied to both optimization-based and sampling-based schemes, providing an effective and flexible zero-shot solution for imaging inverse problems. Extensive experiments demonstrate that our method achieves superior performance across multiple inverse problems, particularly in compressive sensing, delivering high-quality reconstructions even at an extremely low sampling rate (5%).", "date": "2025-06-16", "categories": []}
{"id": "2506.13484", "title": "Deep Diffusion Models and Unsupervised Hyperspectral Unmixing for Realistic Abundance Map Synthesis", "abstract": "This paper presents a novel methodology for generating realistic abundance maps from hyperspectral imagery using an unsupervised, deep-learning-driven approach. Our framework integrates blind linear hyperspectral unmixing with state-of-the-art diffusion models to enhance the realism and diversity of synthetic abundance maps. First, we apply blind unmixing to extract endmembers and abundance maps directly from raw hyperspectral data. These abundance maps then serve as inputs to a diffusion model, which acts as a generative engine to synthesize highly realistic spatial distributions. Diffusion models have recently revolutionized image synthesis by offering superior performance, flexibility, and stability, making them well-suited for high-dimensional spectral data. By leveraging this combination of physically interpretable unmixing and deep generative modeling, our approach enables the simulation of hyperspectral sensor outputs under diverse imaging conditions--critical for data augmentation, algorithm benchmarking, and model evaluation in hyperspectral analysis. Notably, our method is entirely unsupervised, ensuring adaptability to different datasets without the need for labeled training data. We validate our approach using real hyperspectral imagery from the PRISMA space mission for Earth observation, demonstrating its effectiveness in producing realistic synthetic abundance maps that capture the spatial and spectral characteristics of natural scenes.", "date": "2025-06-16", "categories": []}
{"id": "2506.13488", "title": "Imaging at the quantum limit with convolutional neural networks", "abstract": "Deep neural networks have been shown to achieve exceptional performance for computer vision tasks like image recognition, segmentation, and reconstruction or denoising. Here, we evaluate the ultimate performance limits of deep convolutional neural network models for image reconstruction, by comparing them against the standard quantum limit set by shot-noise and the Heisenberg limit on precision. We train U-Net models on images of natural objects illuminated with coherent states of light, and find that the average mean-squared error of the reconstructions can surpass the standard quantum limit, and in some cases reaches the Heisenberg limit. Further, we train models on well-parameterized images for which we can calculate the quantum Cram\\'er-Rao bound to determine the minimum possible measurable variance of an estimated parameter for a given probe state. We find the mean-squared error of the model predictions reaches these bounds calculated for the parameters, across a variety of parameterized images. These results suggest that deep convolutional neural networks can learn to become the optimal estimators allowed by the laws of physics, performing parameter estimation and image reconstruction at the ultimate possible limits of precision for the case of classical illumination of the object.", "date": "2025-06-16", "categories": []}
{"id": "2506.13545", "title": "Limited-Angle CBCT Reconstruction via Geometry-Integrated Cycle-domain Denoising Diffusion Probabilistic Models", "abstract": "Cone-beam CT (CBCT) is widely used in clinical radiotherapy for image-guided treatment, improving setup accuracy, adaptive planning, and motion management. However, slow gantry rotation limits performance by introducing motion artifacts, blurring, and increased dose. This work aims to develop a clinically feasible method for reconstructing high-quality CBCT volumes from consecutive limited-angle acquisitions, addressing imaging challenges in time- or dose-constrained settings. We propose a limited-angle (LA) geometry-integrated cycle-domain (LA-GICD) framework for CBCT reconstruction, comprising two denoising diffusion probabilistic models (DDPMs) connected via analytic cone-beam forward and back projectors. A Projection-DDPM completes missing projections, followed by back-projection, and an Image-DDPM refines the volume. This dual-domain design leverages complementary priors from projection and image spaces to achieve high-quality reconstructions from limited-angle (<= 90 degrees) scans. Performance was evaluated against full-angle reconstruction. Four board-certified medical physicists conducted assessments. A total of 78 planning CTs in common CBCT geometries were used for training and evaluation. The method achieved a mean absolute error of 35.5 HU, SSIM of 0.84, and PSNR of 29.8 dB, with visibly reduced artifacts and improved soft-tissue clarity. LA-GICD's geometry-aware dual-domain learning, embedded in analytic forward/backward operators, enabled artifact-free, high-contrast reconstructions from a single 90-degree scan, reducing acquisition time and dose four-fold. LA-GICD improves limited-angle CBCT reconstruction with strong data fidelity and anatomical realism. It offers a practical solution for short-arc acquisitions, enhancing CBCT use in radiotherapy by providing clinically applicable images with reduced scan time and dose for more accurate, personalized treatments.", "date": "2025-06-16", "categories": []}
{"id": "2506.13614", "title": "Exploiting the Exact Denoising Posterior Score in Training-Free Guidance of Diffusion Models", "abstract": "The success of diffusion models has driven interest in performing conditional sampling via training-free guidance of the denoising process to solve image restoration and other inverse problems. A popular class of methods, based on Diffusion Posterior Sampling (DPS), attempts to approximate the intractable posterior score function directly. In this work, we present a novel expression for the exact posterior score for purely denoising tasks that is tractable in terms of the unconditional score function. We leverage this result to analyze the time-dependent error in the DPS score for denoising tasks and compute step sizes on the fly to minimize the error at each time step. We demonstrate that these step sizes are transferable to related inverse problems such as colorization, random inpainting, and super resolution. Despite its simplicity, this approach is competitive with state-of-the-art techniques and enables sampling with fewer time steps than DPS.", "date": "2025-06-16", "categories": []}
{"id": "2506.11183", "title": "DiffPR: Diffusion-Based Phase Reconstruction via Frequency-Decoupled Learning", "abstract": "Oversmoothing remains a persistent problem when applying deep learning to off-axis quantitative phase imaging (QPI). End-to-end U-Nets favour low-frequency content and under-represent fine, diagnostic detail. We trace this issue to spectral bias and show that the bias is reinforced by high-level skip connections that feed high-frequency features directly into the decoder. Removing those deepest skips thus supervising the network only at a low resolution significantly improves generalisation and fidelity. Building on this insight, we introduce DiffPR, a two-stage frequency-decoupled framework. Stage 1: an asymmetric U-Net with cancelled high-frequency skips predicts a quarter-scale phase map from the interferogram, capturing reliable low-frequency structure while avoiding spectral bias. Stage 2: the upsampled prediction, lightly perturbed with Gaussian noise, is refined by an unconditional diffusion model that iteratively recovers the missing high-frequency residuals through reverse denoising. Experiments on four QPI datasets (B-Cell, WBC, HeLa, 3T3) show that DiffPR outperforms strong U-Net baselines, boosting PSNR by up to 1.1 dB and reducing MAE by 11 percent, while delivering markedly sharper membrane ridges and speckle patterns. The results demonstrate that cancelling high-level skips and delegating detail synthesis to a diffusion prior is an effective remedy for the spectral bias that limits conventional phase-retrieval networks.", "date": "2025-06-12", "categories": []}
{"id": "2506.11732", "title": "Data-driven approaches to inverse problems", "abstract": "Inverse problems are concerned with the reconstruction of unknown physical quantities using indirect measurements and are fundamental across diverse fields such as medical imaging, remote sensing, and material sciences. These problems serve as critical tools for visualizing internal structures beyond what is visible to the naked eye, enabling quantification, diagnosis, prediction, and discovery. However, most inverse problems are ill-posed, necessitating robust mathematical treatment to yield meaningful solutions. While classical approaches provide mathematically rigorous and computationally stable solutions, they are constrained by the ability to accurately model solution properties and implement them efficiently.\n  A more recent paradigm considers deriving solutions to inverse problems in a data-driven manner. Instead of relying on classical mathematical modeling, this approach utilizes highly over-parameterized models, typically deep neural networks, which are adapted to specific inverse problems using carefully selected training data. Current approaches that follow this new paradigm distinguish themselves through solution accuracy paired with computational efficiency that was previously inconceivable.\n  These notes offer an introduction to this data-driven paradigm for inverse problems. The first part of these notes will provide an introduction to inverse problems, discuss classical solution strategies, and present some applications. The second part will delve into modern data-driven approaches, with a particular focus on adversarial regularization and provably convergent linear plug-and-play denoisers. Throughout the presentation of these methodologies, their theoretical properties will be discussed, and numerical examples will be provided. The lecture series will conclude with a discussion of open problems and future perspectives in the field.", "date": "2025-06-13", "categories": []}
{"id": "2506.10141", "title": "Diffusion prior as a direct regularization term for FWI", "abstract": "Diffusion models have recently shown promise as powerful generative priors for inverse problems. However, conventional applications require solving the full reverse diffusion process and operating on noisy intermediate states, which poses challenges for physics-constrained computational seismic imaging. In particular, such instability is pronounced in non-linear solvers like those used in Full Waveform Inversion (FWI), where wave propagation through noisy velocity fields can lead to numerical artifacts and poor inversion quality. In this work, we propose a simple yet effective framework that directly integrates a pretrained Denoising Diffusion Probabilistic Model (DDPM) as a score-based generative diffusion prior into FWI through a score rematching strategy. Unlike traditional diffusion approaches, our method avoids the reverse diffusion sampling and needs fewer iterations. We operate the image inversion entirely in the clean image space, eliminating the need to operate through noisy velocity models. The generative diffusion prior can be introduced as a simple regularization term in the standard FWI update rule, requiring minimal modification to existing FWI pipelines. This promotes stable wave propagation and can improve convergence behavior and inversion quality. Numerical experiments suggest that the proposed method offers enhanced fidelity and robustness compared to conventional and GAN-based FWI approaches, while remaining practical and computationally efficient for seismic imaging and other inverse problem tasks.", "date": "2025-06-11", "categories": []}
{"id": "2506.10309", "title": "DUN-SRE: Deep Unrolling Network with Spatiotemporal Rotation Equivariance for Dynamic MRI Reconstruction", "abstract": "Dynamic Magnetic Resonance Imaging (MRI) exhibits transformation symmetries, including spatial rotation symmetry within individual frames and temporal symmetry along the time dimension. Explicit incorporation of these symmetry priors in the reconstruction model can significantly improve image quality, especially under aggressive undersampling scenarios. Recently, Equivariant convolutional neural network (ECNN) has shown great promise in exploiting spatial symmetry priors. However, existing ECNNs critically fail to model temporal symmetry, arguably the most universal and informative structural prior in dynamic MRI reconstruction. To tackle this issue, we propose a novel Deep Unrolling Network with Spatiotemporal Rotation Equivariance (DUN-SRE) for Dynamic MRI Reconstruction. The DUN-SRE establishes spatiotemporal equivariance through a (2+1)D equivariant convolutional architecture. In particular, it integrates both the data consistency and proximal mapping module into a unified deep unrolling framework. This architecture ensures rigorous propagation of spatiotemporal rotation symmetry constraints throughout the reconstruction process, enabling more physically accurate modeling of cardiac motion dynamics in cine MRI. In addition, a high-fidelity group filter parameterization mechanism is developed to maintain representation precision while enforcing symmetry constraints. Comprehensive experiments on Cardiac CINE MRI datasets demonstrate that DUN-SRE achieves state-of-the-art performance, particularly in preserving rotation-symmetric structures, offering strong generalization capability to a broad range of dynamic MRI reconstruction tasks.", "date": "2025-06-12", "categories": []}
{"id": "2405.18410", "title": "Towards a Sampling Theory for Implicit Neural Representations", "abstract": "Implicit neural representations (INRs) have emerged as a powerful tool for solving inverse problems in computer vision and computational imaging. INRs represent images as continuous domain functions realized by a neural network taking spatial coordinates as inputs. However, unlike traditional pixel representations, little is known about the sample complexity of estimating images using INRs in the context of linear inverse problems. Towards this end, we study the sampling requirements for recovery of a continuous domain image from its low-pass Fourier coefficients by fitting a single hidden-layer INR with ReLU activation and a Fourier features layer using a generalized form of weight decay regularization. Our key insight is to relate minimizers of this non-convex parameter space optimization problem to minimizers of a convex penalty defined over an infinite-dimensional space of measures. We identify a sufficient number of samples for which an image realized by a width-1 INR is exactly recoverable by solving the INR training problem, and give a conjecture for the general width-$W$ case. To validate our theory, we empirically assess the probability of achieving exact recovery of images realized by low-width single hidden-layer INRs, and illustrate the performance of INR on super-resolution recovery of more realistic continuous domain phantom images.", "date": "2024-05-28", "categories": []}
{"id": "2501.03874", "title": "Neuromorphic Optical Tracking and Imaging of Randomly Moving Targets through Strongly Scattering Media", "abstract": "Tracking and acquiring simultaneous optical images of randomly moving targets obscured by scattering media remains a challenging problem of importance to many applications that require precise object localization and identification. In this work we develop an end-to-end neuromorphic optical engineering and computational approach to demonstrate how to track and image normally invisible objects by combining an event detecting camera with a multistage neuromorphic deep learning strategy. Photons emerging from dense scattering media are detected by the event camera and converted to pixel-wise asynchronized spike trains - a first step in isolating object-specific information from the dominant uninformative background. Spiking data is fed into a deep spiking neural network (SNN) engine where object tracking and image reconstruction are performed by two separate yet interconnected modules running in parallel in discrete time steps over the event duration. Through benchtop experiments we demonstrate tracking and imaging randomly moving objects in dense turbid media as well as image reconstruction of spatially stationary but optically dynamic objects. Standardized character sets serve as representative proxies for geometrically complex objects, underscoring the method's generality. The results highlight the advantages of a fully neuromorphic approach in meeting a major imaging technology with high computational efficiency and low power consumption.", "date": "2025-01-07", "categories": []}
{"id": "2501.16583", "title": "Directing Mamba to Complex Textures: An Efficient Texture-Aware State Space Model for Image Restoration", "abstract": "Image restoration aims to recover details and enhance contrast in degraded images. With the growing demand for high-quality imaging (\\textit{e.g.}, 4K and 8K), achieving a balance between restoration quality and computational efficiency has become increasingly critical. Existing methods, primarily based on CNNs, Transformers, or their hybrid approaches, apply uniform deep representation extraction across the image. However, these methods often struggle to effectively model long-range dependencies and largely overlook the spatial characteristics of image degradation (regions with richer textures tend to suffer more severe damage), making it hard to achieve the best trade-off between restoration quality and efficiency. To address these issues, we propose a novel texture-aware image restoration method, TAMambaIR, which simultaneously perceives image textures and achieves a trade-off between performance and efficiency. Specifically, we introduce a novel Texture-Aware State Space Model, which enhances texture awareness and improves efficiency by modulating the transition matrix of the state-space equation and focusing on regions with complex textures. Additionally, we design a {Multi-Directional Perception Block} to improve multi-directional receptive fields while maintaining low computational overhead. Extensive experiments on benchmarks for image super-resolution, deraining, and low-light image enhancement demonstrate that TAMambaIR achieves state-of-the-art performance with significantly improved efficiency, establishing it as a robust and efficient framework for image restoration.", "date": "2025-01-27", "categories": []}
{"id": "2506.08650", "title": "Beyond Calibration: Physically Informed Learning for Raw-to-Raw Mapping", "abstract": "Achieving consistent color reproduction across multiple cameras is essential for seamless image fusion and Image Processing Pipeline (ISP) compatibility in modern devices, but it is a challenging task due to variations in sensors and optics. Existing raw-to-raw conversion methods face limitations such as poor adaptability to changing illumination, high computational costs, or impractical requirements such as simultaneous camera operation and overlapping fields-of-view. We introduce the Neural Physical Model (NPM), a lightweight, physically-informed approach that simulates raw images under specified illumination to estimate transformations between devices. The NPM effectively adapts to varying illumination conditions, can be initialized with physical measurements, and supports training with or without paired data. Experiments on public datasets like NUS and BeyondRGB demonstrate that NPM outperforms recent state-of-the-art methods, providing robust chromatic consistency across different sensors and optical systems.", "date": "2025-06-10", "categories": []}
{"id": "2506.09949", "title": "Sampling Theory for Super-Resolution with Implicit Neural Representations", "abstract": "Implicit neural representations (INRs) have emerged as a powerful tool for solving inverse problems in computer vision and computational imaging. INRs represent images as continuous domain functions realized by a neural network taking spatial coordinates as inputs. However, unlike traditional pixel representations, little is known about the sample complexity of estimating images using INRs in the context of linear inverse problems. Towards this end, we study the sampling requirements for recovery of a continuous domain image from its low-pass Fourier samples by fitting a single hidden-layer INR with ReLU activation and a Fourier features layer using a generalized form of weight decay regularization. Our key insight is to relate minimizers of this non-convex parameter space optimization problem to minimizers of a convex penalty defined over an infinite-dimensional space of measures. We identify a sufficient number of Fourier samples for which an image realized by an INR is exactly recoverable by solving the INR training problem. To validate our theory, we empirically assess the probability of achieving exact recovery of images realized by low-width single hidden-layer INRs, and illustrate the performance of INRs on super-resolution recovery of continuous domain phantom images.", "date": "2025-06-11", "categories": []}
{"id": "2408.11531", "title": "Just Project! Multi-Channel Despeckling, the Easy Way", "abstract": "Reducing speckle fluctuations in multi-channel SAR images is essential in many applications of SAR imaging such as polarimetric classification or interferometric height estimation. While single-channel despeckling has widely benefited from the application of deep learning techniques, extensions to multi-channel SAR images are much more challenging. This paper introduces MuChaPro, a generic framework that exploits existing single-channel despeckling methods. The key idea is to generate numerous single-channel projections, restore these projections, and recombine them into the final multi-channel estimate. This simple approach is shown to be effective in polarimetric and/or interferometric modalities. A special appeal of MuChaPro is the possibility to apply a self-supervised training strategy to learn sensor-specific networks for single-channel despeckling.", "date": "2024-08-21", "categories": []}
{"id": "2504.07549", "title": "STeP: A Framework for Solving Scientific Video Inverse Problems with Spatiotemporal Diffusion Priors", "abstract": "Reconstructing spatially and temporally coherent videos from time-varying measurements is a fundamental challenge in many scientific domains. A major difficulty arises from the sparsity of measurements, which hinders accurate recovery of temporal dynamics. Existing image diffusion-based methods rely on extracting temporal consistency directly from measurements, limiting their effectiveness on scientific tasks with high spatiotemporal uncertainty. We address this difficulty by proposing a plug-and-play framework that incorporates a learned spatiotemporal diffusion prior. Due to its plug-and-play nature, our framework can be flexibly applied to different video inverse problems without the need for task-specific design and temporal heuristics. We further demonstrate that a spatiotemporal diffusion model can be trained efficiently with limited video data. We validate our approach on two challenging scientific video reconstruction tasks: black hole video reconstruction and dynamic MRI. While baseline methods struggle to provide temporally coherent reconstructions, our approach achieves significantly improved recovery of the spatiotemporal structure of the underlying ground truth videos.", "date": "2025-04-10", "categories": []}
{"id": "2506.08361", "title": "Image Demoir\\'eing Using Dual Camera Fusion on Mobile Phones", "abstract": "When shooting electronic screens, moir\\'e patterns usually appear in captured images, which seriously affects the image quality. Existing image demoir\\'eing methods face great challenges in removing large and heavy moir\\'e. To address the issue, we propose to utilize Dual Camera fusion for Image Demoir\\'eing (DCID), \\ie, using the ultra-wide-angle (UW) image to assist the moir\\'e removal of wide-angle (W) image. This is inspired by two motivations: (1) the two lenses are commonly equipped with modern smartphones, (2) the UW image generally can provide normal colors and textures when moir\\'e exists in the W image mainly due to their different focal lengths. In particular, we propose an efficient DCID method, where a lightweight UW image encoder is integrated into an existing demoir\\'eing network and a fast two-stage image alignment manner is present. Moreover, we construct a large-scale real-world dataset with diverse mobile phones and monitors, containing about 9,000 samples. Experiments on the dataset show our method performs better than state-of-the-art methods. Code and dataset are available at https://github.com/Mrduckk/DCID.", "date": "2025-06-10", "categories": []}
{"id": "2506.08470", "title": "MARMOT: Masked Autoencoder for Modeling Transient Imaging", "abstract": "Pretrained models have demonstrated impressive success in many modalities such as language and vision. Recent works facilitate the pretraining paradigm in imaging research. Transients are a novel modality, which are captured for an object as photon counts versus arrival times using a precisely time-resolved sensor. In particular for non-line-of-sight (NLOS) scenarios, transients of hidden objects are measured beyond the sensor's direct line of sight. Using NLOS transients, the majority of previous works optimize volume density or surfaces to reconstruct the hidden objects and do not transfer priors learned from datasets. In this work, we present a masked autoencoder for modeling transient imaging, or MARMOT, to facilitate NLOS applications. Our MARMOT is a self-supervised model pretrianed on massive and diverse NLOS transient datasets. Using a Transformer-based encoder-decoder, MARMOT learns features from partially masked transients via a scanning pattern mask (SPM), where the unmasked subset is functionally equivalent to arbitrary sampling, and predicts full measurements. Pretrained on TransVerse-a synthesized transient dataset of 500K 3D models-MARMOT adapts to downstream imaging tasks using direct feature transfer or decoder finetuning. Comprehensive experiments are carried out in comparisons with state-of-the-art methods. Quantitative and qualitative results demonstrate the efficiency of our MARMOT.", "date": "2025-06-10", "categories": []}
{"id": "2305.09986", "title": "A robust multi-domain network for short-scanning amyloid PET reconstruction", "abstract": "This paper presents a robust multi-domain network designed to restore low-quality amyloid PET images acquired in a short period of time. The proposed method is trained on pairs of PET images from short (2 minutes) and standard (20 minutes) scanning times, sourced from multiple domains. Learning relevant image features between these domains with a single network is challenging. Our key contribution is the introduction of a mapping label, which enables effective learning of specific representations between different domains. The network, trained with various mapping labels, can efficiently correct amyloid PET datasets in multiple training domains and unseen domains, such as those obtained with new radiotracers, acquisition protocols, or PET scanners. Internal, temporal, and external validations demonstrate the effectiveness of the proposed method. Notably, for external validation datasets from unseen domains, the proposed method achieved comparable or superior results relative to methods trained with these datasets, in terms of quantitative metrics such as normalized root mean-square error and structure similarity index measure. Two nuclear medicine physicians evaluated the amyloid status as positive or negative for the external validation datasets, with accuracies of 0.970 and 0.930 for readers 1 and 2, respectively.", "date": "2023-05-17", "categories": []}
{"id": "2307.06343", "title": "Sequential Experimental Design for X-Ray CT Using Deep Reinforcement Learning", "abstract": "In X-ray Computed Tomography (CT), projections from many angles are acquired and used for 3D reconstruction. To make CT suitable for in-line quality control, reducing the number of angles while maintaining reconstruction quality is necessary. Sparse-angle tomography is a popular approach for obtaining 3D reconstructions from limited data. To optimize its performance, one can adapt scan angles sequentially to select the most informative angles for each scanned object. Mathematically, this corresponds to solving an optimal experimental design (OED) problem. OED problems are high-dimensional, non-convex, bi-level optimization problems that cannot be solved online, i.e., during the scan. To address these challenges, we pose the OED problem as a partially observable Markov decision process in a Bayesian framework, and solve it through deep reinforcement learning. The approach learns efficient non-greedy policies to solve a given class of OED problems through extensive offline training rather than solving a given OED problem directly via numerical optimization. As such, the trained policy can successfully find the most informative scan angles online. We use a policy training method based on the Actor-Critic approach and evaluate its performance on 2D tomography with synthetic data.", "date": "2023-07-12", "categories": []}
{"id": "2410.08646", "title": "Fully Unsupervised Dynamic MRI Reconstruction via Diffeo-Temporal Equivariance", "abstract": "Reconstructing dynamic MRI image sequences from undersampled accelerated measurements is crucial for faster and higher spatiotemporal resolution real-time imaging of cardiac motion, free breathing motion and many other applications. Classical paradigms, such as gated cine MRI, assume periodicity, disallowing imaging of true motion. Supervised deep learning methods are fundamentally flawed as, in dynamic imaging, ground truth fully-sampled videos are impossible to truly obtain. We propose an unsupervised framework to learn to reconstruct dynamic MRI sequences from undersampled measurements alone by leveraging natural geometric spatiotemporal equivariances of MRI. Dynamic Diffeomorphic Equivariant Imaging (DDEI) significantly outperforms state-of-the-art unsupervised methods such as SSDU on highly accelerated dynamic cardiac imaging. Our method is agnostic to the underlying neural network architecture and can be used to adapt the latest models and post-processing approaches. Our code and video demos are at https://github.com/Andrewwango/ddei.", "date": "2024-10-11", "categories": []}
{"id": "2411.18970", "title": "FiRe: Fixed-points of Restoration Priors for Solving Inverse Problems", "abstract": "Selecting an appropriate prior to compensate for information loss due to the measurement operator is a fundamental challenge in imaging inverse problems. Implicit priors based on denoising neural networks have become central to widely-used frameworks such as Plug-and-Play (PnP) algorithms. In this work, we introduce Fixed-points of Restoration (FiRe) priors as a new framework for expanding the notion of priors in PnP to general restoration models beyond traditional denoising models. The key insight behind FiRe is that smooth images emerge as fixed points of the composition of a degradation operator with the corresponding restoration model. This enables us to derive an explicit formula for our implicit prior by quantifying invariance of images under this composite operation. Adopting this fixed-point perspective, we show how various restoration networks can effectively serve as priors for solving inverse problems. The FiRe framework further enables ensemble-like combinations of multiple restoration models as well as acquisition-informed restoration networks, all within a unified optimization approach. Experimental results validate the effectiveness of FiRe across various inverse problems, establishing a new paradigm for incorporating pretrained restoration models into PnP-like algorithms. Code available at https://github.com/matthieutrs/fire.", "date": "2024-11-28", "categories": []}
{"id": "2412.00087", "title": "Physics-Informed Deep Learning Model for Line-integral Diagnostics Across Fusion Devices", "abstract": "Rapid reconstruction of 2D plasma profiles from line-integral measurements is important in nuclear fusion. This paper introduces a physics-informed model architecture called Onion, that can enhance the performance of models and be adapted to various backbone networks. The model under Onion incorporates physical information by a multiplication process and applies the physics-informed loss function according to the principle of line integration. Prediction results demonstrate that the additional input of physical information improves the deep learning model's ability, leading to a reduction in the average relative error E_1 between the reconstruction profiles and the target profiles by approximately 0.84x10^(-2) on synthetic datasets and about 0.06x10^(-2) on experimental datasets. Furthermore, the implementation of the Softplus activation function in the final two fully connected layers improves model performance. This enhancement results in a reduction in the E_1 by approximately 1.06x10^(-2) on synthetic datasets and about 0.11x10^(-2) on experimental datasets. The incorporation of the physics-informed loss function has been shown to correct the model's predictions, bringing the back-projections closer to the actual inputs and reducing the errors associated with inversion algorithms. Besides, we have developed a synthetic data model to generate customized line-integral diagnostic datasets and have also collected soft x-ray diagnostic datasets from EAST and HL-2A. This study achieves reductions in reconstruction errors, and accelerates the development of surrogate models in fusion research.", "date": "2024-11-27", "categories": []}
{"id": "2506.06993", "title": "DM$^3$Net: Dual-Camera Super-Resolution via Domain Modulation and Multi-scale Matching", "abstract": "Dual-camera super-resolution is highly practical for smartphone photography that primarily super-resolve the wide-angle images using the telephoto image as a reference. In this paper, we propose DM$^3$Net, a novel dual-camera super-resolution network based on Domain Modulation and Multi-scale Matching. To bridge the domain gap between the high-resolution domain and the degraded domain, we learn two compressed global representations from image pairs corresponding to the two domains. To enable reliable transfer of high-frequency structural details from the reference image, we design a multi-scale matching module that conducts patch-level feature matching and retrieval across multiple receptive fields to improve matching accuracy and robustness. Moreover, we also introduce Key Pruning to achieve a significant reduction in memory usage and inference time with little model performance sacrificed. Experimental results on three real-world datasets demonstrate that our DM$^3$Net outperforms the state-of-the-art approaches.", "date": "2025-06-08", "categories": []}
{"id": "2504.10984", "title": "Seeing like a Cephalopod: Colour Vision with a Monochrome Event Camera", "abstract": "Cephalopods exhibit unique colour discrimination capabilities despite having one type of photoreceptor, relying instead on chromatic aberration induced by their ocular optics and pupil shapes to perceive spectral information. We took inspiration from this biological mechanism to design a spectral imaging system that combines a ball lens with an event-based camera. Our approach relies on a motorised system that shifts the focal position, mirroring the adaptive lens motion in cephalopods. This approach has enabled us to achieve wavelength-dependent focusing across the visible light and near-infrared spectrum, making the event a spectral sensor. We characterise chromatic aberration effects, using both event-based and conventional frame-based sensors, validating the effectiveness of bio-inspired spectral discrimination both in simulation and in a real setup as well as assessing the spectral discrimination performance. Our proposed approach provides a robust spectral sensing capability without conventional colour filters or computational demosaicing. This approach opens new pathways toward new spectral sensing systems inspired by nature's evolutionary solutions. Code and analysis are available at: https://samiarja.github.io/neuromorphic_octopus_eye/", "date": "2025-04-15", "categories": []}
{"id": "2506.04444", "title": "Photoreal Scene Reconstruction from an Egocentric Device", "abstract": "In this paper, we investigate the challenges associated with using egocentric devices to photorealistic reconstruct the scene in high dynamic range. Existing methodologies typically assume using frame-rate 6DoF pose estimated from the device's visual-inertial odometry system, which may neglect crucial details necessary for pixel-accurate reconstruction. This study presents two significant findings. Firstly, in contrast to mainstream work treating RGB camera as global shutter frame-rate camera, we emphasize the importance of employing visual-inertial bundle adjustment (VIBA) to calibrate the precise timestamps and movement of the rolling shutter RGB sensing camera in a high frequency trajectory format, which ensures an accurate calibration of the physical properties of the rolling-shutter camera. Secondly, we incorporate a physical image formation model based into Gaussian Splatting, which effectively addresses the sensor characteristics, including the rolling-shutter effect of RGB cameras and the dynamic ranges measured by sensors. Our proposed formulation is applicable to the widely-used variants of Gaussian Splats representation. We conduct a comprehensive evaluation of our pipeline using the open-source Project Aria device under diverse indoor and outdoor lighting conditions, and further validate it on a Meta Quest3 device. Across all experiments, we observe a consistent visual enhancement of +1 dB in PSNR by incorporating VIBA, with an additional +1 dB achieved through our proposed image formation model. Our complete implementation, evaluation datasets, and recording profile are available at http://www.projectaria.com/photoreal-reconstruction/", "date": "2025-06-04", "categories": []}
{"id": "2506.04753", "title": "Physics Informed Capsule Enhanced Variational AutoEncoder for Underwater Image Enhancement", "abstract": "We present a novel dual-stream architecture that achieves state-of-the-art underwater image enhancement by explicitly integrating the Jaffe-McGlamery physical model with capsule clustering-based feature representation learning. Our method simultaneously estimates transmission maps and spatially-varying background light through a dedicated physics estimator while extracting entity-level features via capsule clustering in a parallel stream. This physics-guided approach enables parameter-free enhancement that respects underwater formation constraints while preserving semantic structures and fine-grained details. Our approach also features a novel optimization objective ensuring both physical adherence and perceptual quality across multiple spatial frequencies. To validate our approach, we conducted extensive experiments across six challenging benchmarks. Results demonstrate consistent improvements of $+0.5$dB PSNR over the best existing methods while requiring only one-third of their computational complexity (FLOPs), or alternatively, more than $+1$dB PSNR improvement when compared to methods with similar computational budgets. Code and data \\textit{will} be available at https://github.com/iN1k1/.", "date": "2025-06-05", "categories": []}
{"id": "2506.02197", "title": "NTIRE 2025 Challenge on RAW Image Restoration and Super-Resolution", "abstract": "This paper reviews the NTIRE 2025 RAW Image Restoration and Super-Resolution Challenge, highlighting the proposed solutions and results. New methods for RAW Restoration and Super-Resolution could be essential in modern Image Signal Processing (ISP) pipelines, however, this problem is not as explored as in the RGB domain. The goal of this challenge is two fold, (i) restore RAW images with blur and noise degradations, (ii) upscale RAW Bayer images by 2x, considering unknown noise and blur. In the challenge, a total of 230 participants registered, and 45 submitted results during thee challenge period. This report presents the current state-of-the-art in RAW Restoration.", "date": "2025-06-02", "categories": []}
{"id": "2506.03645", "title": "YOND: Practical Blind Raw Image Denoising Free from Camera-Specific Data Dependency", "abstract": "The rapid advancement of photography has created a growing demand for a practical blind raw image denoising method. Recently, learning-based methods have become mainstream due to their excellent performance. However, most existing learning-based methods suffer from camera-specific data dependency, resulting in performance drops when applied to data from unknown cameras. To address this challenge, we introduce a novel blind raw image denoising method named YOND, which represents You Only Need a Denoiser. Trained solely on synthetic data, YOND can generalize robustly to noisy raw images captured by diverse unknown cameras. Specifically, we propose three key modules to guarantee the practicality of YOND: coarse-to-fine noise estimation (CNE), expectation-matched variance-stabilizing transform (EM-VST), and SNR-guided denoiser (SNR-Net). Firstly, we propose CNE to identify the camera noise characteristic, refining the estimated noise parameters based on the coarse denoised image. Secondly, we propose EM-VST to eliminate camera-specific data dependency, correcting the bias expectation of VST according to the noisy image. Finally, we propose SNR-Net to offer controllable raw image denoising, supporting adaptive adjustments and manual fine-tuning. Extensive experiments on unknown cameras, along with flexible solutions for challenging cases, demonstrate the superior practicality of our method. The source code will be publicly available at the \\href{https://fenghansen.github.io/publication/YOND}{project homepage}.", "date": "2025-06-04", "categories": []}
{"id": "2506.03792", "title": "Analytical Reconstruction of Periodically Deformed Objects in Time-resolved CT", "abstract": "Time-resolved CT is an advanced measurement technique that has been widely used to observe dynamic objects, including periodically varying structures such as hearts, lungs, or hearing structures. To reconstruct these objects from CT projections, a common approach is to divide the projections into several collections based on their motion phases and perform reconstruction within each collection, assuming they originate from a static object. This describes the gating-based method, which is the standard approach for time-periodic reconstruction. However, the gating-based reconstruction algorithm only utilizes a limited subset of projections within each collection and ignores the correlation between different collections, leading to inefficient use of the radiation dose. To address this issue, we propose two analytical reconstruction pipelines in this paper, and validate them with experimental data captured using tomographic synchrotron microscopy. We demonstrate that our approaches significantly reduce random noise in the reconstructed images without blurring the sharp features of the observed objects. Equivalently, our methods can achieve the same reconstruction quality as gating-based methods but with a lower radiation dose. Our code is available at github.com/PeriodRecon.", "date": "2025-06-04", "categories": []}
{"id": "2506.04054", "title": "Video Deblurring with Deconvolution and Aggregation Networks", "abstract": "In contrast to single-image deblurring, video deblurring has the advantage that neighbor frames can be utilized to deblur a target frame. However, existing video deblurring algorithms often fail to properly employ the neighbor frames, resulting in sub-optimal performance. In this paper, we propose a deconvolution and aggregation network (DAN) for video deblurring that utilizes the information of neighbor frames well. In DAN, both deconvolution and aggregation strategies are achieved through three sub-networks: the preprocessing network (PPN) and the alignment-based deconvolution network (ABDN) for the deconvolution scheme; the frame aggregation network (FAN) for the aggregation scheme. In the deconvolution part, blurry inputs are first preprocessed by the PPN with non-local operations. Then, the output frames from the PPN are deblurred by the ABDN based on the frame alignment. In the FAN, these deblurred frames from the deconvolution part are combined into a latent frame according to reliability maps which infer pixel-wise sharpness. The proper combination of three sub-networks can achieve favorable performance on video deblurring by using the neighbor frames suitably. In experiments, the proposed DAN was demonstrated to be superior to existing state-of-the-art methods through both quantitative and qualitative evaluations on the public datasets.", "date": "2025-06-04", "categories": []}
{"id": "2412.04339", "title": "Likelihood-Scheduled Score-Based Generative Modeling for Fully 3D PET Image Reconstruction", "abstract": "Medical image reconstruction with pre-trained score-based generative models (SGMs) has advantages over other existing state-of-the-art deep-learned reconstruction methods, including improved resilience to different scanner setups and advanced image distribution modeling. SGM-based reconstruction has recently been applied to simulated positron emission tomography (PET) datasets, showing improved contrast recovery for out-of-distribution lesions relative to the state-of-the-art. However, existing methods for SGM-based reconstruction from PET data suffer from slow reconstruction, burdensome hyperparameter tuning and slice inconsistency effects (in 3D). In this work, we propose a practical methodology for fully 3D reconstruction that accelerates reconstruction and reduces the number of critical hyperparameters by matching the likelihood of an SGM's reverse diffusion process to a current iterate of the maximum-likelihood expectation maximization algorithm. Using the example of low-count reconstruction from simulated [$^{18}$F]DPA-714 datasets, we show our methodology can match or improve on the NRMSE and SSIM of existing state-of-the-art SGM-based PET reconstruction while reducing reconstruction time and the need for hyperparameter tuning. We evaluate our methodology against state-of-the-art supervised and conventional reconstruction algorithms. Finally, we demonstrate a first-ever implementation of SGM-based reconstruction for real 3D PET data, specifically [$^{18}$F]DPA-714 data, where we integrate perpendicular pre-trained SGMs to eliminate slice inconsistency issues.", "date": "2024-12-05", "categories": []}
{"id": "2506.02364", "title": "A TRPCA-Inspired Deep Unfolding Network for Hyperspectral Image Denoising via Thresholded t-SVD and Top-K Sparse Transformer", "abstract": "Hyperspectral images (HSIs) are often degraded by complex mixed noise during acquisition and transmission, making effective denoising essential for subsequent analysis. Recent hybrid approaches that bridge model-driven and data-driven paradigms have shown great promise. However, most of these approaches lack effective alternation between different priors or modules, resulting in loosely coupled regularization and insufficient exploitation of their complementary strengths. Inspired by tensor robust principal component analysis (TRPCA), we propose a novel deep unfolding network (DU-TRPCA) that enforces stage-wise alternation between two tightly integrated modules: low-rank and sparse. The low-rank module employs thresholded tensor singular value decomposition (t-SVD), providing a widely adopted convex surrogate for tensor low-rankness and has been demonstrated to effectively capture the global spatial-spectral structure of HSIs. The Top-K sparse transformer module adaptively imposes sparse constraints, directly matching the sparse regularization in TRPCA and enabling effective removal of localized outliers and complex noise. This tightly coupled architecture preserves the stage-wise alternation between low-rank approximation and sparse refinement inherent in TRPCA, while enhancing representational capacity through attention mechanisms. Extensive experiments on synthetic and real-world HSIs demonstrate that DU-TRPCA surpasses state-of-the-art methods under severe mixed noise, while offering interpretability benefits and stable denoising dynamics inspired by iterative optimization. Code is available at https://github.com/liangli97/TRPCA-Deep-Unfolding-HSI-Denoising.", "date": "2025-06-03", "categories": []}
{"id": "2506.02605", "title": "One-Step Diffusion-based Real-World Image Super-Resolution with Visual Perception Distillation", "abstract": "Diffusion-based models have been widely used in various visual generation tasks, showing promising results in image super-resolution (SR), while typically being limited by dozens or even hundreds of sampling steps. Although existing methods aim to accelerate the inference speed of multi-step diffusion-based SR methods through knowledge distillation, their generated images exhibit insufficient semantic alignment with real images, resulting in suboptimal perceptual quality reconstruction, specifically reflected in the CLIPIQA score. These methods still have many challenges in perceptual quality and semantic fidelity. Based on the challenges, we propose VPD-SR, a novel visual perception diffusion distillation framework specifically designed for SR, aiming to construct an effective and efficient one-step SR model. Specifically, VPD-SR consists of two components: Explicit Semantic-aware Supervision (ESS) and High-Frequency Perception (HFP) loss. Firstly, the ESS leverages the powerful visual perceptual understanding capabilities of the CLIP model to extract explicit semantic supervision, thereby enhancing semantic consistency. Then, Considering that high-frequency information contributes to the visual perception quality of images, in addition to the vanilla distillation loss, the HFP loss guides the student model to restore the missing high-frequency details in degraded images that are critical for enhancing perceptual quality. Lastly, we expand VPD-SR in adversarial training manner to further enhance the authenticity of the generated content. Extensive experiments conducted on synthetic and real-world datasets demonstrate that the proposed VPD-SR achieves superior performance compared to both previous state-of-the-art methods and the teacher model with just one-step sampling.", "date": "2025-06-03", "categories": []}
{"id": "2506.02981", "title": "Astrophotography turbulence mitigation via generative models", "abstract": "Photography is the cornerstone of modern astronomical and space research. However, most astronomical images captured by ground-based telescopes suffer from atmospheric turbulence, resulting in degraded imaging quality. While multi-frame strategies like lucky imaging can mitigate some effects, they involve intensive data acquisition and complex manual processing. In this paper, we propose AstroDiff, a generative restoration method that leverages both the high-quality generative priors and restoration capabilities of diffusion models to mitigate atmospheric turbulence. Extensive experiments demonstrate that AstroDiff outperforms existing state-of-the-art learning-based methods in astronomical image turbulence mitigation, providing higher perceptual quality and better structural fidelity under severe turbulence conditions. Our code and additional results are available at https://web-six-kappa-66.vercel.app/", "date": "2025-06-03", "categories": []}
{"id": "2501.13134", "title": "UniRestore: Unified Perceptual and Task-Oriented Image Restoration Model Using Diffusion Prior", "abstract": "Image restoration aims to recover content from inputs degraded by various factors, such as adverse weather, blur, and noise. Perceptual Image Restoration (PIR) methods improve visual quality but often do not support downstream tasks effectively. On the other hand, Task-oriented Image Restoration (TIR) methods focus on enhancing image utility for high-level vision tasks, sometimes compromising visual quality. This paper introduces UniRestore, a unified image restoration model that bridges the gap between PIR and TIR by using a diffusion prior. The diffusion prior is designed to generate images that align with human visual quality preferences, but these images are often unsuitable for TIR scenarios. To solve this limitation, UniRestore utilizes encoder features from an autoencoder to adapt the diffusion prior to specific tasks. We propose a Complementary Feature Restoration Module (CFRM) to reconstruct degraded encoder features and a Task Feature Adapter (TFA) module to facilitate adaptive feature fusion in the decoder. This design allows UniRestore to optimize images for both human perception and downstream task requirements, addressing discrepancies between visual quality and functional needs. Integrating these modules also enhances UniRestore's adapability and efficiency across diverse tasks. Extensive expertments demonstrate the superior performance of UniRestore in both PIR and TIR scenarios.", "date": "2025-01-22", "categories": []}
{"id": "2505.03149", "title": "Motion-compensated cardiac MRI using low-rank diffeomorphic flow (DMoCo)", "abstract": "We introduce an unsupervised motion-compensated image reconstruction algorithm for free-breathing and ungated 3D cardiac magnetic resonance imaging (MRI). We express the image volume corresponding to each specific motion phase as the deformation of a single static image template. The main contribution of the work is the low-rank model for the compact joint representation of the family of diffeomorphisms, parameterized by the motion phases. The diffeomorphism at a specific motion phase is obtained by integrating a parametric velocity field along a path connecting the reference template phase to the motion phase. The velocity field at different phases is represented using a low-rank model. The static template and the low-rank motion model parameters are learned directly from the k-space data in an unsupervised fashion. The more constrained motion model is observed to offer improved recovery compared to current motion-resolved and motion-compensated algorithms for free-breathing 3D cine MRI.", "date": "2025-05-06", "categories": []}
{"id": "2506.01394", "title": "NTIRE 2025 the 2nd Restore Any Image Model (RAIM) in the Wild Challenge", "abstract": "In this paper, we present a comprehensive overview of the NTIRE 2025 challenge on the 2nd Restore Any Image Model (RAIM) in the Wild. This challenge established a new benchmark for real-world image restoration, featuring diverse scenarios with and without reference ground truth. Participants were tasked with restoring real-captured images suffering from complex and unknown degradations, where both perceptual quality and fidelity were critically evaluated. The challenge comprised two tracks: (1) the low-light joint denoising and demosaicing (JDD) task, and (2) the image detail enhancement/generation task. Each track included two sub-tasks. The first sub-task involved paired data with available ground truth, enabling quantitative evaluation. The second sub-task dealt with real-world yet unpaired images, emphasizing restoration efficiency and subjective quality assessed through a comprehensive user study. In total, the challenge attracted nearly 300 registrations, with 51 teams submitting more than 600 results. The top-performing methods advanced the state of the art in image restoration and received unanimous recognition from all 20+ expert judges. The datasets used in Track 1 and Track 2 are available at https://drive.google.com/drive/folders/1Mgqve-yNcE26IIieI8lMIf-25VvZRs_J and https://drive.google.com/drive/folders/1UB7nnzLwqDZOwDmD9aT8J0KVg2ag4Qae, respectively. The official challenge pages for Track 1 and Track 2 can be found at https://codalab.lisn.upsaclay.fr/competitions/21334#learn_the_details and https://codalab.lisn.upsaclay.fr/competitions/21623#learn_the_details.", "date": "2025-06-02", "categories": []}
{"id": "2506.01947", "title": "RAW Image Reconstruction from RGB on Smartphones. NTIRE 2025 Challenge Report", "abstract": "Numerous low-level vision tasks operate in the RAW domain due to its linear properties, bit depth, and sensor designs. Despite this, RAW image datasets are scarce and more expensive to collect than the already large and public sRGB datasets. For this reason, many approaches try to generate realistic RAW images using sensor information and sRGB images. This paper covers the second challenge on RAW Reconstruction from sRGB (Reverse ISP). We aim to recover RAW sensor images from smartphones given the corresponding sRGB images without metadata and, by doing this, ``reverse\" the ISP transformation. Over 150 participants joined this NTIRE 2025 challenge and submitted efficient models. The proposed methods and benchmark establish the state-of-the-art for generating realistic RAW data.", "date": "2025-06-02", "categories": []}
{"id": "2505.24514", "title": "Digital twins enable full-reference quality assessment of photoacoustic image reconstructions", "abstract": "Quantitative comparison of the quality of photoacoustic image reconstruction algorithms remains a major challenge. No-reference image quality measures are often inadequate, but full-reference measures require access to an ideal reference image. While the ground truth is known in simulations, it is unknown in vivo, or in phantom studies, as the reference depends on both the phantom properties and the imaging system. We tackle this problem by using numerical digital twins of tissue-mimicking phantoms and the imaging system to perform a quantitative calibration to reduce the simulation gap. The contributions of this paper are two-fold: First, we use this digital-twin framework to compare multiple state-of-the-art reconstruction algorithms. Second, among these is a Fourier transform-based reconstruction algorithm for circular detection geometries, which we test on experimental data for the first time. Our results demonstrate the usefulness of digital phantom twins by enabling assessment of the accuracy of the numerical forward model and enabling comparison of image reconstruction schemes with full-reference image quality assessment. We show that the Fourier transform-based algorithm yields results comparable to those of iterative time reversal, but at a lower computational cost. All data and code are publicly available on Zenodo: https://doi.org/10.5281/zenodo.15388429.", "date": "2025-05-30", "categories": []}
{"id": "2505.23068", "title": "URWKV: Unified RWKV Model with Multi-state Perspective for Low-light Image Restoration", "abstract": "Existing low-light image enhancement (LLIE) and joint LLIE and deblurring (LLIE-deblur) models have made strides in addressing predefined degradations, yet they are often constrained by dynamically coupled degradations. To address these challenges, we introduce a Unified Receptance Weighted Key Value (URWKV) model with multi-state perspective, enabling flexible and effective degradation restoration for low-light images. Specifically, we customize the core URWKV block to perceive and analyze complex degradations by leveraging multiple intra- and inter-stage states. First, inspired by the pupil mechanism in the human visual system, we propose Luminance-adaptive Normalization (LAN) that adjusts normalization parameters based on rich inter-stage states, allowing for adaptive, scene-aware luminance modulation. Second, we aggregate multiple intra-stage states through exponential moving average approach, effectively capturing subtle variations while mitigating information loss inherent in the single-state mechanism. To reduce the degradation effects commonly associated with conventional skip connections, we propose the State-aware Selective Fusion (SSF) module, which dynamically aligns and integrates multi-state features across encoder stages, selectively fusing contextual information. In comparison to state-of-the-art models, our URWKV model achieves superior performance on various benchmarks, while requiring significantly fewer parameters and computational resources.", "date": "2025-05-29", "categories": []}
{"id": "2505.23180", "title": "Proximal Algorithm Unrolling: Flexible and Efficient Reconstruction Networks for Single-Pixel Imaging", "abstract": "Deep-unrolling and plug-and-play (PnP) approaches have become the de-facto standard solvers for single-pixel imaging (SPI) inverse problem. PnP approaches, a class of iterative algorithms where regularization is implicitly performed by an off-the-shelf deep denoiser, are flexible for varying compression ratios (CRs) but are limited in reconstruction accuracy and speed. Conversely, unrolling approaches, a class of multi-stage neural networks where a truncated iterative optimization process is transformed into an end-to-end trainable network, typically achieve better accuracy with faster inference but require fine-tuning or even retraining when CR changes. In this paper, we address the challenge of integrating the strengths of both classes of solvers. To this end, we design an efficient deep image restorer (DIR) for the unrolling of HQS (half quadratic splitting) and ADMM (alternating direction method of multipliers). More importantly, a general proximal trajectory (PT) loss function is proposed to train HQS/ADMM-unrolling networks such that learned DIR approximates the proximal operator of an ideal explicit restoration regularizer. Extensive experiments demonstrate that, the resulting proximal unrolling networks can not only flexibly handle varying CRs with a single model like PnP algorithms, but also outperform previous CR-specific unrolling networks in both reconstruction accuracy and speed. Source codes and models are available at https://github.com/pwangcs/ProxUnroll.", "date": "2025-05-29", "categories": []}
{"id": "2505.23594", "title": "Multilook Coherent Imaging: Theoretical Guarantees and Algorithms", "abstract": "Multilook coherent imaging is a widely used technique in applications such as digital holography, ultrasound imaging, and synthetic aperture radar. A central challenge in these systems is the presence of multiplicative noise, commonly known as speckle, which degrades image quality. Despite the widespread use of coherent imaging systems, their theoretical foundations remain relatively underexplored. In this paper, we study both the theoretical and algorithmic aspects of likelihood-based approaches for multilook coherent imaging, providing a rigorous framework for analysis and method development. Our theoretical contributions include establishing the first theoretical upper bound on the Mean Squared Error (MSE) of the maximum likelihood estimator under the deep image prior hypothesis. Our results capture the dependence of MSE on the number of parameters in the deep image prior, the number of looks, the signal dimension, and the number of measurements per look. On the algorithmic side, we employ projected gradient descent (PGD) as an efficient method for computing the maximum likelihood solution. Furthermore, we introduce two key ideas to enhance the practical performance of PGD. First, we incorporate the Newton-Schulz algorithm to compute matrix inverses within the PGD iterations, significantly reducing computational complexity. Second, we develop a bagging strategy to mitigate projection errors introduced during PGD updates. We demonstrate that combining these techniques with PGD yields state-of-the-art performance. Our code is available at https://github.com/Computational-Imaging-RU/Bagged-DIP-Speckle.", "date": "2025-05-29", "categories": []}
{"id": "2505.21530", "title": "High-Fidelity Functional Ultrasound Reconstruction via A Visual Auto-Regressive Framework", "abstract": "Functional ultrasound (fUS) imaging provides exceptional spatiotemporal resolution for neurovascular mapping, yet its practical application is significantly hampered by critical challenges. Foremost among these are data scarcity, arising from ethical considerations and signal degradation through the cranium, which collectively limit dataset diversity and compromise the fairness of downstream machine learning models.", "date": "2025-05-23", "categories": []}
{"id": "2505.21532", "title": "EvidenceMoE: A Physics-Guided Mixture-of-Experts with Evidential Critics for Advancing Fluorescence Light Detection and Ranging in Scattering Media", "abstract": "Fluorescence LiDAR (FLiDAR), a Light Detection and Ranging (LiDAR) technology employed for distance and depth estimation across medical, automotive, and other fields, encounters significant computational challenges in scattering media. The complex nature of the acquired FLiDAR signal, particularly in such environments, makes isolating photon time-of-flight (related to target depth) and intrinsic fluorescence lifetime exceptionally difficult, thus limiting the effectiveness of current analytical and computational methodologies. To overcome this limitation, we present a Physics-Guided Mixture-of-Experts (MoE) framework tailored for specialized modeling of diverse temporal components. In contrast to the conventional MoE approaches our expert models are informed by underlying physics, such as the radiative transport equation governing photon propagation in scattering media. Central to our approach is EvidenceMoE, which integrates Evidence-Based Dirichlet Critics (EDCs). These critic models assess the reliability of each expert's output by providing per-expert quality scores and corrective feedback. A Decider Network then leverages this information to fuse expert predictions into a robust final estimate adaptively. We validate our method using realistically simulated Fluorescence LiDAR (FLiDAR) data for non-invasive cancer cell depth detection generated from photon transport models in tissue. Our framework demonstrates strong performance, achieving a normalized root mean squared error (NRMSE) of 0.030 for depth estimation and 0.074 for fluorescence lifetime.", "date": "2025-05-23", "categories": []}
{"id": "2505.21637", "title": "BaryIR: Learning Multi-Source Unified Representation in Continuous Barycenter Space for Generalizable All-in-One Image Restoration", "abstract": "Despite remarkable advances made in all-in-one image restoration (AIR) for handling different types of degradations simultaneously, existing methods remain vulnerable to out-of-distribution degradations and images, limiting their real-world applicability. In this paper, we propose a multi-source representation learning framework BaryIR, which decomposes the latent space of multi-source degraded images into a continuous barycenter space for unified feature encoding and source-specific subspaces for specific semantic encoding. Specifically, we seek the multi-source unified representation by introducing a multi-source latent optimal transport barycenter problem, in which a continuous barycenter map is learned to transport the latent representations to the barycenter space. The transport cost is designed such that the representations from source-specific subspaces are contrasted with each other while maintaining orthogonality to those from the barycenter space. This enables BaryIR to learn compact representations with unified degradation-agnostic information from the barycenter space, as well as degradation-specific semantics from source-specific subspaces, capturing the inherent geometry of multi-source data manifold for generalizable AIR. Extensive experiments demonstrate that BaryIR achieves competitive performance compared to state-of-the-art all-in-one methods. Particularly, BaryIR exhibits superior generalization ability to real-world data and unseen degradations. The code will be publicly available at https://github.com/xl-tang3/BaryIR.", "date": "2025-05-27", "categories": []}
{"id": "2505.22481", "title": "Hypothesis Testing in Imaging Inverse Problems", "abstract": "This paper proposes a framework for semantic hypothesis testing tailored to imaging inverse problems. Modern imaging methods struggle to support hypothesis testing, a core component of the scientific method that is essential for the rigorous interpretation of experiments and robust interfacing with decision-making processes. There are three main reasons why image-based hypothesis testing is challenging. First, the difficulty of using a single observation to simultaneously reconstruct an image, formulate hypotheses, and quantify their statistical significance. Second, the hypotheses encountered in imaging are mostly of semantic nature, rather than quantitative statements about pixel values. Third, it is challenging to control test error probabilities because the null and alternative distributions are often unknown. Our proposed approach addresses these difficulties by leveraging concepts from self-supervised computational imaging, vision-language models, and non-parametric hypothesis testing with e-values. We demonstrate our proposed framework through numerical experiments related to image-based phenotyping, where we achieve excellent power while robustly controlling Type I errors.", "date": "2025-05-28", "categories": []}
{"id": "2406.19239", "title": "ALMA: a mathematics-driven approach for determining tuning parameters in generalized LASSO problems, with applications to MRI", "abstract": "Magnetic Resonance Imaging (MRI) is a powerful technique employed for non-invasive in vivo visualization of internal structures. Sparsity is often deployed to accelerate the signal acquisition or overcome the presence of motion artifacts, improving the quality of image reconstruction. Image reconstruction algorithms use TV-regularized LASSO (Total Variation-regularized LASSO) to retrieve the missing information of undersampled signals, by cleaning the data of noise and while optimizing sparsity. A tuning parameter moderates the balance between these two aspects; its choice affecting the quality of the reconstructions. Currently, there is a lack of general deterministic techniques to choose these parameters, which are oftentimes manually selected and thus hinder the reliability of the reconstructions. Here, we present ALMA (Algorithm for Lagrange Multipliers Approximation), an iterative mathematics-inspired technique that computes tuning parameters for generalized LASSO problems during MRI reconstruction. We analyze quantitatively the performance of these parameters for imaging reconstructions via TV-LASSO in an MRI context on phantoms. Although our study concentrates on TV-LASSO, the techniques developed here hold significant promise for a wide array of applications. ALMA is not only adaptable to more generalized LASSO problems but is also robust to accommodate other forms of regularization beyond total variation. Moreover, it extends effectively to handle non-Cartesian sampling trajectories, broadening its utility in complex data reconstruction scenarios. More generally, ALMA provides a powerful tool for numerically solving constrained optimization problems across various disciplines, offering a versatile and impactful solution for advanced computational challenges.", "date": "2024-06-27", "categories": []}
{"id": "2503.12688", "title": "Dynamic Angle Selection in X-Ray CT: A Reinforcement Learning Approach to Optimal Stopping", "abstract": "In industrial X-ray Computed Tomography (CT), the need for rapid in-line inspection is critical. Sparse-angle tomography plays a significant role in this by reducing the required number of projections, thereby accelerating processing and conserving resources. Most existing methods aim to balance reconstruction quality and scanning time, typically relying on fixed scan durations. Adaptive adjustment of the number of angles is essential; for instance, more angles may be required for objects with complex geometries or noisier projections. The concept of optimal stopping, which dynamically adjusts this balance according to varying industrial needs, remains overlooked. Building on our previous work, we integrate optimal stopping into sequential Optimal Experimental Design (sOED) and Reinforcement Learning (RL). We propose a novel method for computing the policy gradient within the Actor-Critic framework, enabling the development of adaptive policies for informative angle selection and scan termination. Additionally, we investigate the gap between simulation and real-world applications in the context of the developed learning-based method. Our trained model, developed using synthetic data, demonstrates reliable performance when applied to experimental X-ray CT data. This approach enhances the flexibility of CT operations and expands the applicability of sparse-angle tomography in industrial settings.", "date": "2025-03-16", "categories": []}
{"id": "2505.18365", "title": "Brightness-Invariant Tracking Estimation in Tagged MRI", "abstract": "Magnetic resonance (MR) tagging is an imaging technique for noninvasively tracking tissue motion in vivo by creating a visible pattern of magnetization saturation (tags) that deforms with the tissue. Due to longitudinal relaxation and progression to steady-state, the tags and tissue brightnesses change over time, which makes tracking with optical flow methods error-prone. Although Fourier methods can alleviate these problems, they are also sensitive to brightness changes as well as spectral spreading due to motion. To address these problems, we introduce the brightness-invariant tracking estimation (BRITE) technique for tagged MRI. BRITE disentangles the anatomy from the tag pattern in the observed tagged image sequence and simultaneously estimates the Lagrangian motion. The inherent ill-posedness of this problem is addressed by leveraging the expressive power of denoising diffusion probabilistic models to represent the probabilistic distribution of the underlying anatomy and the flexibility of physics-informed neural networks to estimate biologically-plausible motion. A set of tagged MR images of a gel phantom was acquired with various tag periods and imaging flip angles to demonstrate the impact of brightness variations and to validate our method. The results show that BRITE achieves more accurate motion and strain estimates as compared to other state of the art methods, while also being resistant to tag fading.", "date": "2025-05-23", "categories": []}
{"id": "2502.08634", "title": "Rapid Whole Brain Motion-robust Mesoscale In-vivo MR Imaging using Multi-scale Implicit Neural Representation", "abstract": "High-resolution whole-brain in vivo MR imaging at mesoscale resolutions remains challenging due to long scan durations, motion artifacts, and limited signal-to-noise ratio (SNR). This study proposes Rotating-view super-resolution (ROVER)-MRI, an unsupervised framework based on multi-scale implicit neural representations (INR), enabling efficient recovery of fine anatomical details from multi-view thick-slice acquisitions. ROVER-MRI employs coordinate-based neural networks to implicitly and continuously encode image structures at multiple spatial scales, simultaneously modeling anatomical continuity and correcting inter-view motion through an integrated registration mechanism. Validation on ex-vivo monkey brain data and multiple in-vivo human datasets demonstrates substantially improved reconstruction performance compared to bicubic interpolation and state-of-the-art regularized least-squares super-resolution reconstruction (LS-SRR) with 2-fold reduction in scan time. Notably, ROVER-MRI achieves an unprecedented whole-brain in-vivo T2-weighted imaging at 180 micron isotropic resolution in only 17 minutes of scan time on a 7T scanner with 22.4% lower relative error compared to LS-SRR. We also demonstrate improved SNR using ROVER-MRI compared to a time-matched 3D GRE acquisition. Quantitative results on several datasets demonstrate better sharpness of the reconstructed images with ROVER-MRI for different super-resolution factors (5 to 11). These findings highlight ROVER-MRI's potential as a rapid, accurate, and motion-resilient mesoscale imaging solution, promising substantial advantages for neuroimaging studies.", "date": "2025-02-12", "categories": []}
{"id": "2505.17358", "title": "Repurposing Marigold for Zero-Shot Metric Depth Estimation via Defocus Blur Cues", "abstract": "Recent monocular metric depth estimation (MMDE) methods have made notable progress towards zero-shot generalization. However, they still exhibit a significant performance drop on out-of-distribution datasets. We address this limitation by injecting defocus blur cues at inference time into Marigold, a \\textit{pre-trained} diffusion model for zero-shot, scale-invariant monocular depth estimation (MDE). Our method effectively turns Marigold into a metric depth predictor in a training-free manner. To incorporate defocus cues, we capture two images with a small and a large aperture from the same viewpoint. To recover metric depth, we then optimize the metric depth scaling parameters and the noise latents of Marigold at inference time using gradients from a loss function based on the defocus-blur image formation model. We compare our method against existing state-of-the-art zero-shot MMDE methods on a self-collected real dataset, showing quantitative and qualitative improvements.", "date": "2025-05-23", "categories": []}
{"id": "2505.16373", "title": "PCMamba: Physics-Informed Cross-Modal State Space Model for Dual-Camera Compressive Hyperspectral Imaging", "abstract": "Panchromatic (PAN) -assisted Dual-Camera Compressive Hyperspectral Imaging (DCCHI) is a key technology in snapshot hyperspectral imaging. Existing research primarily focuses on exploring spectral information from 2D compressive measurements and spatial information from PAN images in an explicit manner, leading to a bottleneck in HSI reconstruction. Various physical factors, such as temperature, emissivity, and multiple reflections between objects, play a critical role in the process of a sensor acquiring hyperspectral thermal signals. Inspired by this, we attempt to investigate the interrelationships between physical properties to provide deeper theoretical insights for HSI reconstruction. In this paper, we propose a Physics-Informed Cross-Modal State Space Model Network (PCMamba) for DCCHI, which incorporates the forward physical imaging process of HSI into the linear complexity of Mamba to facilitate lightweight and high-quality HSI reconstruction. Specifically, we analyze the imaging process of hyperspectral thermal signals to enable the network to disentangle the three key physical properties-temperature, emissivity, and texture. By fully exploiting the potential information embedded in 2D measurements and PAN images, the HSIs are reconstructed through a physics-driven synthesis process. Furthermore, we design a Cross-Modal Scanning Mamba Block (CSMB) that introduces inter-modal pixel-wise interaction with positional inductive bias by cross-scanning the backbone features and PAN features. Extensive experiments conducted on both real and simulated datasets demonstrate that our method significantly outperforms SOTA methods in both quantitative and qualitative metrics.", "date": "2025-05-22", "categories": []}
{"id": "2505.15777", "title": "Projection-Based Correction for Enhancing Deep Inverse Networks", "abstract": "Deep learning-based models have demonstrated remarkable success in solving illposed inverse problems; however, many fail to strictly adhere to the physical constraints imposed by the measurement process. In this work, we introduce a projection-based correction method to enhance the inference of deep inverse networks by ensuring consistency with the forward model. Specifically, given an initial estimate from a learned reconstruction network, we apply a projection step that constrains the solution to lie within the valid solution space of the inverse problem. We theoretically demonstrate that if the recovery model is a well-trained deep inverse network, the solution can be decomposed into range-space and null-space components, where the projection-based correction reduces to an identity transformation. Extensive simulations and experiments validate the proposed method, demonstrating improved reconstruction accuracy across diverse inverse problems and deep network architectures.", "date": "2025-05-21", "categories": []}
{"id": "2502.17177", "title": "Joint multiband deconvolution for Euclid and Vera C. Rubin images", "abstract": "With the advent of surveys like Euclid and Vera C. Rubin, astrophysicists will have access to both deep, high-resolution images and multiband images. However, these two types are not simultaneously available in any single dataset. It is therefore vital to devise image deconvolution algorithms that exploit the best of both worlds and that can jointly analyze datasets spanning a range of resolutions and wavelengths. In this work we introduce a novel multiband deconvolution technique aimed at improving the resolution of ground-based astronomical images by leveraging higher-resolution space-based observations. The method capitalizes on the fortunate fact that the Rubin $r$, $i$, and $z$ bands lie within the Euclid VIS band. The algorithm jointly de-convolves all the data to convert the $r$-, $i$-, and $z$-band Rubin images to the resolution of Euclid by leveraging the correlations between the different bands. We also investigate the performance of deep-learning-based denoising with DRUNet to further improve the results. We illustrate the effectiveness of our method in terms of resolution and morphology recovery, flux preservation, and generalization to different noise levels. This approach extends beyond the specific Euclid-Rubin combination, offering a versatile solution to improving the resolution of ground-based images in multiple photometric bands by jointly using any space-based images with overlapping filters.", "date": "2025-02-24", "categories": []}
{"id": "2505.13579", "title": "Learning Wavelet-Sparse FDK for 3D Cone-Beam CT Reconstruction", "abstract": "Cone-Beam Computed Tomography (CBCT) is essential in medical imaging, and the Feldkamp-Davis-Kress (FDK) algorithm is a popular choice for reconstruction due to its efficiency. However, FDK is susceptible to noise and artifacts. While recent deep learning methods offer improved image quality, they often increase computational complexity and lack the interpretability of traditional methods. In this paper, we introduce an enhanced FDK-based neural network that maintains the classical algorithm's interpretability by selectively integrating trainable elements into the cosine weighting and filtering stages. Recognizing the challenge of a large parameter space inherent in 3D CBCT data, we leverage wavelet transformations to create sparse representations of the cosine weights and filters. This strategic sparsification reduces the parameter count by $93.75\\%$ without compromising performance, accelerates convergence, and importantly, maintains the inference computational cost equivalent to the classical FDK algorithm. Our method not only ensures volumetric consistency and boosts robustness to noise, but is also designed for straightforward integration into existing CT reconstruction pipelines. This presents a pragmatic enhancement that can benefit clinical applications, particularly in environments with computational limitations.", "date": "2025-05-19", "categories": []}
{"id": "2505.14560", "title": "Neural Inverse Scattering with Score-based Regularization", "abstract": "Inverse scattering is a fundamental challenge in many imaging applications, ranging from microscopy to remote sensing. Solving this problem often requires jointly estimating two unknowns -- the image and the scattering field inside the object -- necessitating effective image prior to regularize the inference. In this paper, we propose a regularized neural field (NF) approach which integrates the denoising score function used in score-based generative models. The neural field formulation offers convenient flexibility to performing joint estimation, while the denoising score function imposes the rich structural prior of images. Our results on three high-contrast simulated objects show that the proposed approach yields a better imaging quality compared to the state-of-the-art NF approach, where regularization is based on total variation.", "date": "2025-05-20", "categories": []}
{"id": "2505.14634", "title": "A General Framework for Group Sparsity in Hyperspectral Unmixing Using Endmember Bundles", "abstract": "Due to low spatial resolution, hyperspectral data often consists of mixtures of contributions from multiple materials. This limitation motivates the task of hyperspectral unmixing (HU), a fundamental problem in hyperspectral imaging. HU aims to identify the spectral signatures (\\textit{endmembers}) of the materials present in an observed scene, along with their relative proportions (\\textit{fractional abundance}) in each pixel. A major challenge lies in the class variability in materials, which hinders accurate representation by a single spectral signature, as assumed in the conventional linear mixing model. Moreover, To address this issue, we propose using group sparsity after representing each material with a set of spectral signatures, known as endmember bundles, where each group corresponds to a specific material. In particular, we develop a bundle-based framework that can enforce either inter-group sparsity or sparsity within and across groups (SWAG) on the abundance coefficients. Furthermore, our framework offers the flexibility to incorporate a variety of sparsity-promoting penalties, among which the transformed $\\ell_1$ (TL1) penalty is a novel regularization in the HU literature. Extensive experiments conducted on both synthetic and real hyperspectral data demonstrate the effectiveness and superiority of the proposed approaches.", "date": "2025-05-20", "categories": []}
{"id": "2408.02866", "title": "Back-Projection Diffusion: Solving the Wideband Inverse Scattering Problem with Diffusion Models", "abstract": "We present Wideband Back-Projection Diffusion, an end-to-end probabilistic framework for approximating the posterior distribution induced by the inverse scattering map from wideband scattering data. This framework produces highly accurate reconstructions, leveraging conditional diffusion models to draw samples, and also honors the symmetries of the underlying physics of wave-propagation. The procedure is factored into two steps: the first step, inspired by the filtered back-propagation formula, transforms data into a physics-based latent representation, while the second step learns a conditional score function conditioned on this latent representation. These two steps individually obey their associated symmetries and are amenable to compression by imposing the rank structure found in the filtered back-projection formula. Empirically, our framework has both low sample and computational complexity, with its number of parameters scaling only sub-linearly with the target resolution, and has stable training dynamics. It provides sharp reconstructions effortlessly and is capable of recovering even sub-Nyquist features in the multiple-scattering regime.", "date": "2024-08-05", "categories": []}
{"id": "2505.11720", "title": "UGoDIT: Unsupervised Group Deep Image Prior Via Transferable Weights", "abstract": "Recent advances in data-centric deep generative models have led to significant progress in solving inverse imaging problems. However, these models (e.g., diffusion models (DMs)) typically require large amounts of fully sampled (clean) training data, which is often impractical in medical and scientific settings such as dynamic imaging.\n  On the other hand, training-data-free approaches like the Deep Image Prior (DIP) do not require clean ground-truth images but suffer from noise overfitting and can be computationally expensive as the network parameters need to be optimized for each measurement set independently. Moreover, DIP-based methods often overlook the potential of learning a prior using a small number of sub-sampled measurements (or degraded images) available during training. In this paper, we propose UGoDIT, an Unsupervised Group DIP via Transferable weights, designed for the low-data regime where only a very small number, M, of sub-sampled measurement vectors are available during training. Our method learns a set of transferable weights by optimizing a shared encoder and M disentangled decoders. At test time, we reconstruct the unseen degraded image using a DIP network, where part of the parameters are fixed to the learned weights, while the remaining are optimized to enforce measurement consistency. We evaluate UGoDIT on both medical (multi-coil MRI) and natural (super resolution and non-linear deblurring) image recovery tasks under various settings. Compared to recent standalone DIP methods, UGoDIT provides accelerated convergence and notable improvement in reconstruction quality. Furthermore, our method achieves performance competitive with SOTA DM-based and supervised approaches, despite not requiring large amounts of clean training data.", "date": "2025-05-16", "categories": []}
{"id": "2505.11800", "title": "Self-Learning Hyperspectral and Multispectral Image Fusion via Adaptive Residual Guided Subspace Diffusion Model", "abstract": "Hyperspectral and multispectral image (HSI-MSI) fusion involves combining a low-resolution hyperspectral image (LR-HSI) with a high-resolution multispectral image (HR-MSI) to generate a high-resolution hyperspectral image (HR-HSI). Most deep learning-based methods for HSI-MSI fusion rely on large amounts of hyperspectral data for supervised training, which is often scarce in practical applications. In this paper, we propose a self-learning Adaptive Residual Guided Subspace Diffusion Model (ARGS-Diff), which only utilizes the observed images without any extra training data. Specifically, as the LR-HSI contains spectral information and the HR-MSI contains spatial information, we design two lightweight spectral and spatial diffusion models to separately learn the spectral and spatial distributions from them. Then, we use these two models to reconstruct HR-HSI from two low-dimensional components, i.e, the spectral basis and the reduced coefficient, during the reverse diffusion process. Furthermore, we introduce an Adaptive Residual Guided Module (ARGM), which refines the two components through a residual guided function at each sampling step, thereby stabilizing the sampling process. Extensive experimental results demonstrate that ARGS-Diff outperforms existing state-of-the-art methods in terms of both performance and computational efficiency in the field of HSI-MSI fusion. Code is available at https://github.com/Zhu1116/ARGS-Diff.", "date": "2025-05-17", "categories": []}
{"id": "2505.11913", "title": "Joint Manifold Learning and Optimal Transport for Dynamic Imaging", "abstract": "Dynamic imaging is critical for understanding and visualizing dynamic biological processes in medicine and cell biology. These applications often encounter the challenge of a limited amount of time series data and time points, which hinders learning meaningful patterns. Regularization methods provide valuable prior knowledge to address this challenge, enabling the extraction of relevant information despite the scarcity of time-series data and time points. In particular, low-dimensionality assumptions on the image manifold address sample scarcity, while time progression models, such as optimal transport (OT), provide priors on image development to mitigate the lack of time points. Existing approaches using low-dimensionality assumptions disregard a temporal prior but leverage information from multiple time series. OT-prior methods, however, incorporate the temporal prior but regularize only individual time series, ignoring information from other time series of the same image modality. In this work, we investigate the effect of integrating a low-dimensionality assumption of the underlying image manifold with an OT regularizer for time-evolving images. In particular, we propose a latent model representation of the underlying image manifold and promote consistency between this representation, the time series data, and the OT prior on the time-evolving images. We discuss the advantages of enriching OT interpolations with latent models and integrating OT priors into latent models.", "date": "2025-05-17", "categories": []}
{"id": "2505.12089", "title": "NTIRE 2025 Challenge on Efficient Burst HDR and Restoration: Datasets, Methods, and Results", "abstract": "This paper reviews the NTIRE 2025 Efficient Burst HDR and Restoration Challenge, which aims to advance efficient multi-frame high dynamic range (HDR) and restoration techniques. The challenge is based on a novel RAW multi-frame fusion dataset, comprising nine noisy and misaligned RAW frames with various exposure levels per scene. Participants were tasked with developing solutions capable of effectively fusing these frames while adhering to strict efficiency constraints: fewer than 30 million model parameters and a computational budget under 4.0 trillion FLOPs. A total of 217 participants registered, with six teams finally submitting valid solutions. The top-performing approach achieved a PSNR of 43.22 dB, showcasing the potential of novel methods in this domain. This paper provides a comprehensive overview of the challenge, compares the proposed solutions, and serves as a valuable reference for researchers and practitioners in efficient burst HDR and restoration.", "date": "2025-05-17", "categories": []}
{"id": "2505.12630", "title": "Degradation-Aware Feature Perturbation for All-in-One Image Restoration", "abstract": "All-in-one image restoration aims to recover clear images from various degradation types and levels with a unified model. Nonetheless, the significant variations among degradation types present challenges for training a universal model, often resulting in task interference, where the gradient update directions of different tasks may diverge due to shared parameters. To address this issue, motivated by the routing strategy, we propose DFPIR, a novel all-in-one image restorer that introduces Degradation-aware Feature Perturbations(DFP) to adjust the feature space to align with the unified parameter space. In this paper, the feature perturbations primarily include channel-wise perturbations and attention-wise perturbations. Specifically, channel-wise perturbations are implemented by shuffling the channels in high-dimensional space guided by degradation types, while attention-wise perturbations are achieved through selective masking in the attention space. To achieve these goals, we propose a Degradation-Guided Perturbation Block (DGPB) to implement these two functions, positioned between the encoding and decoding stages of the encoder-decoder architecture. Extensive experimental results demonstrate that DFPIR achieves state-of-the-art performance on several all-in-one image restoration tasks including image denoising, image dehazing, image deraining, motion deblurring, and low-light image enhancement. Our codes are available at https://github.com/TxpHome/DFPIR.", "date": "2025-05-19", "categories": []}
{"id": "2505.13250", "title": "Joint Depth and Reflectivity Estimation using Single-Photon LiDAR", "abstract": "Single-Photon Light Detection and Ranging (SP-LiDAR is emerging as a leading technology for long-range, high-precision 3D vision tasks. In SP-LiDAR, timestamps encode two complementary pieces of information: pulse travel time (depth) and the number of photons reflected by the object (reflectivity). Existing SP-LiDAR reconstruction methods typically recover depth and reflectivity separately or sequentially use one modality to estimate the other. Moreover, the conventional 3D histogram construction is effective mainly for slow-moving or stationary scenes. In dynamic scenes, however, it is more efficient and effective to directly process the timestamps. In this paper, we introduce an estimation method to simultaneously recover both depth and reflectivity in fast-moving scenes. We offer two contributions: (1) A theoretical analysis demonstrating the mutual correlation between depth and reflectivity and the conditions under which joint estimation becomes beneficial. (2) A novel reconstruction method, \"SPLiDER\", which exploits the shared information to enhance signal recovery. On both synthetic and real SP-LiDAR data, our method outperforms existing approaches, achieving superior joint reconstruction quality.", "date": "2025-05-19", "categories": []}
{"id": "2505.11062", "title": "HSRMamba: Efficient Wavelet Stripe State Space Model for Hyperspectral Image Super-Resolution", "abstract": "Single hyperspectral image super-resolution (SHSR) aims to restore high-resolution images from low-resolution hyperspectral images. Recently, the Visual Mamba model has achieved an impressive balance between performance and computational efficiency. However, due to its 1D scanning paradigm, the model may suffer from potential artifacts during image generation. To address this issue, we propose HSRMamba. While maintaining the computational efficiency of Visual Mamba, we introduce a strip-based scanning scheme to effectively reduce artifacts from global unidirectional scanning. Additionally, HSRMamba uses wavelet decomposition to alleviate modal conflicts between high-frequency spatial features and low-frequency spectral features, further improving super-resolution performance. Extensive experiments show that HSRMamba not only excels in reducing computational load and model size but also outperforms existing methods, achieving state-of-the-art results.", "date": "2025-05-16", "categories": []}
{"id": "2505.11267", "title": "Equal is Not Always Fair: A New Perspective on Hyperspectral Representation Non-Uniformity", "abstract": "Hyperspectral image (HSI) representation is fundamentally challenged by pervasive non-uniformity, where spectral dependencies, spatial continuity, and feature efficiency exhibit complex and often conflicting behaviors. Most existing models rely on a unified processing paradigm that assumes homogeneity across dimensions, leading to suboptimal performance and biased representations. To address this, we propose FairHyp, a fairness-directed framework that explicitly disentangles and resolves the threefold non-uniformity through cooperative yet specialized modules. We introduce a Runge-Kutta-inspired spatial variability adapter to restore spatial coherence under resolution discrepancies, a multi-receptive field convolution module with sparse-aware refinement to enhance discriminative features while respecting inherent sparsity, and a spectral-context state space model that captures stable and long-range spectral dependencies via bidirectional Mamba scanning and statistical aggregation. Unlike one-size-fits-all solutions, FairHyp achieves dimension-specific adaptation while preserving global consistency and mutual reinforcement. This design is grounded in the view that non-uniformity arises from the intrinsic structure of HSI representations, rather than any particular task setting. To validate this, we apply FairHyp across four representative tasks including classification, denoising, super-resolution, and inpaintin, demonstrating its effectiveness in modeling a shared structural flaw. Extensive experiments show that FairHyp consistently outperforms state-of-the-art methods under varied imaging conditions. Our findings redefine fairness as a structural necessity in HSI modeling and offer a new paradigm for balancing adaptability, efficiency, and fidelity in high-dimensional vision tasks.", "date": "2025-05-16", "categories": []}
{"id": "2505.10420", "title": "Learned Lightweight Smartphone ISP with Unpaired Data", "abstract": "The Image Signal Processor (ISP) is a fundamental component in modern smartphone cameras responsible for conversion of RAW sensor image data to RGB images with a strong focus on perceptual quality. Recent work highlights the potential of deep learning approaches and their ability to capture details with a quality increasingly close to that of professional cameras. A difficult and costly step when developing a learned ISP is the acquisition of pixel-wise aligned paired data that maps the raw captured by a smartphone camera sensor to high-quality reference images. In this work, we address this challenge by proposing a novel training method for a learnable ISP that eliminates the need for direct correspondences between raw images and ground-truth data with matching content. Our unpaired approach employs a multi-term loss function guided by adversarial training with multiple discriminators processing feature maps from pre-trained networks to maintain content structure while learning color and texture characteristics from the target RGB dataset. Using lightweight neural network architectures suitable for mobile devices as backbones, we evaluated our method on the Zurich RAW to RGB and Fujifilm UltraISP datasets. Compared to paired training methods, our unpaired learning strategy shows strong potential and achieves high fidelity across multiple evaluation metrics. The code and pre-trained models are available at https://github.com/AndreiiArhire/Learned-Lightweight-Smartphone-ISP-with-Unpaired-Data .", "date": "2025-05-15", "categories": []}
{"id": "2505.08843", "title": "Total Variation-Based Image Decomposition and Denoising for Microscopy Images", "abstract": "Experimentally acquired microscopy images are unavoidably affected by the presence of noise and other unwanted signals, which degrade their quality and might hide relevant features. With the recent increase in image acquisition rate, modern denoising and restoration solutions become necessary. This study focuses on image decomposition and denoising of microscopy images through a workflow based on total variation (TV), addressing images obtained from various microscopy techniques, including atomic force microscopy (AFM), scanning tunneling microscopy (STM), and scanning electron microscopy (SEM). Our approach consists in restoring an image by extracting its unwanted signal components and subtracting them from the raw one, or by denoising it. We evaluate the performance of TV-$L^1$, Huber-ROF, and TGV-$L^1$ in achieving this goal in distinct study cases. Huber-ROF proved to be the most flexible one, while TGV-$L^1$ is the most suitable for denoising. Our results suggest a wider applicability of this method in microscopy, restricted not only to STM, AFM, and SEM images. The Python code used for this study is publicly available as part of AiSurf. It is designed to be integrated into experimental workflows for image acquisition or can be used to denoise previously acquired images.", "date": "2025-05-13", "categories": []}
{"id": "2505.09528", "title": "Conformal Bounds on Full-Reference Image Quality for Imaging Inverse Problems", "abstract": "In imaging inverse problems, we would like to know how close the recovered image is to the true image in terms of full-reference image quality (FRIQ) metrics like PSNR, SSIM, LPIPS, etc. This is especially important in safety-critical applications like medical imaging, where knowing that, say, the SSIM was poor could potentially avoid a costly misdiagnosis. But since we don't know the true image, computing FRIQ is non-trivial. In this work, we combine conformal prediction with approximate posterior sampling to construct bounds on FRIQ that are guaranteed to hold up to a user-specified error probability. We demonstrate our approach on image denoising and accelerated magnetic resonance imaging (MRI) problems. Code is available at https://github.com/jwen307/quality_uq.", "date": "2025-05-14", "categories": []}
{"id": "2504.02697", "title": "Learning Phase Distortion with Selective State Space Models for Video Turbulence Mitigation", "abstract": "Atmospheric turbulence is a major source of image degradation in long-range imaging systems. Although numerous deep learning-based turbulence mitigation (TM) methods have been proposed, many are slow, memory-hungry, and do not generalize well. In the spatial domain, methods based on convolutional operators have a limited receptive field, so they cannot handle a large spatial dependency required by turbulence. In the temporal domain, methods relying on self-attention can, in theory, leverage the lucky effects of turbulence, but their quadratic complexity makes it difficult to scale to many frames. Traditional recurrent aggregation methods face parallelization challenges.\n  In this paper, we present a new TM method based on two concepts: (1) A turbulence mitigation network based on the Selective State Space Model (MambaTM). MambaTM provides a global receptive field in each layer across spatial and temporal dimensions while maintaining linear computational complexity. (2) Learned Latent Phase Distortion (LPD). LPD guides the state space model. Unlike classical Zernike-based representations of phase distortion, the new LPD map uniquely captures the actual effects of turbulence, significantly improving the model's capability to estimate degradation by reducing the ill-posedness. Our proposed method exceeds current state-of-the-art networks on various synthetic and real-world TM benchmarks with significantly faster inference speed.", "date": "2025-04-03", "categories": []}
{"id": "2402.17410", "title": "Image space formalism of convolutional neural networks for k-space interpolation", "abstract": "Purpose: Noise resilience in image reconstructions by scan-specific robust artificial neural networks for k-space interpolation (RAKI) is linked to nonlinear activations in k-space. To gain a deeper understanding of this relationship, an image space formalism of RAKI is introduced for analyzing noise propagation analytically, identifying and characterizing image reconstruction features and to describe the role of nonlinear activations in a human readable manner. Methods: The image space formalism for RAKI inference is employed by expressing nonlinear activations in k-space as element-wise multiplications with activation masks, which transform into convolutions in image space. Jacobians of the de-aliased, coil-combined image relative to the aliased coil images can be expressed algebraically, and thus, the noise amplification is quantified analytically (g-factor maps). We analyze the role of nonlinearity for noise resilience by controlling the degree of nonlinearity in the reconstruction model via the negative slope parameter in leaky ReLU. Results: The analytical g-factor maps correspond with those obtained from Monte Carlo simulations and from an auto differentiation approach for in vivo brain images. Apparent blurring and contrast loss artifacts are identified as implications of enhanced noise resilience. These residual artifacts can be traded against noise resilience by adjusting the degree of nonlinearity in the model (Tikhonov-like regularization) in case of limited training data. The inspection of image space activations reveals an autocorrelation pattern leading to a potential center artifact. Conclusion: The image space formalism of RAKI provides the means for analytical quantitative noisepropagation analysis and human-readable visualization of the effects of the nonlinear activation functions in k-space.", "date": "2024-02-27", "categories": []}
{"id": "2505.05488", "title": "From Events to Enhancement: A Survey on Event-Based Imaging Technologies", "abstract": "Event cameras offering high dynamic range and low latency have emerged as disruptive technologies in imaging. Despite growing research on leveraging these benefits for different imaging tasks, a comprehensive study of recently advances and challenges are still lacking. This limits the broader understanding of how to utilize events in universal imaging applications. In this survey, we first introduce a physical model and the characteristics of different event sensors as the foundation. Following this, we highlight the advancement and interaction of image/video enhancement tasks with events. Additionally, we explore advanced tasks, which capture richer light information with events, \\eg~light field estimation, multi-view generation, and photometric. Finally, we discuss new challenges and open questions offering a perspective for this rapidly evolving field. More continuously updated resources are at this link: https://github.com/yunfanLu/Awesome-Event-Imaging", "date": "2025-04-30", "categories": []}
{"id": "2505.05643", "title": "UltraGauss: Ultrafast Gaussian Reconstruction of 3D Ultrasound Volumes", "abstract": "Ultrasound imaging is widely used due to its safety, affordability, and real-time capabilities, but its 2D interpretation is highly operator-dependent, leading to variability and increased cognitive demand. 2D-to-3D reconstruction mitigates these challenges by providing standardized volumetric views, yet existing methods are often computationally expensive, memory-intensive, or incompatible with ultrasound physics. We introduce UltraGauss: the first ultrasound-specific Gaussian Splatting framework, extending view synthesis techniques to ultrasound wave propagation. Unlike conventional perspective-based splatting, UltraGauss models probe-plane intersections in 3D, aligning with acoustic image formation. We derive an efficient rasterization boundary formulation for GPU parallelization and introduce a numerically stable covariance parametrization, improving computational efficiency and reconstruction accuracy. On real clinical ultrasound data, UltraGauss achieves state-of-the-art reconstructions in 5 minutes, and reaching 0.99 SSIM within 20 minutes on a single GPU. A survey of expert clinicians confirms UltraGauss' reconstructions are the most realistic among competing methods. Our CUDA implementation will be released upon publication.", "date": "2025-05-08", "categories": []}
{"id": "2505.04836", "title": "Integrated Image Reconstruction and Target Recognition based on Deep\n  Learning Technique", "abstract": "  Computational microwave imaging (CMI) has gained attention as an alternative\ntechnique for conventional microwave imaging techniques, addressing their\nlimitations such as hardware-intensive physical layer and slow data collection\nacquisition speed to name a few. Despite these advantages, CMI still encounters\nnotable computational bottlenecks, especially during the image reconstruction\nstage. In this setting, both image recovery and object classification present\nsignificant processing demands. To address these challenges, our previous work\nintroduced ClassiGAN, which is a generative deep learning model designed to\nsimultaneously reconstruct images and classify targets using only\nback-scattered signals. In this study, we build upon that framework by\nincorporating attention gate modules into ClassiGAN. These modules are intended\nto refine feature extraction and improve the identification of relevant\ninformation. By dynamically focusing on important features and suppressing\nirrelevant ones, the attention mechanism enhances the overall model\nperformance. The proposed architecture, named Att-ClassiGAN, significantly\nreduces the reconstruction time compared to traditional CMI approaches.\nFurthermore, it outperforms current advanced methods, delivering improved\nNormalized Mean Squared Error (NMSE), higher Structural Similarity Index\n(SSIM), and better classification outcomes for the reconstructed targets.\n", "date": "2025-05-07", "categories": []}
{"id": "2505.04369", "title": "WDMamba: When Wavelet Degradation Prior Meets Vision Mamba for Image\n  Dehazing", "abstract": "  In this paper, we reveal a novel haze-specific wavelet degradation prior\nobserved through wavelet transform analysis, which shows that haze-related\ninformation predominantly resides in low-frequency components. Exploiting this\ninsight, we propose a novel dehazing framework, WDMamba, which decomposes the\nimage dehazing task into two sequential stages: low-frequency restoration\nfollowed by detail enhancement. This coarse-to-fine strategy enables WDMamba to\neffectively capture features specific to each stage of the dehazing process,\nresulting in high-quality restored images. Specifically, in the low-frequency\nrestoration stage, we integrate Mamba blocks to reconstruct global structures\nwith linear complexity, efficiently removing overall haze and producing a\ncoarse restored image. Thereafter, the detail enhancement stage reinstates\nfine-grained information that may have been overlooked during the previous\nphase, culminating in the final dehazed output. Furthermore, to enhance detail\nretention and achieve more natural dehazing, we introduce a self-guided\ncontrastive regularization during network training. By utilizing the coarse\nrestored output as a hard negative example, our model learns more\ndiscriminative representations, substantially boosting the overall dehazing\nperformance. Extensive evaluations on public dehazing benchmarks demonstrate\nthat our method surpasses state-of-the-art approaches both qualitatively and\nquantitatively. Code is available at https://github.com/SunJ000/WDMamba.\n", "date": "2025-05-07", "categories": []}
{"id": "2505.03463", "title": "Nonperiodic dynamic CT reconstruction using backward-warping INR with\n  regularization of diffeomorphism (BIRD)", "abstract": "  Dynamic computed tomography (CT) reconstruction faces significant challenges\nin addressing motion artifacts, particularly for nonperiodic rapid movements\nsuch as cardiac imaging with fast heart rates. Traditional methods struggle\nwith the extreme limited-angle problems inherent in nonperiodic cases. Deep\nlearning methods have improved performance but face generalization challenges.\nRecent implicit neural representation (INR) techniques show promise through\nself-supervised deep learning, but have critical limitations: computational\ninefficiency due to forward-warping modeling, difficulty balancing DVF\ncomplexity with anatomical plausibility, and challenges in preserving fine\ndetails without additional patient-specific pre-scans. This paper presents a\nnovel INR-based framework, BIRD, for nonperiodic dynamic CT reconstruction. It\naddresses these challenges through four key contributions: (1) backward-warping\ndeformation that enables direct computation of each dynamic voxel with\nsignificantly reduced computational cost, (2) diffeomorphism-based DVF\nregularization that ensures anatomically plausible deformations while\nmaintaining representational capacity, (3) motion-compensated analytical\nreconstruction that enhances fine details without requiring additional\npre-scans, and (4) dimensional-reduction design for efficient 4D coordinate\nencoding. Through various simulations and practical studies, including digital\nand physical phantoms and retrospective patient data, we demonstrate the\neffectiveness of our approach for nonperiodic dynamic CT reconstruction with\nenhanced details and reduced motion artifacts. The proposed framework enables\nmore accurate dynamic CT reconstruction with potential clinical applications,\nsuch as one-beat cardiac reconstruction, cinematic image sequences for\nfunctional imaging, and motion artifact reduction in conventional CT scans.\n", "date": "2025-05-06", "categories": []}
{"id": "2404.02225", "title": "CHOSEN: Contrastive Hypothesis Selection for Multi-View Depth Refinement", "abstract": "  We propose CHOSEN, a simple yet flexible, robust and effective multi-view\ndepth refinement framework. It can be employed in any existing multi-view\nstereo pipeline, with straightforward generalization capability for different\nmulti-view capture systems such as camera relative positioning and lenses.\nGiven an initial depth estimation, CHOSEN iteratively re-samples and selects\nthe best hypotheses, and automatically adapts to different metric or intrinsic\nscales determined by the capture system. The key to our approach is the\napplication of contrastive learning in an appropriate solution space and a\ncarefully designed hypothesis feature, based on which positive and negative\nhypotheses can be effectively distinguished. Integrated in a simple baseline\nmulti-view stereo pipeline, CHOSEN delivers impressive quality in terms of\ndepth and normal accuracy compared to many current deep learning based\nmulti-view stereo pipelines.\n", "date": "2024-04-02", "categories": []}
{"id": "2409.09779", "title": "Underwater Image Enhancement via Dehazing and Color Restoration", "abstract": "  Underwater visual imaging is crucial for marine engineering, but it suffers\nfrom low contrast, blurriness, and color degradation, which hinders downstream\nanalysis. Existing underwater image enhancement methods often treat the haze\nand color cast as a unified degradation process, neglecting their inherent\nindependence while overlooking their synergistic relationship. To overcome this\nlimitation, we propose a Vision Transformer (ViT)-based network (referred to as\nWaterFormer) to improve underwater image quality. WaterFormer contains three\nmajor components: a dehazing block (DehazeFormer Block) to capture the\nself-correlated haze features and extract deep-level features, a Color\nRestoration Block (CRB) to capture self-correlated color cast features, and a\nChannel Fusion Block (CFB) that dynamically integrates these decoupled features\nto achieve comprehensive enhancement. To ensure authenticity, a soft\nreconstruction layer based on the underwater imaging physics model is included.\nFurther, a Chromatic Consistency Loss and Sobel Color Loss are designed to\nrespectively preserve color fidelity and enhance structural details during\nnetwork training. Comprehensive experimental results demonstrate that\nWaterFormer outperforms other state-of-the-art methods in enhancing underwater\nimages.\n", "date": "2024-09-15", "categories": []}
{"id": "2412.06806", "title": "A Physics-Inspired Deep Learning Framework with Polar Coordinate\n  Attention for Ptychographic Imaging", "abstract": "  Ptychographic imaging confronts inherent challenges in applying deep learning\nfor phase retrieval from diffraction patterns. Conventional neural\narchitectures, both convolutional neural networks and Transformer-based\nmethods, are optimized for natural images with Euclidean spatial\nneighborhood-based inductive biases that exhibit geometric mismatch with the\nconcentric coherent patterns characteristic of diffraction data in reciprocal\nspace. In this paper, we present PPN, a physics-inspired deep learning network\nwith Polar Coordinate Attention (PoCA) for ptychographic imaging, that aligns\nneural inductive biases with diffraction physics through a dual-branch\narchitecture separating local feature extraction from non-local coherence\nmodeling. It consists of a PoCA mechanism that replaces Euclidean spatial\npriors with physically consistent radial-angular correlations. PPN outperforms\nexisting end-to-end models, with spectral and spatial analysis confirming its\ngreater preservation of high-frequency details. Notably, PPN maintains robust\nperformance compared to iterative methods even at low overlap ratios, making it\nwell suited for high-throughput imaging in real-world acquisition scenarios for\nsamples with consistent structural characteristics.\n", "date": "2024-11-25", "categories": []}
{"id": "2502.20824", "title": "MFSR-GAN: Multi-Frame Super-Resolution with Handheld Motion Modeling", "abstract": "  Smartphone cameras have become ubiquitous imaging tools, yet their small\nsensors and compact optics often limit spatial resolution and introduce\ndistortions. Combining information from multiple low-resolution (LR) frames to\nproduce a high-resolution (HR) image has been explored to overcome the inherent\nlimitations of smartphone cameras. Despite the promise of multi-frame\nsuper-resolution (MFSR), current approaches are hindered by datasets that fail\nto capture the characteristic noise and motion patterns found in real-world\nhandheld burst images. In this work, we address this gap by introducing a novel\nsynthetic data engine that uses multi-exposure static images to synthesize\nLR-HR training pairs while preserving sensor-specific noise characteristics and\nimage motion found during handheld burst photography. We also propose MFSR-GAN:\na multi-scale RAW-to-RGB network for MFSR. Compared to prior approaches,\nMFSR-GAN emphasizes a \"base frame\" throughout its architecture to mitigate\nartifacts. Experimental results on both synthetic and real data demonstrates\nthat MFSR-GAN trained with our synthetic engine yields sharper, more realistic\nreconstructions than existing methods for real-world MFSR.\n", "date": "2025-02-28", "categories": []}
{"id": "2505.00220", "title": "Towards Robust and Generalizable Gerchberg Saxton based Physics Inspired\n  Neural Networks for Computer Generated Holography: A Sensitivity Analysis\n  Framework", "abstract": "  Computer-generated holography (CGH) enables applications in holographic\naugmented reality (AR), 3D displays, systems neuroscience, and optical\ntrapping. The fundamental challenge in CGH is solving the inverse problem of\nphase retrieval from intensity measurements. Physics-inspired neural networks\n(PINNs), especially Gerchberg-Saxton-based PINNs (GS-PINNs), have advanced\nphase retrieval capabilities. However, their performance strongly depends on\nforward models (FMs) and their hyperparameters (FMHs), limiting generalization,\ncomplicating benchmarking, and hindering hardware optimization. We present a\nsystematic sensitivity analysis framework based on Saltelli's extension of\nSobol's method to quantify FMH impacts on GS-PINN performance. Our analysis\ndemonstrates that SLM pixel-resolution is the primary factor affecting neural\nnetwork sensitivity, followed by pixel-pitch, propagation distance, and\nwavelength. Free space propagation forward models demonstrate superior neural\nnetwork performance compared to Fourier holography, providing enhanced\nparameterization and generalization. We introduce a composite evaluation metric\ncombining performance consistency, generalization capability, and\nhyperparameter perturbation resilience, establishing a unified benchmarking\nstandard across CGH configurations. Our research connects physics-inspired deep\nlearning theory with practical CGH implementations through concrete guidelines\nfor forward model selection, neural network architecture, and performance\nevaluation. Our contributions advance the development of robust, interpretable,\nand generalizable neural networks for diverse holographic applications,\nsupporting evidence-based decisions in CGH research and implementation.\n", "date": "2025-04-30", "categories": []}
{"id": "2505.00643", "title": "Deep Learning Assisted Outer Volume Removal for Highly-Accelerated\n  Real-Time Dynamic MRI", "abstract": "  Real-time (RT) dynamic MRI plays a vital role in capturing rapid\nphysiological processes, offering unique insights into organ motion and\nfunction. Among these applications, RT cine MRI is particularly important for\nfunctional assessment of the heart with high temporal resolution. RT imaging\nenables free-breathing, ungated imaging of cardiac motion, making it a crucial\nalternative for patients who cannot tolerate conventional breath-hold,\nECG-gated acquisitions. However, achieving high acceleration rates in RT cine\nMRI is challenging due to aliasing artifacts from extra-cardiac tissues,\nparticularly at high undersampling factors. In this study, we propose a novel\nouter volume removal (OVR) method to address this challenge by eliminating\naliasing contributions from non-cardiac regions in a post-processing framework.\nOur approach estimates the outer volume signal for each timeframe using\ncomposite temporal images from time-interleaved undersampling patterns, which\ninherently contain pseudo-periodic ghosting artifacts. A deep learning (DL)\nmodel is trained to identify and remove these artifacts, producing a clean\nouter volume estimate that is subsequently subtracted from the corresponding\nk-space data. The final reconstruction is performed with a physics-driven DL\n(PD-DL) method trained using an OVR-specific loss function to restore high\nspatio-temporal resolution images. Experimental results show that the proposed\nmethod at high accelerations achieves image quality that is visually comparable\nto clinical baseline images, while outperforming conventional reconstruction\ntechniques, both qualitatively and quantitatively. The proposed approach\nprovides a practical and effective solution for artifact reduction in RT cine\nMRI without requiring acquisition modifications, offering a pathway to higher\nacceleration rates while preserving diagnostic quality.\n", "date": "2025-05-01", "categories": []}
{"id": "2406.00834", "title": "End-to-End Hybrid Refractive-Diffractive Lens Design with Differentiable\n  Ray-Wave Model", "abstract": "  Hybrid refractive-diffractive lenses combine the light efficiency of\nrefractive lenses with the information encoding power of diffractive optical\nelements (DOE), showing great potential as the next generation of imaging\nsystems. However, accurately simulating such hybrid designs is generally\ndifficult, and in particular, there are no existing differentiable image\nformation models for hybrid lenses with sufficient accuracy.\n  In this work, we propose a new hybrid ray-tracing and wave-propagation\n(ray-wave) model for accurate simulation of both optical aberrations and\ndiffractive phase modulation, where the DOE is placed between the last\nrefractive surface and the image sensor, i.e. away from the Fourier plane that\nis often used as a DOE position. The proposed ray-wave model is fully\ndifferentiable, enabling gradient back-propagation for end-to-end co-design of\nrefractive-diffractive lens optimization and the image reconstruction network.\nWe validate the accuracy of the proposed model by comparing the simulated point\nspread functions (PSFs) with theoretical results, as well as simulation\nexperiments that show our model to be more accurate than solutions implemented\nin commercial software packages like Zemax. We demonstrate the effectiveness of\nthe proposed model through real-world experiments and show significant\nimprovements in both aberration correction and extended depth-of-field (EDoF)\nimaging. We believe the proposed model will motivate further investigation into\na wide range of applications in computational imaging, computational\nphotography, and advanced optical design. Code will be released upon\npublication.\n", "date": "2024-06-02", "categories": []}
{"id": "2501.00142", "title": "Minimalist Vision with Freeform Pixels", "abstract": "  A minimalist vision system uses the smallest number of pixels needed to solve\na vision task. While traditional cameras use a large grid of square pixels, a\nminimalist camera uses freeform pixels that can take on arbitrary shapes to\nincrease their information content. We show that the hardware of a minimalist\ncamera can be modeled as the first layer of a neural network, where the\nsubsequent layers are used for inference. Training the network for any given\ntask yields the shapes of the camera's freeform pixels, each of which is\nimplemented using a photodetector and an optical mask. We have designed\nminimalist cameras for monitoring indoor spaces (with 8 pixels), measuring room\nlighting (with 8 pixels), and estimating traffic flow (with 8 pixels). The\nperformance demonstrated by these systems is on par with a traditional camera\nwith orders of magnitude more pixels. Minimalist vision has two major\nadvantages. First, it naturally tends to preserve the privacy of individuals in\nthe scene since the captured information is inadequate for extracting visual\ndetails. Second, since the number of measurements made by a minimalist camera\nis very small, we show that it can be fully self-powered, i.e., function\nwithout an external power supply or a battery.\n", "date": "2024-12-30", "categories": []}
{"id": "2501.15309", "title": "Investigating the Feasibility of Patch-based Inference for Generalized\n  Diffusion Priors in Inverse Problems for Medical Images", "abstract": "  Plug-and-play approaches to solving inverse problems such as restoration and\nsuper-resolution have recently benefited from Diffusion-based generative priors\nfor natural as well as medical images. However, solutions often use the\nstandard albeit computationally intensive route of training and inferring with\nthe whole image on the diffusion prior. While patch-based approaches to\nevaluating diffusion priors in plug-and-play methods have received some\ninterest, they remain an open area of study. In this work, we explore the\nfeasibility of the usage of patches for training and inference of a diffusion\nprior on MRI images. We explore the minor adaptation necessary for artifact\navoidance, the performance and the efficiency of memory usage of patch-based\nmethods as well as the adaptability of whole image training to patch-based\nevaluation - evaluating across multiple plug-and-play methods, tasks and\ndatasets.\n", "date": "2025-01-25", "categories": []}
{"id": "2504.18067", "title": "Physics-Driven Neural Compensation For Electrical Impedance Tomography", "abstract": "  Electrical Impedance Tomography (EIT) provides a non-invasive, portable\nimaging modality with significant potential in medical and industrial\napplications. Despite its advantages, EIT encounters two primary challenges:\nthe ill-posed nature of its inverse problem and the spatially variable,\nlocation-dependent sensitivity distribution. Traditional model-based methods\nmitigate ill-posedness through regularization but overlook sensitivity\nvariability, while supervised deep learning approaches require extensive\ntraining data and lack generalization. Recent developments in neural fields\nhave introduced implicit regularization techniques for image reconstruction,\nbut these methods typically neglect the physical principles underlying EIT,\nthus limiting their effectiveness. In this study, we propose PhyNC\n(Physics-driven Neural Compensation), an unsupervised deep learning framework\nthat incorporates the physical principles of EIT. PhyNC addresses both the\nill-posed inverse problem and the sensitivity distribution by dynamically\nallocating neural representational capacity to regions with lower sensitivity,\nensuring accurate and balanced conductivity reconstructions. Extensive\nevaluations on both simulated and experimental data demonstrate that PhyNC\noutperforms existing methods in terms of detail preservation and artifact\nresistance, particularly in low-sensitivity regions. Our approach enhances the\nrobustness of EIT reconstructions and provides a flexible framework that can be\nadapted to other imaging modalities with similar challenges.\n", "date": "2025-04-25", "categories": []}
{"id": "2503.23313", "title": "SpINR: Neural Volumetric Reconstruction for FMCW Radars", "abstract": "  In this paper, we introduce SpINR, a novel framework for volumetric\nreconstruction using Frequency-Modulated Continuous-Wave (FMCW) radar data.\nTraditional radar imaging techniques, such as backprojection, often assume\nideal signal models and require dense aperture sampling, leading to limitations\nin resolution and generalization. To address these challenges, SpINR integrates\na fully differentiable forward model that operates natively in the frequency\ndomain with implicit neural representations (INRs). This integration leverages\nthe linear relationship between beat frequency and scatterer distance inherent\nin FMCW radar systems, facilitating more efficient and accurate learning of\nscene geometry. Additionally, by computing outputs for only the relevant\nfrequency bins, our forward model achieves greater computational efficiency\ncompared to time-domain approaches that process the entire signal before\ntransformation. Through extensive experiments, we demonstrate that SpINR\nsignificantly outperforms classical backprojection methods and existing\nlearning-based approaches, achieving higher resolution and more accurate\nreconstructions of complex scenes. This work represents the first application\nof neural volumetic reconstruction in the radar domain, offering a promising\ndirection for future research in radar-based imaging and perception systems.\n", "date": "2025-03-30", "categories": []}
{"id": "2504.18520", "title": "RSFR: A Coarse-to-Fine Reconstruction Framework for Diffusion Tensor\n  Cardiac MRI with Semantic-Aware Refinement", "abstract": "  Cardiac diffusion tensor imaging (DTI) offers unique insights into\ncardiomyocyte arrangements, bridging the gap between microscopic and\nmacroscopic cardiac function. However, its clinical utility is limited by\ntechnical challenges, including a low signal-to-noise ratio, aliasing\nartefacts, and the need for accurate quantitative fidelity. To address these\nlimitations, we introduce RSFR (Reconstruction, Segmentation, Fusion &\nRefinement), a novel framework for cardiac diffusion-weighted image\nreconstruction. RSFR employs a coarse-to-fine strategy, leveraging zero-shot\nsemantic priors via the Segment Anything Model and a robust Vision Mamba-based\nreconstruction backbone. Our framework integrates semantic features effectively\nto mitigate artefacts and enhance fidelity, achieving state-of-the-art\nreconstruction quality and accurate DT parameter estimation under high\nundersampling rates. Extensive experiments and ablation studies demonstrate the\nsuperior performance of RSFR compared to existing methods, highlighting its\nrobustness, scalability, and potential for clinical translation in quantitative\ncardiac DTI.\n", "date": "2025-04-25", "categories": []}
{"id": "2504.16636", "title": "Dual-Camera All-in-Focus Neural Radiance Fields", "abstract": "  We present the first framework capable of synthesizing the all-in-focus\nneural radiance field (NeRF) from inputs without manual refocusing. Without\nrefocusing, the camera will automatically focus on the fixed object for all\nviews, and current NeRF methods typically using one camera fail due to the\nconsistent defocus blur and a lack of sharp reference. To restore the\nall-in-focus NeRF, we introduce the dual-camera from smartphones, where the\nultra-wide camera has a wider depth-of-field (DoF) and the main camera\npossesses a higher resolution. The dual camera pair saves the high-fidelity\ndetails from the main camera and uses the ultra-wide camera's deep DoF as\nreference for all-in-focus restoration. To this end, we first implement spatial\nwarping and color matching to align the dual camera, followed by a\ndefocus-aware fusion module with learnable defocus parameters to predict a\ndefocus map and fuse the aligned camera pair. We also build a multi-view\ndataset that includes image pairs of the main and ultra-wide cameras in a\nsmartphone. Extensive experiments on this dataset verify that our solution,\ntermed DC-NeRF, can produce high-quality all-in-focus novel views and compares\nfavorably against strong baselines quantitatively and qualitatively. We further\nshow DoF applications of DC-NeRF with adjustable blur intensity and focal\nplane, including refocusing and split diopter.\n", "date": "2025-04-23", "categories": []}
{"id": "2504.07758", "title": "PIDSR: Complementary Polarized Image Demosaicing and Super-Resolution", "abstract": "  Polarization cameras can capture multiple polarized images with different\npolarizer angles in a single shot, bringing convenience to polarization-based\ndownstream tasks. However, their direct outputs are color-polarization filter\narray (CPFA) raw images, requiring demosaicing to reconstruct full-resolution,\nfull-color polarized images; unfortunately, this necessary step introduces\nartifacts that make polarization-related parameters such as the degree of\npolarization (DoP) and angle of polarization (AoP) prone to error. Besides,\nlimited by the hardware design, the resolution of a polarization camera is\noften much lower than that of a conventional RGB camera. Existing polarized\nimage demosaicing (PID) methods are limited in that they cannot enhance\nresolution, while polarized image super-resolution (PISR) methods, though\ndesigned to obtain high-resolution (HR) polarized images from the demosaicing\nresults, tend to retain or even amplify errors in the DoP and AoP introduced by\ndemosaicing artifacts. In this paper, we propose PIDSR, a joint framework that\nperforms complementary Polarized Image Demosaicing and Super-Resolution,\nshowing the ability to robustly obtain high-quality HR polarized images with\nmore accurate DoP and AoP from a CPFA raw image in a direct manner. Experiments\nshow our PIDSR not only achieves state-of-the-art performance on both synthetic\nand real data, but also facilitates downstream tasks.\n", "date": "2025-04-10", "categories": []}
{"id": "2504.15311", "title": "RINN: One Sample Radio Frequency Imaging based on Physics Informed\n  Neural Network", "abstract": "  Due to its ability to work in non-line-of-sight and low-light environments,\nradio frequency (RF) imaging technology is expected to bring new possibilities\nfor embodied intelligence and multimodal sensing. However, widely used RF\ndevices (such as Wi-Fi) often struggle to provide high-precision\nelectromagnetic measurements and large-scale datasets, hindering the\napplication of RF imaging technology. In this paper, we combine the ideas of\nPINN to design the RINN network, using physical constraints instead of true\nvalue comparison constraints and adapting it with the characteristics of\nubiquitous RF signals, allowing the RINN network to achieve RF imaging using\nonly one sample without phase and with amplitude noise. Our numerical\nevaluation results show that compared with 5 classic algorithms based on phase\ndata for imaging results, RINN's imaging results based on phaseless data are\ngood, with indicators such as RRMSE (0.11) performing similarly well. RINN\nprovides new possibilities for the universal development of radio frequency\nimaging technology.\n", "date": "2025-04-19", "categories": []}
{"id": "2410.02073", "title": "Depth Pro: Sharp Monocular Metric Depth in Less Than a Second", "abstract": "  We present a foundation model for zero-shot metric monocular depth\nestimation. Our model, Depth Pro, synthesizes high-resolution depth maps with\nunparalleled sharpness and high-frequency details. The predictions are metric,\nwith absolute scale, without relying on the availability of metadata such as\ncamera intrinsics. And the model is fast, producing a 2.25-megapixel depth map\nin 0.3 seconds on a standard GPU. These characteristics are enabled by a number\nof technical contributions, including an efficient multi-scale vision\ntransformer for dense prediction, a training protocol that combines real and\nsynthetic datasets to achieve high metric accuracy alongside fine boundary\ntracing, dedicated evaluation metrics for boundary accuracy in estimated depth\nmaps, and state-of-the-art focal length estimation from a single image.\nExtensive experiments analyze specific design choices and demonstrate that\nDepth Pro outperforms prior work along multiple dimensions. We release code and\nweights at https://github.com/apple/ml-depth-pro\n", "date": "2024-10-02", "categories": []}
{"id": "2410.05413", "title": "Implicitly Learned Neural Phase Functions for Basis-Free Point Spread\n  Function Engineering", "abstract": "  Point spread function (PSF) engineering is vital for precisely controlling\nthe focus of light in computational imaging, with applications in neural\nimaging, fluorescence microscopy, and biophotonics. The PSF is derived from the\nmagnitude of the Fourier transform of a phase function, making the construction\nof the phase function given the PSF (PSF engineering) an ill-posed inverse\nproblem. Traditional PSF engineering methods rely on physical basis functions,\nlimiting their ability to generalize across the range of PSFs required for\nimaging tasks. We introduce a novel approach leveraging implicit neural\nrepresentations that overcome the limitations of pixel-wise optimization\nmethods. Our approach achieves a median MSSIM of 0.8162 and a mean MSSIM of\n0.5634, compared to a median MSSIM of 0.0 and a mean MSSIM of 0.1841 with\npixel-wise optimization when learning randomly generated phase functions. Our\napproach also achieves a median PSNR of 10.38 dB and a mean PSNR of 8.672 dB,\ncompared to a median PSNR of 6.653 dB and a mean PSNR of 6.660 dB with\npixel-wise optimization for this task.\n", "date": "2024-10-07", "categories": []}
{"id": "2504.14156", "title": "Breaking the Diffraction Barrier for Passive Sources:\n  Parameter-Decoupled Superresolution Assisted by Physics-Informed Machine\n  Learning", "abstract": "  We present a parameter-decoupled superresolution framework for estimating\nsub-wavelength separations of passive two-point sources without requiring prior\nknowledge or control of the source. Our theoretical foundation circumvents the\nneed to estimate multiple challenging parameters such as partial coherence,\nbrightness imbalance, random relative phase, and photon statistics. A\nphysics-informed machine learning (ML) model (trained with a standard desktop\nworkstation), synergistically integrating this theory, further addresses\npractical imperfections including background noise, photon loss, and\ncentroid/orientation misalignment. The integrated parameter-decoupling\nsuperresolution method achieves resolution 14 and more times below the\ndiffraction limit (corresponding to ~ 13.5 nm in optical microscopy) on\nexperimentally generated realistic images with >82% fidelity, performance\nrivaling state-of-the-art techniques for actively controllable sources.\nCritically, our method's robustness against source parameter variability and\nsource-independent noises enables potential applications in realistic scenarios\nwhere source control is infeasible, such as astrophysical imaging, live-cell\nmicroscopy, and quantum metrology. This work bridges a critical gap between\ntheoretical superresolution limits and practical implementations for passive\nsystems.\n", "date": "2025-04-19", "categories": []}
{"id": "2504.14157", "title": "DeepPD: Joint Phase and Object Estimation from Phase Diversity with\n  Neural Calibration of a Deformable Mirror", "abstract": "  Sample-induced aberrations and optical imperfections limit the resolution of\nfluorescence microscopy. Phase diversity is a powerful technique that leverages\ncomplementary phase information in sequentially acquired images with\ndeliberately introduced aberrations--the phase diversities--to enable phase and\nobject reconstruction and restore diffraction-limited resolution. These phase\ndiversities are typically introduced into the optical path via a deformable\nmirror. Existing phase-diversity-based methods are limited to Zernike modes,\nrequire large numbers of diversity images, or depend on accurate mirror\ncalibration--which are all suboptimal. We present DeepPD, a deep learning-based\nframework that combines neural representations of the object and wavefront with\na learned model of the deformable mirror to jointly estimate both object and\nphase from only five images. DeepPD improves robustness and reconstruction\nquality over previous approaches, even under severe aberrations. We demonstrate\nits performance on calibration targets and biological samples, including\nimmunolabeled myosin in fixed PtK2 cells.\n", "date": "2025-04-19", "categories": []}
{"id": "2504.14664", "title": "Frequency-domain Learning with Kernel Prior for Blind Image Deblurring", "abstract": "  While achieving excellent results on various datasets, many deep learning\nmethods for image deblurring suffer from limited generalization capabilities\nwith out-of-domain data. This limitation is likely caused by their dependence\non certain domain-specific datasets. To address this challenge, we argue that\nit is necessary to introduce the kernel prior into deep learning methods, as\nthe kernel prior remains independent of the image context. For effective fusion\nof kernel prior information, we adopt a rational implementation method inspired\nby traditional deblurring algorithms that perform deconvolution in the\nfrequency domain. We propose a module called Frequency Integration Module (FIM)\nfor fusing the kernel prior and combine it with a frequency-based deblurring\nTransfomer network. Experimental results demonstrate that our method\noutperforms state-of-the-art methods on multiple blind image deblurring tasks,\nshowcasing robust generalization abilities. Source code will be available soon.\n", "date": "2025-04-20", "categories": []}
{"id": "2504.14699", "title": "IXGS-Intraoperative 3D Reconstruction from Sparse, Arbitrarily Posed\n  Real X-rays", "abstract": "  Spine surgery is a high-risk intervention demanding precise execution, often\nsupported by image-based navigation systems. Recently, supervised learning\napproaches have gained attention for reconstructing 3D spinal anatomy from\nsparse fluoroscopic data, significantly reducing reliance on\nradiation-intensive 3D imaging systems. However, these methods typically\nrequire large amounts of annotated training data and may struggle to generalize\nacross varying patient anatomies or imaging conditions. Instance-learning\napproaches like Gaussian splatting could offer an alternative by avoiding\nextensive annotation requirements. While Gaussian splatting has shown promise\nfor novel view synthesis, its application to sparse, arbitrarily posed real\nintraoperative X-rays has remained largely unexplored. This work addresses this\nlimitation by extending the $R^2$-Gaussian splatting framework to reconstruct\nanatomically consistent 3D volumes under these challenging conditions. We\nintroduce an anatomy-guided radiographic standardization step using style\ntransfer, improving visual consistency across views, and enhancing\nreconstruction quality. Notably, our framework requires no pretraining, making\nit inherently adaptable to new patients and anatomies. We evaluated our\napproach using an ex-vivo dataset. Expert surgical evaluation confirmed the\nclinical utility of the 3D reconstructions for navigation, especially when\nusing 20 to 30 views, and highlighted the standardization's benefit for\nanatomical clarity. Benchmarking via quantitative 2D metrics (PSNR/SSIM)\nconfirmed performance trade-offs compared to idealized settings, but also\nvalidated the improvement gained from standardization over raw inputs. This\nwork demonstrates the feasibility of instance-based volumetric reconstruction\nfrom arbitrary sparse-view X-rays, advancing intraoperative 3D imaging for\nsurgical navigation.\n", "date": "2025-04-20", "categories": []}
{"id": "2504.13042", "title": "Event-Enhanced Blurry Video Super-Resolution", "abstract": "  In this paper, we tackle the task of blurry video super-resolution (BVSR),\naiming to generate high-resolution (HR) videos from low-resolution (LR) and\nblurry inputs. Current BVSR methods often fail to restore sharp details at high\nresolutions, resulting in noticeable artifacts and jitter due to insufficient\nmotion information for deconvolution and the lack of high-frequency details in\nLR frames. To address these challenges, we introduce event signals into BVSR\nand propose a novel event-enhanced network, Ev-DeblurVSR. To effectively fuse\ninformation from frames and events for feature deblurring, we introduce a\nreciprocal feature deblurring module that leverages motion information from\nintra-frame events to deblur frame features while reciprocally using global\nscene context from the frames to enhance event features. Furthermore, to\nenhance temporal consistency, we propose a hybrid deformable alignment module\nthat fully exploits the complementary motion information from inter-frame\nevents and optical flow to improve motion estimation in the deformable\nalignment process. Extensive evaluations demonstrate that Ev-DeblurVSR\nestablishes a new state-of-the-art performance on both synthetic and real-world\ndatasets. Notably, on real data, our method is +2.59 dB more accurate and\n7.28$\\times$ faster than the recent best BVSR baseline FMA-Net. Code:\nhttps://github.com/DachunKai/Ev-DeblurVSR.\n", "date": "2025-04-17", "categories": []}
{"id": "2504.12401", "title": "NTIRE 2025 Challenge on Event-Based Image Deblurring: Methods and\n  Results", "abstract": "  This paper presents an overview of NTIRE 2025 the First Challenge on\nEvent-Based Image Deblurring, detailing the proposed methodologies and\ncorresponding results. The primary goal of the challenge is to design an\nevent-based method that achieves high-quality image deblurring, with\nperformance quantitatively assessed using Peak Signal-to-Noise Ratio (PSNR).\nNotably, there are no restrictions on computational complexity or model size.\nThe task focuses on leveraging both events and images as inputs for\nsingle-image deblurring. A total of 199 participants registered, among whom 15\nteams successfully submitted valid results, offering valuable insights into the\ncurrent state of event-based image deblurring. We anticipate that this\nchallenge will drive further advancements in event-based vision research.\n", "date": "2025-04-16", "categories": []}
{"id": "2504.13060", "title": "Imaging for All-Day Wearable Smart Glasses", "abstract": "  In recent years smart glasses technology has rapidly advanced, opening up\nentirely new areas for mobile computing. We expect future smart glasses will\nneed to be all-day wearable, adopting a small form factor to meet the\nrequirements of volume, weight, fashionability and social acceptability, which\nputs significant constraints on the space of possible solutions. Additional\nchallenges arise due to the fact that smart glasses are worn in arbitrary\nenvironments while their wearer moves and performs everyday activities. In this\npaper, we systematically analyze the space of imaging from smart glasses and\nderive several fundamental limits that govern this imaging domain. We discuss\nthe impact of these limits on achievable image quality and camera module size\n-- comparing in particular to related devices such as mobile phones. We then\npropose a novel distributed imaging approach that allows to minimize the size\nof the individual camera modules when compared to a standard monolithic camera\ndesign. Finally, we demonstrate the properties of this novel approach in a\nseries of experiments using synthetic data as well as images captured with two\ndifferent prototype implementations.\n", "date": "2025-04-17", "categories": []}
{"id": "2504.12112", "title": "A Diffusion-Based Framework for Terrain-Aware Remote Sensing Image\n  Reconstruction", "abstract": "  Remote sensing imagery is essential for environmental monitoring,\nagricultural management, and disaster response. However, data loss due to cloud\ncover, sensor failures, or incomplete acquisition-especially in high-resolution\nand high-frequency tasks-severely limits satellite imagery's effectiveness.\nTraditional interpolation methods struggle with large missing areas and complex\nstructures. Remote sensing imagery consists of multiple bands, each with\ndistinct meanings, and ensuring consistency across bands is critical to avoid\nanomalies in the combined images. This paper proposes SatelliteMaker, a\ndiffusion-based method that reconstructs missing data across varying levels of\ndata loss while maintaining spatial, spectral, and temporal consistency. We\nalso propose Digital Elevation Model (DEM) as a conditioning input and use\ntailored prompts to generate realistic images, making diffusion models\napplicable to quantitative remote sensing tasks. Additionally, we propose a\nVGG-Adapter module based on Distribution Loss, which reduces distribution\ndiscrepancy and ensures style consistency. Extensive experiments show that\nSatelliteMaker achieves state-of-the-art performance across multiple tasks.\n", "date": "2025-04-16", "categories": []}
{"id": "2504.12276", "title": "The Tenth NTIRE 2025 Image Denoising Challenge Report", "abstract": "  This paper presents an overview of the NTIRE 2025 Image Denoising Challenge\n({\\sigma} = 50), highlighting the proposed methodologies and corresponding\nresults. The primary objective is to develop a network architecture capable of\nachieving high-quality denoising performance, quantitatively evaluated using\nPSNR, without constraints on computational complexity or model size. The task\nassumes independent additive white Gaussian noise (AWGN) with a fixed noise\nlevel of 50. A total of 290 participants registered for the challenge, with 20\nteams successfully submitting valid results, providing insights into the\ncurrent state-of-the-art in image denoising.\n", "date": "2025-04-16", "categories": []}
{"id": "2504.11202", "title": "Focal Split: Untethered Snapshot Depth from Differential Defocus", "abstract": "  We introduce Focal Split, a handheld, snapshot depth camera with fully\nonboard power and computing based on depth-from-differential-defocus (DfDD).\nFocal Split is passive, avoiding power consumption of light sources. Its\nachromatic optical system simultaneously forms two differentially defocused\nimages of the scene, which can be independently captured using two photosensors\nin a snapshot. The data processing is based on the DfDD theory, which\nefficiently computes a depth and a confidence value for each pixel with only\n500 floating point operations (FLOPs) per pixel from the camera measurements.\nWe demonstrate a Focal Split prototype, which comprises a handheld custom\ncamera system connected to a Raspberry Pi 5 for real-time data processing. The\nsystem consumes 4.9 W and is powered on a 5 V, 10,000 mAh battery. The\nprototype can measure objects with distances from 0.4 m to 1.2 m, outputting\n480$\\times$360 sparse depth maps at 2.1 frames per second (FPS) using\nunoptimized Python scripts. Focal Split is DIY friendly. A comprehensive guide\nto building your own Focal Split depth camera, code, and additional data can be\nfound at https://focal-split.qiguo.org.\n", "date": "2025-04-15", "categories": []}
{"id": "2412.00664", "title": "Improving Decoupled Posterior Sampling for Inverse Problems using Data\n  Consistency Constraint", "abstract": "  Diffusion models have shown strong performances in solving inverse problems\nthrough posterior sampling while they suffer from errors during earlier steps.\nTo mitigate this issue, several Decoupled Posterior Sampling methods have been\nrecently proposed. However, the reverse process in these methods ignores\nmeasurement information, leading to errors that impede effective optimization\nin subsequent steps. To solve this problem, we propose Guided Decoupled\nPosterior Sampling (GDPS) by integrating a data consistency constraint in the\nreverse process. The constraint performs a smoother transition within the\noptimization process, facilitating a more effective convergence toward the\ntarget distribution. Furthermore, we extend our method to latent diffusion\nmodels and Tweedie's formula, demonstrating its scalability. We evaluate GDPS\non the FFHQ and ImageNet datasets across various linear and nonlinear tasks\nunder both standard and challenging conditions. Experimental results\ndemonstrate that GDPS achieves state-of-the-art performance, improving accuracy\nover existing methods.\n", "date": "2024-12-01", "categories": []}
{"id": "2412.18417", "title": "Ultra-Low Complexity On-Orbit Compression for Remote Sensing Imagery via\n  Block Modulated Imaging", "abstract": "  The growing field of remote sensing faces a challenge: the ever-increasing\nsize and volume of imagery data are exceeding the storage and transmission\ncapabilities of satellite platforms. Efficient compression of remote sensing\nimagery is a critical solution to alleviate these burdens on satellites.\nHowever, existing compression methods are often too computationally expensive\nfor satellites. With the continued advancement of compressed sensing theory,\nsingle-pixel imaging emerges as a powerful tool that brings new possibilities\nfor on-orbit image compression. However, it still suffers from prolonged\nimaging times and the inability to perform high-resolution imaging, hindering\nits practical application. This paper advances the study of compressed sensing\nin remote sensing image compression, proposing Block Modulated Imaging (BMI).\nBy requiring only a single exposure, BMI significantly enhances imaging\nacquisition speeds. Additionally, BMI obviates the need for digital micromirror\ndevices and surpasses limitations in image resolution. Furthermore, we propose\na novel decoding network specifically designed to reconstruct images compressed\nunder the BMI framework. Leveraging the gated 3D convolutions and promoting\nefficient information flow across stages through a Two-Way Cross-Attention\nmodule, our decoding network exhibits demonstrably superior reconstruction\nperformance. Extensive experiments conducted on multiple renowned remote\nsensing datasets unequivocally demonstrate the efficacy of our proposed method.\nTo further validate its practical applicability, we developed and tested a\nprototype of the BMI-based camera, which has shown promising potential for\non-orbit image compression. The code is available at\nhttps://github.com/Johnathan218/BMNet.\n", "date": "2024-12-24", "categories": []}
{"id": "2504.09069", "title": "UniFlowRestore: A General Video Restoration Framework via Flow Matching\n  and Prompt Guidance", "abstract": "  Video imaging is often affected by complex degradations such as blur, noise,\nand compression artifacts. Traditional restoration methods follow a\n\"single-task single-model\" paradigm, resulting in poor generalization and high\ncomputational cost, limiting their applicability in real-world scenarios with\ndiverse degradation types. We propose UniFlowRestore, a general video\nrestoration framework that models restoration as a time-continuous evolution\nunder a prompt-guided and physics-informed vector field. A physics-aware\nbackbone PhysicsUNet encodes degradation priors as potential energy, while\nPromptGenerator produces task-relevant prompts as momentum. These components\ndefine a Hamiltonian system whose vector field integrates inertial dynamics,\ndecaying physical gradients, and prompt-based guidance. The system is optimized\nvia a fixed-step ODE solver to achieve efficient and unified restoration across\ntasks. Experiments show that UniFlowRestore delivers stateof-the-art\nperformance with strong generalization and efficiency. Quantitative results\ndemonstrate that UniFlowRestore achieves state-of-the-art performance,\nattaining the highest PSNR (33.89 dB) and SSIM (0.97) on the video denoising\ntask, while maintaining top or second-best scores across all evaluated tasks.\n", "date": "2025-04-12", "categories": []}
{"id": "2504.09379", "title": "Low-Light Image Enhancement using Event-Based Illumination Estimation", "abstract": "  Low-light image enhancement (LLIE) aims to improve the visibility of images\ncaptured in poorly lit environments. Prevalent event-based solutions primarily\nutilize events triggered by motion, i.e., ''motion events'' to strengthen only\nthe edge texture, while leaving the high dynamic range and excellent low-light\nresponsiveness of event cameras largely unexplored. This paper instead opens a\nnew avenue from the perspective of estimating the illumination using\n''temporal-mapping'' events, i.e., by converting the timestamps of events\ntriggered by a transmittance modulation into brightness values. The resulting\nfine-grained illumination cues facilitate a more effective decomposition and\nenhancement of the reflectance component in low-light images through the\nproposed Illumination-aided Reflectance Enhancement module. Furthermore, the\ndegradation model of temporal-mapping events under low-light condition is\ninvestigated for realistic training data synthesizing. To address the lack of\ndatasets under this regime, we construct a beam-splitter setup and collect\nEvLowLight dataset that includes images, temporal-mapping events, and motion\nevents. Extensive experiments across 5 synthetic datasets and our real-world\nEvLowLight dataset substantiate that the devised pipeline, dubbed RetinEV,\nexcels in producing well-illuminated, high dynamic range images, outperforming\nprevious state-of-the-art event-based methods by up to 6.62 dB, while\nmaintaining an efficient inference speed of 35.6 frame-per-second on a 640X480\nimage.\n", "date": "2025-04-13", "categories": []}
{"id": "2504.08591", "title": "ZipIR: Latent Pyramid Diffusion Transformer for High-Resolution Image\n  Restoration", "abstract": "  Recent progress in generative models has significantly improved image\nrestoration capabilities, particularly through powerful diffusion models that\noffer remarkable recovery of semantic details and local fidelity. However,\ndeploying these models at ultra-high resolutions faces a critical trade-off\nbetween quality and efficiency due to the computational demands of long-range\nattention mechanisms. To address this, we introduce ZipIR, a novel framework\nthat enhances efficiency, scalability, and long-range modeling for high-res\nimage restoration. ZipIR employs a highly compressed latent representation that\ncompresses image 32x, effectively reducing the number of spatial tokens, and\nenabling the use of high-capacity models like the Diffusion Transformer (DiT).\nToward this goal, we propose a Latent Pyramid VAE (LP-VAE) design that\nstructures the latent space into sub-bands to ease diffusion training. Trained\non full images up to 2K resolution, ZipIR surpasses existing diffusion-based\nmethods, offering unmatched speed and quality in restoring high-resolution\nimages from severely degraded inputs.\n", "date": "2025-04-11", "categories": []}
{"id": "2504.07927", "title": "Zero-Shot Low-dose CT Denoising via Sinogram Flicking", "abstract": "  Many low-dose CT imaging methods rely on supervised learning, which requires\na large number of paired noisy and clean images. However, obtaining paired\nimages in clinical practice is challenging. To address this issue, zero-shot\nself-supervised methods train denoising networks using only the information\nwithin a single image, such as ZS-N2N. However, these methods often employ\ndownsampling operations that degrade image resolution. Additionally, the\ntraining dataset is inherently constrained to the image itself. In this paper,\nwe propose a zero-shot low-dose CT imaging method based on sinogram flicking,\nwhich operates within a single image but generates many copies via random\nconjugate ray matching. Specifically, two conjugate X-ray pencil beams measure\nthe same path; their expected values should be identical, while their noise\nlevels vary during measurements. By randomly swapping portions of the conjugate\nX-rays in the sinogram domain, we generate a large set of sinograms with\nconsistent content but varying noise patterns. When displayed dynamically,\nthese sinograms exhibit a flickering effect due to their identical structural\ncontent but differing noise patterns-hence the term sinogram flicking. We train\nthe network on pairs of sinograms with the same content but different noise\ndistributions using a lightweight model adapted from ZS-NSN. This process is\nrepeated to obtain the final results. A simulation study demonstrates that our\nmethod outperforms state-of-the-art approaches such as ZS-N2N.\n", "date": "2025-04-10", "categories": []}
{"id": "2310.07887", "title": "Unsupervised Denoising for Signal-Dependent and Row-Correlated Imaging\n  Noise", "abstract": "  Accurate analysis of microscopy images is hindered by the presence of noise.\nThis noise is usually signal-dependent and often additionally correlated along\nrows or columns of pixels. Current self- and unsupervised denoisers can address\nsignal-dependent noise, but none can reliably remove noise that is also row- or\ncolumn-correlated. Here, we present the first fully unsupervised deep\nlearning-based denoiser capable of handling imaging noise that is\nrow-correlated as well as signal-dependent. Our approach uses a Variational\nAutoencoder (VAE) with a specially designed autoregressive decoder. This\ndecoder is capable of modeling row-correlated and signal-dependent noise but is\nincapable of independently modeling underlying clean signal. The VAE therefore\nproduces latent variables containing only clean signal information, and these\nare mapped back into image space using a proposed second decoder network. Our\nmethod does not require a pre-trained noise model and can be trained from\nscratch using unpaired noisy data. We benchmark our approach on microscopy\ndatatsets from a range of imaging modalities and sensor types, each with row-\nor column-correlated, signal-dependent noise, and show that it outperforms\nexisting self- and unsupervised denoisers.\n", "date": "2023-10-11", "categories": []}
{"id": "2504.05590", "title": "CoA: Towards Real Image Dehazing via Compression-and-Adaptation", "abstract": "  Learning-based image dehazing algorithms have shown remarkable success in\nsynthetic domains. However, real image dehazing is still in suspense due to\ncomputational resource constraints and the diversity of real-world scenes.\nTherefore, there is an urgent need for an algorithm that excels in both\nefficiency and adaptability to address real image dehazing effectively. This\nwork proposes a Compression-and-Adaptation (CoA) computational flow to tackle\nthese challenges from a divide-and-conquer perspective. First, model\ncompression is performed in the synthetic domain to develop a compact dehazing\nparameter space, satisfying efficiency demands. Then, a bilevel adaptation in\nthe real domain is introduced to be fearless in unknown real environments by\naggregating the synthetic dehazing capabilities during the learning process.\nLeveraging a succinct design free from additional constraints, our CoA exhibits\ndomain-irrelevant stability and model-agnostic flexibility, effectively\nbridging the model chasm between synthetic and real domains to further improve\nits practical utility. Extensive evaluations and analyses underscore the\napproach's superiority and effectiveness. The code is publicly available at\nhttps://github.com/fyxnl/COA.\n", "date": "2025-04-08", "categories": []}
{"id": "2404.14414", "title": "Removing Reflections from RAW Photos", "abstract": "  We describe a system to remove real-world reflections from images for\nconsumer photography. Our system operates on linear (RAW) photos, and accepts\nan optional contextual photo looking in the opposite direction (e.g., the\n\"selfie\" camera on a mobile device). This optional photo disambiguates what\nshould be considered the reflection. The system is trained solely on synthetic\nmixtures of real RAW photos, which we combine using a reflection simulation\nthat is photometrically and geometrically accurate. Our system comprises a base\nmodel that accepts the captured photo and optional context photo as input, and\nruns at 256p, followed by an up-sampling model that transforms 256p images to\nfull resolution. The system produces preview images at 1K in 4.5-6.5s on a\nMacBook or iPhone 14 Pro. We show SOTA results on RAW photos that were captured\nin the field to embody typical consumer photos, and show that training on RAW\nsimulation data improves performance more than the architectural variations\namong prior works.\n", "date": "2024-02-28", "categories": []}
{"id": "2405.03384", "title": "GLIP: Electromagnetic Field Exposure Map Completion by Deep Generative\n  Networks", "abstract": "  In Spectrum cartography (SC), the generation of exposure maps for radio\nfrequency electromagnetic fields (RF-EMF) spans dimensions of frequency, space,\nand time, which relies on a sparse collection of sensor data, posing a\nchallenging ill-posed inverse problem. Cartography methods based on models\nintegrate designed priors, such as sparsity and low-rank structures, to refine\nthe solution of this inverse problem. In our previous work, EMF exposure map\nreconstruction was achieved by Generative Adversarial Networks (GANs) where\nphysical laws or structural constraints were employed as a prior, but they\nrequire a large amount of labeled data or simulated full maps for training to\nproduce efficient results. In this paper, we present a method to reconstruct\nEMF exposure maps using only the generator network in GANs which does not\nrequire explicit training, thus overcoming the limitations of GANs, such as\nusing reference full exposure maps. This approach uses a prior from sensor data\nas Local Image Prior (LIP) captured by deep convolutional generative networks\nindependent of learning the network parameters from images in an urban\nenvironment. Experimental results show that, even when only sparse sensor data\nare available, our method can produce accurate estimates.\n", "date": "2024-05-06", "categories": []}
{"id": "2411.09911", "title": "DiffFNO: Diffusion Fourier Neural Operator", "abstract": "  We introduce DiffFNO, a novel diffusion framework for arbitrary-scale\nsuper-resolution strengthened by a Weighted Fourier Neural Operator (WFNO).\nMode Rebalancing in WFNO effectively captures critical frequency components,\nsignificantly improving the reconstruction of high-frequency image details that\nare crucial for super-resolution tasks. Gated Fusion Mechanism (GFM) adaptively\ncomplements WFNO's spectral features with spatial features from an\nAttention-based Neural Operator (AttnNO). This enhances the network's\ncapability to capture both global structures and local details. Adaptive\nTime-Step (ATS) ODE solver, a deterministic sampling strategy, accelerates\ninference without sacrificing output quality by dynamically adjusting\nintegration step sizes ATS. Extensive experiments demonstrate that DiffFNO\nachieves state-of-the-art (SOTA) results, outperforming existing methods across\nvarious scaling factors by a margin of 2-4 dB in PSNR, including those beyond\nthe training distribution. It also achieves this at lower inference time. Our\napproach sets a new standard in super-resolution, delivering both superior\naccuracy and computational efficiency.\n", "date": "2024-11-15", "categories": []}
{"id": "2411.17313", "title": "Event Ellipsometer: Event-based Mueller-Matrix Video Imaging", "abstract": "  Light-matter interactions modify both the intensity and polarization state of\nlight. Changes in polarization, represented by a Mueller matrix, encode\ndetailed scene information. Existing optical ellipsometers capture\nMueller-matrix images; however, they are often limited to capturing static\nscenes due to long acquisition times. Here, we introduce Event Ellipsometer, a\nmethod for acquiring a Mueller-matrix video for dynamic scenes. Our imaging\nsystem employs fast-rotating quarter-wave plates (QWPs) in front of a light\nsource and an event camera that asynchronously captures intensity changes\ninduced by the rotating QWPs. We develop an ellipsometric-event image formation\nmodel, a calibration method, and an ellipsometric-event reconstruction method.\nWe experimentally demonstrate that Event Ellipsometer enables Mueller-matrix\nvideo imaging at 30fps, extending ellipsometry to dynamic scenes.\n", "date": "2024-11-26", "categories": []}
{"id": "2504.04228", "title": "Autoregressive High-Order Finite Difference Modulo Imaging: High-Dynamic\n  Range for Computer Vision Applications", "abstract": "  High dynamic range (HDR) imaging is vital for capturing the full range of\nlight tones in scenes, essential for computer vision tasks such as autonomous\ndriving. Standard commercial imaging systems face limitations in capacity for\nwell depth, and quantization precision, hindering their HDR capabilities.\nModulo imaging, based on unlimited sampling (US) theory, addresses these\nlimitations by using a modulo analog-to-digital approach that resets signals\nupon saturation, enabling estimation of pixel resets through neighboring pixel\nintensities. Despite the effectiveness of (US) algorithms in one-dimensional\nsignals, their optimization problem for two-dimensional signals remains\nunclear. This work formulates the US framework as an autoregressive $\\ell_2$\nphase unwrapping problem, providing computationally efficient solutions in the\ndiscrete cosine domain jointly with a stride removal algorithm also based on\nspatial differences. By leveraging higher-order finite differences for\ntwo-dimensional images, our approach enhances HDR image reconstruction from\nmodulo images, demonstrating its efficacy in improving object detection in\nautonomous driving scenes without retraining.\n", "date": "2025-04-05", "categories": []}
{"id": "2404.00916", "title": "Gyro-based Neural Single Image Deblurring", "abstract": "  In this paper, we present GyroDeblurNet, a novel single-image deblurring\nmethod that utilizes a gyro sensor to resolve the ill-posedness of image\ndeblurring. The gyro sensor provides valuable information about camera motion\nthat can improve deblurring quality. However, exploiting real-world gyro data\nis challenging due to errors from various sources. To handle these errors,\nGyroDeblurNet is equipped with two novel neural network blocks: a gyro\nrefinement block and a gyro deblurring block. The gyro refinement block refines\nthe erroneous gyro data using the blur information from the input image. The\ngyro deblurring block removes blur from the input image using the refined gyro\ndata and further compensates for gyro error by leveraging the blur information\nfrom the input image. For training a neural network with erroneous gyro data,\nwe propose a training strategy based on the curriculum learning. We also\nintroduce a novel gyro data embedding scheme to represent real-world intricate\ncamera shakes. Finally, we present both synthetic and real-world datasets for\ntraining and evaluating gyro-based single image deblurring. Our experiments\ndemonstrate that our approach achieves state-of-the-art deblurring quality by\neffectively utilizing erroneous gyro data.\n", "date": "2024-04-01", "categories": []}
{"id": "2410.16290", "title": "A Unified Model for Compressed Sensing MRI Across Undersampling Patterns", "abstract": "  Compressed Sensing MRI reconstructs images of the body's internal anatomy\nfrom undersampled measurements, thereby reducing scan time. Recently, deep\nlearning has shown great potential for reconstructing high-fidelity images from\nhighly undersampled measurements. However, one needs to train multiple models\nfor different undersampling patterns and desired output image resolutions,\nsince most networks operate on a fixed discretization. Such approaches are\nhighly impractical in clinical settings, where undersampling patterns and image\nresolutions are frequently changed to accommodate different real-time imaging\nand diagnostic requirements.\n  We propose a unified MRI reconstruction model robust to various measurement\nundersampling patterns and image resolutions. Our approach uses neural\noperators, a discretization-agnostic architecture applied in both image and\nmeasurement spaces, to capture local and global features. Empirically, our\nmodel improves SSIM by 11% and PSNR by 4 dB over a state-of-the-art CNN\n(End-to-End VarNet), with 600$\\times$ faster inference than diffusion methods.\nThe resolution-agnostic design also enables zero-shot super-resolution and\nextended field-of-view reconstruction, offering a versatile and efficient\nsolution for clinical MR imaging. Our unified model offers a versatile solution\nfor MRI, adapting seamlessly to various measurement undersampling and imaging\nresolutions, making it highly effective for flexible and reliable clinical\nimaging. Our code is available at https://armeet.ca/nomri.\n", "date": "2024-10-05", "categories": []}
{"id": "2412.06978", "title": "Edge-SD-SR: Low Latency and Parameter Efficient On-device\n  Super-Resolution with Stable Diffusion via Bidirectional Conditioning", "abstract": "  There has been immense progress recently in the visual quality of Stable\nDiffusion-based Super Resolution (SD-SR). However, deploying large diffusion\nmodels on computationally restricted devices such as mobile phones remains\nimpractical due to the large model size and high latency. This is compounded\nfor SR as it often operates at high res (e.g. 4Kx3K). In this work, we\nintroduce Edge-SD-SR, the first parameter efficient and low latency diffusion\nmodel for image super-resolution. Edge-SD-SR consists of ~169M parameters,\nincluding UNet, encoder and decoder, and has a complexity of only ~142 GFLOPs.\nTo maintain a high visual quality on such low compute budget, we introduce a\nnumber of training strategies: (i) A novel conditioning mechanism on the low\nresolution input, coined bidirectional conditioning, which tailors the SD model\nfor the SR task. (ii) Joint training of the UNet and encoder, while decoupling\nthe encodings of the HR and LR images and using a dedicated schedule. (iii)\nFinetuning the decoder using the UNet's output to directly tailor the decoder\nto the latents obtained at inference time. Edge-SD-SR runs efficiently on\ndevice, e.g. it can upscale a 128x128 patch to 512x512 in 38 msec while running\non a Samsung S24 DSP, and of a 512x512 to 2048x2048 (requiring 25 model\nevaluations) in just ~1.1 sec. Furthermore, we show that Edge-SD-SR matches or\neven outperforms state-of-the-art SR approaches on the most established SR\nbenchmarks.\n", "date": "2024-12-09", "categories": []}
{"id": "2311.02003", "title": "Efficient Model-Based Deep Learning via Network Pruning and Fine-Tuning", "abstract": "  Model-based deep learning (MBDL) is a powerful methodology for designing deep\nmodels to solve imaging inverse problems. MBDL networks can be seen as\niterative algorithms that estimate the desired image using a physical\nmeasurement model and a learned image prior specified using a convolutional\nneural net (CNNs). The iterative nature of MBDL networks increases the\ntest-time computational complexity, which limits their applicability in certain\nlarge-scale applications. Here we make two contributions to address this issue:\nFirst, we show how structured pruning can be adopted to reduce the number of\nparameters in MBDL networks. Second, we present three methods to fine-tune the\npruned MBDL networks to mitigate potential performance loss. Each fine-tuning\nstrategy has a unique benefit that depends on the presence of a pre-trained\nmodel and a high-quality ground truth. We show that our pruning and fine-tuning\napproach can accelerate image reconstruction using popular deep equilibrium\nlearning (DEQ) and deep unfolding (DU) methods by 50% and 32%, respectively,\nwith nearly no performance loss. This work thus offers a step forward for\nsolving inverse problems by showing the potential of pruning to improve the\nscalability of MBDL. Code is available at\nhttps://github.com/wustl-cig/MBDL_Pruning .\n", "date": "2023-11-03", "categories": []}
{"id": "2410.11247", "title": "A Unified Framework for Forward and Inverse Problems in Subsurface\n  Imaging using Latent Space Translations", "abstract": "  In subsurface imaging, learning the mapping from velocity maps to seismic\nwaveforms (forward problem) and waveforms to velocity (inverse problem) is\nimportant for several applications. While traditional techniques for solving\nforward and inverse problems are computationally prohibitive, there is a\ngrowing interest in leveraging recent advances in deep learning to learn the\nmapping between velocity maps and seismic waveform images directly from data.\nDespite the variety of architectures explored in previous works, several open\nquestions still remain unanswered such as the effect of latent space sizes, the\nimportance of manifold learning, the complexity of translation models, and the\nvalue of jointly solving forward and inverse problems. We propose a unified\nframework to systematically characterize prior research in this area termed the\nGeneralized Forward-Inverse (GFI) framework, building on the assumption of\nmanifolds and latent space translations. We show that GFI encompasses previous\nworks in deep learning for subsurface imaging, which can be viewed as specific\ninstantiations of GFI. We also propose two new model architectures within the\nframework of GFI: Latent U-Net and Invertible X-Net, leveraging the power of\nU-Nets for domain translation and the ability of IU-Nets to simultaneously\nlearn forward and inverse translations, respectively. We show that our proposed\nmodels achieve state-of-the-art (SOTA) performance for forward and inverse\nproblems on a wide range of synthetic datasets, and also investigate their\nzero-shot effectiveness on two real-world-like datasets. Our code is available\nat https://github.com/KGML-lab/Generalized-Forward-Inverse-Framework-for-DL4SI\n", "date": "2024-10-15", "categories": []}
{"id": "2403.07326", "title": "SGE: Structured Light System Based on Gray Code with an Event Camera", "abstract": "  Fast and accurate depth sensing has long been a significant research\nchallenge. Event camera, as a device that quickly responds to intensity\nchanges, provides a new solution for structured light (SL) systems. In this\npaper, we introduce Gray code into event-based SL systems for the first time.\nOur setup includes an event camera and a Digital Light Processing (DLP)\nprojector, enabling depth estimation through high-speed projection and decoding\nof Gray code patterns. By employing Gray code for point matching in event-based\nSL system, our method is immune to timestamp noise, realizing high-speed depth\nestimation without loss of accuracy and spatial resolution. The binary nature\nof events and Gray code minimizes data redundancy, enabling us to fully utilize\nsensor bandwidth at 100%. Experimental results show that our approach achieves\naccuracy comparable to state-of-the-art scanning methods while surpassing them\nin data acquisition speed (up to 41 times improvement) without sacrificing\naccuracy and spatial resolution. Our proposed approach offers a highly\npromising solution for ultra-fast, real-time, and high-precision dense depth\nestimation.\n", "date": "2024-03-12", "categories": []}
{"id": "2407.18456", "title": "Diffusion-driven lensless fiber endomicroscopic quantitative phase\n  imaging towards digital pathology", "abstract": "  Lensless fiber endomicroscope is an emerging tool for in-vivo microscopic\nimaging, where quantitative phase imaging (QPI) can be utilized as a label-free\nmethod to enhance image contrast. However, existing single-shot phase\nreconstruction methods through lensless fiber endomicroscope typically perform\nwell on simple images but struggle with complex microscopic structures. Here,\nwe propose a speckle-conditioned diffusion model (SpecDiffusion), which\nreconstructs phase images directly from speckles captured at the detection side\nof a multi-core fiber (MCF). Unlike conventional neural networks, SpecDiffusion\nemploys iterative phase denoising steps for speckle-driven phase\nreconstruction. The iteration scheme allows SpecDiffusion to break down the\nphase reconstruction process into multiple steps, gradually building up to the\nfinal phase image. This attribute alleviates the computation challenge at each\nstep and enables the reconstruction of rich details in complex microscopic\nimages. To validate its efficacy, we build an optical system to capture\nspeckles from MCF and construct a dataset consisting of 100,000 paired images.\nSpecDiffusion provides high-fidelity phase reconstruction results and shows\npowerful generalization capacity for unseen objects, such as test charts and\nbiological tissues, reducing the average mean absolute error of the\nreconstructed tissue images by 7 times. Furthermore, the reconstructed tissue\nimages using SpecDiffusion shows higher accuracy in zero-shot cell segmentation\ntasks compared to the conventional method, demonstrating the potential for\nfurther cell morphology analysis through the learning-based lensless fiber\nendomicroscope. SpecDiffusion offers a precise and generalized method to phase\nreconstruction through scattering media, including MCFs, opening new\nperspective in lensless fiber endomicroscopic imaging.\n", "date": "2024-07-26", "categories": []}
{"id": "2501.08163", "title": "DH-Mamba: Exploring Dual-domain Hierarchical State Space Models for MRI\n  Reconstruction", "abstract": "  The accelerated MRI reconstruction poses a challenging ill-posed inverse\nproblem due to the significant undersampling in k-space. Deep neural networks,\nsuch as CNNs and ViTs, have shown substantial performance improvements for this\ntask while encountering the dilemma between global receptive fields and\nefficient computation. To this end, this paper explores selective state space\nmodels (Mamba), a new paradigm for long-range dependency modeling with linear\ncomplexity, for efficient and effective MRI reconstruction. However, directly\napplying Mamba to MRI reconstruction faces three significant issues: (1) Mamba\ntypically flattens 2D images into distinct 1D sequences along rows and columns,\ndisrupting k-space's unique spectrum and leaving its potential in k-space\nlearning unexplored. (2) Existing approaches adopt multi-directional lengthy\nscanning to unfold images at the pixel level, leading to long-range forgetting\nand high computational burden. (3) Mamba struggles with spatially-varying\ncontents, resulting in limited diversity of local representations. To address\nthese, we propose a dual-domain hierarchical Mamba for MRI reconstruction from\nthe following perspectives: (1) We pioneer vision Mamba in k-space learning. A\ncircular scanning is customized for spectrum unfolding, benefiting the global\nmodeling of k-space. (2) We propose a hierarchical Mamba with an efficient\nscanning strategy in both image and k-space domains. It mitigates long-range\nforgetting and achieves a better trade-off between efficiency and performance.\n(3) We develop a local diversity enhancement module to improve the\nspatially-varying representation of Mamba. Extensive experiments are conducted\non three public datasets for MRI reconstruction under various undersampling\npatterns. Comprehensive results demonstrate that our method significantly\noutperforms state-of-the-art methods with lower computational cost.\n", "date": "2025-01-14", "categories": []}
{"id": "2503.13147", "title": "Iterative Predictor-Critic Code Decoding for Real-World Image Dehazing", "abstract": "  We propose a novel Iterative Predictor-Critic Code Decoding framework for\nreal-world image dehazing, abbreviated as IPC-Dehaze, which leverages the\nhigh-quality codebook prior encapsulated in a pre-trained VQGAN. Apart from\nprevious codebook-based methods that rely on one-shot decoding, our method\nutilizes high-quality codes obtained in the previous iteration to guide the\nprediction of the Code-Predictor in the subsequent iteration, improving code\nprediction accuracy and ensuring stable dehazing performance. Our idea stems\nfrom the observations that 1) the degradation of hazy images varies with haze\ndensity and scene depth, and 2) clear regions play crucial cues in restoring\ndense haze regions. However, it is non-trivial to progressively refine the\nobtained codes in subsequent iterations, owing to the difficulty in determining\nwhich codes should be retained or replaced at each iteration. Another key\ninsight of our study is to propose Code-Critic to capture interrelations among\ncodes. The Code-Critic is used to evaluate code correlations and then resample\na set of codes with the highest mask scores, i.e., a higher score indicates\nthat the code is more likely to be rejected, which helps retain more accurate\ncodes and predict difficult ones. Extensive experiments demonstrate the\nsuperiority of our method over state-of-the-art methods in real-world dehazing.\n", "date": "2025-03-17", "categories": []}
{"id": "2503.23684", "title": "Detail-aware multi-view stereo network for depth estimation", "abstract": "  Multi-view stereo methods have achieved great success for depth estimation\nbased on the coarse-to-fine depth learning frameworks, however, the existing\nmethods perform poorly in recovering the depth of object boundaries and detail\nregions. To address these issues, we propose a detail-aware multi-view stereo\nnetwork (DA-MVSNet) with a coarse-to-fine framework. The geometric depth clues\nhidden in the coarse stage are utilized to maintain the geometric structural\nrelationships between object surfaces and enhance the expressive capability of\nimage features. In addition, an image synthesis loss is employed to constrain\nthe gradient flow for detailed regions and further strengthen the supervision\nof object boundaries and texture-rich areas. Finally, we propose an adaptive\ndepth interval adjustment strategy to improve the accuracy of object\nreconstruction. Extensive experiments on the DTU and Tanks & Temples datasets\ndemonstrate that our method achieves competitive results. The code is available\nat https://github.com/wsmtht520-/DAMVSNet.\n", "date": "2025-03-31", "categories": []}
{"id": "2503.21907", "title": "KernelFusion: Assumption-Free Blind Super-Resolution via Patch Diffusion", "abstract": "  Traditional super-resolution (SR) methods assume an ``ideal'' downscaling\nSR-kernel (e.g., bicubic downscaling) between the high-resolution (HR) image\nand the low-resolution (LR) image. Such methods fail once the LR images are\ngenerated differently. Current blind-SR methods aim to remove this assumption,\nbut are still fundamentally restricted to rather simplistic downscaling\nSR-kernels (e.g., anisotropic Gaussian kernels), and fail on more complex (out\nof distribution) downscaling degradations. However, using the correct SR-kernel\nis often more important than using a sophisticated SR algorithm. In\n``KernelFusion'' we introduce a zero-shot diffusion-based method that makes no\nassumptions about the kernel. Our method recovers the unique image-specific\nSR-kernel directly from the LR input image, while simultaneously recovering its\ncorresponding HR image. KernelFusion exploits the principle that the correct\nSR-kernel is the one that maximizes patch similarity across different scales of\nthe LR image. We first train an image-specific patch-based diffusion model on\nthe single LR input image, capturing its unique internal patch statistics. We\nthen reconstruct a larger HR image with the same learned patch distribution,\nwhile simultaneously recovering the correct downscaling SR-kernel that\nmaintains this cross-scale relation between the HR and LR images. Empirical\nresults show that KernelFusion vastly outperforms all SR baselines on complex\ndownscaling degradations, where existing SotA Blind-SR methods fail miserably.\nBy breaking free from predefined kernel assumptions, KernelFusion pushes\nBlind-SR into a new assumption-free paradigm, handling downscaling kernels\npreviously thought impossible.\n", "date": "2025-03-27", "categories": []}
{"id": "2410.02619", "title": "GI-GS: Global Illumination Decomposition on Gaussian Splatting for\n  Inverse Rendering", "abstract": "  We present GI-GS, a novel inverse rendering framework that leverages 3D\nGaussian Splatting (3DGS) and deferred shading to achieve photo-realistic novel\nview synthesis and relighting. In inverse rendering, accurately modeling the\nshading processes of objects is essential for achieving high-fidelity results.\nTherefore, it is critical to incorporate global illumination to account for\nindirect lighting that reaches an object after multiple bounces across the\nscene. Previous 3DGS-based methods have attempted to model indirect lighting by\ncharacterizing indirect illumination as learnable lighting volumes or\nadditional attributes of each Gaussian, while using baked occlusion to\nrepresent shadow effects. These methods, however, fail to accurately model the\ncomplex physical interactions between light and objects, making it impossible\nto construct realistic indirect illumination during relighting. To address this\nlimitation, we propose to calculate indirect lighting using efficient path\ntracing with deferred shading. In our framework, we first render a G-buffer to\ncapture the detailed geometry and material properties of the scene. Then, we\nperform physically-based rendering (PBR) only for direct lighting. With the\nG-buffer and previous rendering results, the indirect lighting can be\ncalculated through a lightweight path tracing. Our method effectively models\nindirect lighting under any given lighting conditions, thereby achieving better\nnovel view synthesis and competitive relighting. Quantitative and qualitative\nresults show that our GI-GS outperforms existing baselines in both rendering\nquality and efficiency.\n", "date": "2024-10-03", "categories": []}
{"id": "2412.02798", "title": "Grayscale to Hyperspectral at Any Resolution Using a Phase-Only Lens", "abstract": "  We consider the problem of reconstructing a HxWx31 hyperspectral image from a\nHxW grayscale snapshot measurement that is captured using only a single\ndiffractive optic and a filterless panchromatic photosensor. This problem is\nseverely ill-posed, but we present the first model that produces high-quality\nresults. We make efficient use of limited data by training a conditional\ndenoising diffusion model that operates on small patches in a shift-invariant\nmanner. During inference, we synchronize per-patch hyperspectral predictions\nusing guidance derived from the optical point spread function. Surprisingly,\nour experiments reveal that patch sizes as small as the PSFs support achieve\nexcellent results, and they show that local optical cues are sufficient to\ncapture full spectral information. Moreover, by drawing multiple samples, our\nmodel provides per-pixel uncertainty estimates that strongly correlate with\nreconstruction error. Our work lays the foundation for a new class of\nhigh-resolution snapshot hyperspectral imagers that are compact and\nlight-efficient.\n", "date": "2024-12-03", "categories": []}
{"id": "2305.04268", "title": "MS-NeRF: Multi-Space Neural Radiance Fields", "abstract": "  Existing Neural Radiance Fields (NeRF) methods suffer from the existence of\nreflective objects, often resulting in blurry or distorted rendering. Instead\nof calculating a single radiance field, we propose a multi-space neural\nradiance field (MS-NeRF) that represents the scene using a group of feature\nfields in parallel sub-spaces, which leads to a better understanding of the\nneural network toward the existence of reflective and refractive objects. Our\nmulti-space scheme works as an enhancement to existing NeRF methods, with only\nsmall computational overheads needed for training and inferring the extra-space\noutputs. We design different multi-space modules for representative MLP-based\nand grid-based NeRF methods, which improve Mip-NeRF 360 by 4.15 dB in PSNR with\n0.5% extra parameters and further improve TensoRF by 2.71 dB with 0.046% extra\nparameters on reflective regions without degrading the rendering quality on\nother regions. We further construct a novel dataset consisting of 33 synthetic\nscenes and 7 real captured scenes with complex reflection and refraction, where\nwe design complex camera paths to fully benchmark the robustness of NeRF-based\nmethods. Extensive experiments show that our approach significantly outperforms\nthe existing single-space NeRF methods for rendering high-quality scenes\nconcerned with complex light paths through mirror-like objects. The source\ncode, dataset, and results are available via our project page:\nhttps://zx-yin.github.io/msnerf/.\n", "date": "2023-05-07", "categories": []}
{"id": "2412.04591", "title": "Aberration Correcting Vision Transformers for High-Fidelity Metalens\n  Imaging", "abstract": "  Metalens is an emerging optical system with an irreplaceable merit in that it\ncan be manufactured in ultra-thin and compact sizes, which shows great promise\nin various applications. Despite its advantage in miniaturization, its\npracticality is constrained by spatially varying aberrations and distortions,\nwhich significantly degrade the image quality. Several previous arts have\nattempted to address different types of aberrations, yet most of them are\nmainly designed for the traditional bulky lens and ineffective to remedy harsh\naberrations of the metalens. While there have existed aberration correction\nmethods specifically for metalens, they still fall short of restoration\nquality. In this work, we propose a novel aberration correction framework for\nmetalens-captured images, harnessing Vision Transformers (ViT) that have the\npotential to restore metalens images with non-uniform aberrations.\nSpecifically, we devise a Multiple Adaptive Filters Guidance (MAFG), where\nmultiple Wiener filters enrich the degraded input images with various\nnoise-detail balances and a cross-attention module reweights the features\nconsidering the different degrees of aberrations. In addition, we introduce a\nSpatial and Transposed self-Attention Fusion (STAF) module, which aggregates\nfeatures from spatial self-attention and transposed self-attention modules to\nfurther ameliorate aberration correction. We conduct extensive experiments,\nincluding correcting aberrated images and videos, and clean 3D reconstruction.\nThe proposed method outperforms the previous arts by a significant margin. We\nfurther fabricate a metalens and verify the practicality of our method by\nrestoring the images captured with the manufactured metalens. Code and\npre-trained models are available at\nhttps://benhenryl.github.io/Metalens-Transformer.\n", "date": "2024-12-05", "categories": []}
{"id": "2502.17435", "title": "GCC: Generative Color Constancy via Diffusing a Color Checker", "abstract": "  Color constancy methods often struggle to generalize across different camera\nsensors due to varying spectral sensitivities. We present GCC, which leverages\ndiffusion models to inpaint color checkers into images for illumination\nestimation. Our key innovations include (1) a single-step deterministic\ninference approach that inpaints color checkers reflecting scene illumination,\n(2) a Laplacian decomposition technique that preserves checker structure while\nallowing illumination-dependent color adaptation, and (3) a mask-based data\naugmentation strategy for handling imprecise color checker annotations. By\nharnessing rich priors from pre-trained diffusion models, GCC demonstrates\nstrong robustness in challenging cross-camera scenarios. These results\nhighlight our method's effective generalization capability across different\ncamera characteristics without requiring sensor-specific training, making it a\nversatile and practical solution for real-world applications.\n", "date": "2025-02-24", "categories": []}
{"id": "2503.00260", "title": "Seeing A 3D World in A Grain of Sand", "abstract": "  We present a snapshot imaging technique for recovering 3D surrounding views\nof miniature scenes. Due to their intricacy, miniature scenes with objects\nsized in millimeters are difficult to reconstruct, yet miniatures are common in\nlife and their 3D digitalization is desirable. We design a catadioptric imaging\nsystem with a single camera and eight pairs of planar mirrors for snapshot 3D\nreconstruction from a dollhouse perspective. We place paired mirrors on nested\npyramid surfaces for capturing surrounding multi-view images in a single shot.\nOur mirror design is customizable based on the size of the scene for optimized\nview coverage. We use the 3D Gaussian Splatting (3DGS) representation for scene\nreconstruction and novel view synthesis. We overcome the challenge posed by our\nsparse view input by integrating visual hull-derived depth constraint. Our\nmethod demonstrates state-of-the-art performance on a variety of synthetic and\nreal miniature scenes.\n", "date": "2025-03-01", "categories": []}
{"id": "2503.19253", "title": "$L^2$FMamba: Lightweight Light Field Image Super-Resolution with State\n  Space Model", "abstract": "  Transformers bring significantly improved performance to the light field\nimage super-resolution task due to their long-range dependency modeling\ncapability. However, the inherently high computational complexity of their core\nself-attention mechanism has increasingly hindered their advancement in this\ntask. To address this issue, we first introduce the LF-VSSM block, a novel\nmodule inspired by progressive feature extraction, to efficiently capture\ncritical long-range spatial-angular dependencies in light field images. LF-VSSM\nsuccessively extracts spatial features within sub-aperture images,\nspatial-angular features between sub-aperture images, and spatial-angular\nfeatures between light field image pixels. On this basis, we propose a\nlightweight network, $L^2$FMamba (Lightweight Light Field Mamba), which\nintegrates the LF-VSSM block to leverage light field features for\nsuper-resolution tasks while overcoming the computational challenges of\nTransformer-based approaches. Extensive experiments on multiple light field\ndatasets demonstrate that our method reduces the number of parameters and\ncomplexity while achieving superior super-resolution performance with faster\ninference speed.\n", "date": "2025-03-25", "categories": []}
{"id": "2503.19634", "title": "Burst Image Super-Resolution with Mamba", "abstract": "  Burst image super-resolution (BISR) aims to enhance the resolution of a\nkeyframe by leveraging information from multiple low-resolution images captured\nin quick succession. In the deep learning era, BISR methods have evolved from\nfully convolutional networks to transformer-based architectures, which, despite\ntheir effectiveness, suffer from the quadratic complexity of self-attention. We\nsee Mamba as the next natural step in the evolution of this field, offering a\ncomparable global receptive field and selective information routing with only\nlinear time complexity. In this work, we introduce BurstMamba, a Mamba-based\narchitecture for BISR. Our approach decouples the task into two specialized\nbranches: a spatial module for keyframe super-resolution and a temporal module\nfor subpixel prior extraction, striking a balance between computational\nefficiency and burst information integration. To further enhance burst\nprocessing with Mamba, we propose two novel strategies: (i) optical flow-based\nserialization, which aligns burst sequences only during state updates to\npreserve subpixel details, and (ii) a wavelet-based reparameterization of the\nstate-space update rules, prioritizing high-frequency features for improved\nburst-to-keyframe information passing. Our framework achieves SOTA performance\non public benchmarks of SyntheticSR, RealBSR-RGB, and RealBSR-RAW.\n", "date": "2025-03-25", "categories": []}
{"id": "2310.16102", "title": "Learned, uncertainty-driven adaptive acquisition for photon-efficient\n  scanning microscopy", "abstract": "  Scanning microscopy systems, such as confocal and multiphoton microscopy, are\npowerful imaging tools for probing deep into biological tissue. However,\nscanning systems have an inherent trade-off between acquisition time, field of\nview, phototoxicity, and image quality, often resulting in noisy measurements\nwhen fast, large field of view, and/or gentle imaging is needed. Deep learning\ncould be used to denoise noisy microscopy measurements, but these algorithms\ncan be prone to hallucination, which can be disastrous for medical and\nscientific applications. We propose a method to simultaneously denoise and\npredict pixel-wise uncertainty for scanning microscopy systems, improving\nalgorithm trustworthiness and providing statistical guarantees for deep\nlearning predictions. Furthermore, we propose to leverage this learned,\npixel-wise uncertainty to drive an adaptive acquisition technique that rescans\nonly the most uncertain regions of a sample, saving time and reducing the total\nlight dose to the sample. We demonstrate our method on experimental confocal\nand multiphoton microscopy systems, showing that our uncertainty maps can\npinpoint hallucinations in the deep learned predictions. Finally, with our\nadaptive acquisition technique, we demonstrate up to 16X reduction in\nacquisition time and total light dose while successfully recovering fine\nfeatures in the sample and reducing hallucinations. We are the first to\ndemonstrate distribution-free uncertainty quantification for a denoising task\nwith real experimental data and the first to propose adaptive acquisition based\non reconstruction uncertainty.\n", "date": "2023-10-24", "categories": []}
{"id": "2406.05475", "title": "HDRT: A Large-Scale Dataset for Infrared-Guided HDR Imaging", "abstract": "  Capturing images with enough details to solve imaging tasks is a\nlong-standing challenge in imaging, particularly due to the limitations of\nstandard dynamic range (SDR) images which often lose details in underexposed or\noverexposed regions. Traditional high dynamic range (HDR) methods, like\nmulti-exposure fusion or inverse tone mapping, struggle with ghosting and\nincomplete data reconstruction. Infrared (IR) imaging offers a unique advantage\nby being less affected by lighting conditions, providing consistent detail\ncapture regardless of visible light intensity. In this paper, we introduce the\nHDRT dataset, the first comprehensive dataset that consists of HDR and thermal\nIR images. The HDRT dataset comprises 50,000 images captured across three\nseasons over six months in eight cities, providing a diverse range of lighting\nconditions and environmental contexts. Leveraging this dataset, we propose\nHDRTNet, a novel deep neural method that fuses IR and SDR content to generate\nHDR images. Extensive experiments validate HDRTNet against the\nstate-of-the-art, showing substantial quantitative and qualitative quality\nimprovements. The HDRT dataset not only advances IR-guided HDR imaging but also\noffers significant potential for broader research in HDR imaging, multi-modal\nfusion, domain transfer, and beyond. The dataset is available at\nhttps://huggingface.co/datasets/jingchao-peng/HDRTDataset.\n", "date": "2024-06-08", "categories": []}
{"id": "2407.07052", "title": "Latent Space Imaging", "abstract": "  Digital imaging systems have traditionally relied on brute-force measurement\nand processing of pixels arranged on regular grids. In contrast, the human\nvisual system performs significant data reduction from the large number of\nphotoreceptors to the optic nerve, effectively encoding visual information into\na low-bandwidth latent space representation optimized for brain processing.\nInspired by this, we propose a similar approach to advance artificial vision\nsystems. Latent Space Imaging introduces a new paradigm that combines optics\nand software to encode image information directly into the semantically rich\nlatent space of a generative model. This approach substantially reduces\nbandwidth and memory demands during image capture and enables a range of\ndownstream tasks focused on the latent space. We validate this principle\nthrough an initial hardware prototype based on a single-pixel camera. By\nimplementing an amplitude modulation scheme that encodes into the generative\nmodel's latent space, we achieve compression ratios ranging from 1:100 to\n1:1000 during imaging, and up to 1:16384 for downstream applications. This\napproach leverages the model's intrinsic linear boundaries, demonstrating the\npotential of latent space imaging for highly efficient imaging hardware,\nadaptable future applications in high-speed imaging, and task-specific cameras\nwith significantly reduced hardware complexity.\n", "date": "2024-07-09", "categories": []}
{"id": "2412.20066", "title": "MaIR: A Locality- and Continuity-Preserving Mamba for Image Restoration", "abstract": "  Recent advancements in Mamba have shown promising results in image\nrestoration. These methods typically flatten 2D images into multiple distinct\n1D sequences along rows and columns, process each sequence independently using\nselective scan operation, and recombine them to form the outputs. However, such\na paradigm overlooks two vital aspects: i) the local relationships and spatial\ncontinuity inherent in natural images, and ii) the discrepancies among\nsequences unfolded through totally different ways. To overcome the drawbacks,\nwe explore two problems in Mamba-based restoration methods: i) how to design a\nscanning strategy preserving both locality and continuity while facilitating\nrestoration, and ii) how to aggregate the distinct sequences unfolded in\ntotally different ways. To address these problems, we propose a novel\nMamba-based Image Restoration model (MaIR), which consists of Nested S-shaped\nScanning strategy (NSS) and Sequence Shuffle Attention block (SSA).\nSpecifically, NSS preserves locality and continuity of the input images through\nthe stripe-based scanning region and the S-shaped scanning path, respectively.\nSSA aggregates sequences through calculating attention weights within the\ncorresponding channels of different sequences. Thanks to NSS and SSA, MaIR\nsurpasses 40 baselines across 14 challenging datasets, achieving\nstate-of-the-art performance on the tasks of image super-resolution, denoising,\ndeblurring and dehazing. The code is available at\nhttps://github.com/XLearning-SCU/2025-CVPR-MaIR.\n", "date": "2024-12-28", "categories": []}
{"id": "2503.17915", "title": "Cat-AIR: Content and Task-Aware All-in-One Image Restoration", "abstract": "  All-in-one image restoration seeks to recover high-quality images from\nvarious types of degradation using a single model, without prior knowledge of\nthe corruption source. However, existing methods often struggle to effectively\nand efficiently handle multiple degradation types. We present Cat-AIR, a novel\n\\textbf{C}ontent \\textbf{A}nd \\textbf{T}ask-aware framework for\n\\textbf{A}ll-in-one \\textbf{I}mage \\textbf{R}estoration. Cat-AIR incorporates\nan alternating spatial-channel attention mechanism that adaptively balances the\nlocal and global information for different tasks. Specifically, we introduce\ncross-layer channel attentions and cross-feature spatial attentions that\nallocate computations based on content and task complexity. Furthermore, we\npropose a smooth learning strategy that allows for seamless adaptation to new\nrestoration tasks while maintaining performance on existing ones. Extensive\nexperiments demonstrate that Cat-AIR achieves state-of-the-art results across a\nwide range of restoration tasks, requiring fewer FLOPs than previous methods,\nestablishing new benchmarks for efficient all-in-one image restoration.\n", "date": "2025-03-23", "categories": []}
{"id": "2503.17992", "title": "Geometric Constrained Non-Line-of-Sight Imaging", "abstract": "  Normal reconstruction is crucial in non-line-of-sight (NLOS) imaging, as it\nprovides key geometric and lighting information about hidden objects, which\nsignificantly improves reconstruction accuracy and scene understanding.\nHowever, jointly estimating normals and albedo expands the problem from\nmatrix-valued functions to tensor-valued functions that substantially\nincreasing complexity and computational difficulty. In this paper, we propose a\nnovel joint albedo-surface reconstruction method, which utilizes the Frobenius\nnorm of the shape operator to control the variation rate of the normal field.\nIt is the first attempt to apply regularization methods to the reconstruction\nof surface normals for hidden objects. By improving the accuracy of the normal\nfield, it enhances detail representation and achieves high-precision\nreconstruction of hidden object geometry. The proposed method demonstrates\nrobustness and effectiveness on both synthetic and experimental datasets. On\ntransient data captured within 15 seconds, our surface normal-regularized\nreconstruction model produces more accurate surfaces than recently proposed\nmethods and is 30 times faster than the existing surface reconstruction\napproach.\n", "date": "2025-03-23", "categories": []}
{"id": "2503.18038", "title": "Multiple-Particle Autofocusing Algorithm Using Axial Resolution and\n  Morphological Analyses Based on Digital Holography", "abstract": "  We propose an autofocusing algorithm to obtain, relatively accurately, the 3D\nposition of each particle, particularly its axial location, and particle number\nof a dense transparent particle solution via its hologram. First, morphological\nanalyses and constrained intensity are used on raw reconstructed images to\nobtain information on candidate focused particles. Second, axial resolution is\nused to obtain the real focused particles. Based on the mean intensity and\nequivalent diameter of each candidate focused particle, all focused particles\nare eventually secured. Our proposed method can rapidly provide relatively\naccurate ground-truth axial positions to solve the autofocusing problem that\noccurs with dense particles.\n", "date": "2025-03-23", "categories": []}
{"id": "2503.18055", "title": "PolarFree: Polarization-based Reflection-free Imaging", "abstract": "  Reflection removal is challenging due to complex light interactions, where\nreflections obscure important details and hinder scene understanding.\nPolarization naturally provides a powerful cue to distinguish between reflected\nand transmitted light, enabling more accurate reflection removal. However,\nexisting methods often rely on small-scale or synthetic datasets, which fail to\ncapture the diversity and complexity of real-world scenarios. To this end, we\nconstruct a large-scale dataset, PolaRGB, for Polarization-based reflection\nremoval of RGB images, which enables us to train models that generalize\neffectively across a wide range of real-world scenarios. The PolaRGB dataset\ncontains 6,500 well-aligned mixed-transmission image pairs, 8x larger than\nexisting polarization datasets, and is the first to include both RGB and\npolarization images captured across diverse indoor and outdoor environments\nwith varying lighting conditions. Besides, to fully exploit the potential of\npolarization cues for reflection removal, we introduce PolarFree, which\nleverages diffusion process to generate reflection-free cues for accurate\nreflection removal. Extensive experiments show that PolarFree significantly\nenhances image clarity in challenging reflective scenarios, setting a new\nbenchmark for polarized imaging and reflection removal. Code and dataset are\navailable at https://github.com/mdyao/PolarFree.\n", "date": "2025-03-23", "categories": []}
{"id": "2503.18341", "title": "PS-EIP: Robust Photometric Stereo Based on Event Interval Profile", "abstract": "  Recently, the energy-efficient photometric stereo method using an event\ncamera has been proposed to recover surface normals from events triggered by\nchanges in logarithmic Lambertian reflections under a moving directional light\nsource. However, EventPS treats each event interval independently, making it\nsensitive to noise, shadows, and non-Lambertian reflections. This paper\nproposes Photometric Stereo based on Event Interval Profile (PS-EIP), a robust\nmethod that recovers pixelwise surface normals from a time-series profile of\nevent intervals. By exploiting the continuity of the profile and introducing an\noutlier detection method based on profile shape, our approach enhances\nrobustness against outliers from shadows and specular reflections. Experiments\nusing real event data from 3D-printed objects demonstrate that PS-EIP\nsignificantly improves robustness to outliers compared to EventPS's\ndeep-learning variant, EventPS-FCN, without relying on deep learning.\n", "date": "2025-03-24", "categories": []}
{"id": "2503.18705", "title": "Benchmarking Burst Super-Resolution for Polarization Images: Noise\n  Dataset and Analysis", "abstract": "  Snapshot polarization imaging calculates polarization states from linearly\npolarized subimages. To achieve this, a polarization camera employs a double\nBayer-patterned sensor to capture both color and polarization. It demonstrates\nlow light efficiency and low spatial resolution, resulting in increased noise\nand compromised polarization measurements. Although burst super-resolution\neffectively reduces noise and enhances spatial resolution, applying it to\npolarization imaging poses challenges due to the lack of tailored datasets and\nreliable ground truth noise statistics. To address these issues, we introduce\nPolarNS and PolarBurstSR, two innovative datasets developed specifically for\npolarization imaging. PolarNS provides characterization of polarization noise\nstatistics, facilitating thorough analysis, while PolarBurstSR functions as a\nbenchmark for burst super-resolution in polarization images. These datasets,\ncollected under various real-world conditions, enable comprehensive evaluation.\nAdditionally, we present a model for analyzing polarization noise to quantify\nnoise propagation, tested on a large dataset captured in a darkroom\nenvironment. As part of our application, we compare the latest burst\nsuper-resolution models, highlighting the advantages of training tailored to\npolarization compared to RGB-based methods. This work establishes a benchmark\nfor polarization burst super-resolution and offers critical insights into noise\npropagation, thereby enhancing polarization image reconstruction.\n", "date": "2025-03-24", "categories": []}
{"id": "2403.10346", "title": "End-to-end Adaptive Dynamic Subsampling and Reconstruction for Cardiac\n  MRI", "abstract": "  $\\textbf{Background:}$ Accelerating dynamic MRI is vital for advancing\nclinical applications and improving patient comfort. Commonly, deep learning\n(DL) methods for accelerated dynamic MRI reconstruction typically rely on\nuniformly applying non-adaptive predetermined or random subsampling patterns\nacross all temporal frames of the dynamic acquisition. This approach fails to\nexploit temporal correlations or optimize subsampling on a case-by-case basis.\n  $\\textbf{Purpose:}$ To develop an end-to-end approach for adaptive dynamic\nMRI subsampling and reconstruction, capable of generating customized sampling\npatterns maximizing at the same time reconstruction quality.\n  $\\textbf{Methods:}$ We introduce the End-to-end Adaptive Dynamic Sampling and\nReconstruction (E2E-ADS-Recon) for MRI framework, which integrates an adaptive\ndynamic sampler (ADS) that adapts the acquisition trajectory to each case for a\ngiven acceleration factor with a state-of-the-art dynamic reconstruction\nnetwork, vSHARP, for reconstructing the adaptively sampled data into a dynamic\nimage. The ADS can produce either frame-specific patterns or unified patterns\napplied to all temporal frames. E2E-ADS-Recon is evaluated under both\nframe-specific and unified 1D or 2D sampling settings, using dynamic cine\ncardiac MRI data and compared with vSHARP models employing standard subsampling\ntrajectories, as well as pipelines where ADS was replaced by parameterized\nsamplers optimized for dataset-specific schemes.\n  $\\textbf{Results:}$ E2E-ADS-Recon exhibited superior reconstruction quality,\nespecially at high accelerations, in terms of standard quantitative metrics\n(SSIM, pSNR, NMSE).\n  $\\textbf{Conclusion:}$ The proposed framework improves reconstruction\nquality, highlighting the importance of case-specific subsampling optimization\nin dynamic MRI applications.\n", "date": "2024-03-15", "categories": []}
{"id": "2503.16690", "title": "Making the unmodulated pyramid wavefront sensor smart II. First on-sky\n  demonstration of extreme adaptive optics with deep learning", "abstract": "  Pyramid wavefront sensors (PWFSs) are the preferred choice for current and\nfuture extreme adaptive optics (XAO) systems. Almost all instruments use the\nPWFS in its modulated form to mitigate its limited linearity range. However,\nthis modulation comes at the cost of a reduction in sensitivity, a blindness to\npetal-piston modes, and a limit to the sensor's ability to operate at high\nspeeds. Therefore, there is strong interest to use the PWFS without modulation,\nwhich can be enabled with nonlinear reconstructors. Here, we present the first\non-sky demonstration of XAO with an unmodulated PWFS using a nonlinear\nreconstructor based on convolutional neural networks. We discuss the real-time\nimplementation on the Magellan Adaptive Optics eXtreme (MagAO-X) instrument\nusing the optimized TensorRT framework and show that inference is fast enough\nto run the control loop at >2 kHz frequencies. Our on-sky results demonstrate a\nsuccessful closed-loop operation using a model calibrated with internal source\ndata that delivers stable and robust correction under varying conditions.\nPerformance analysis reveals that our smart PWFS achieves nearly the same\nStrehl ratio as the highly optimized modulated PWFS under favorable conditions\non bright stars. Notably, we observe an improvement in performance on a fainter\nstar under the influence of strong winds. These findings confirm the\nfeasibility of using the PWFS in its unmodulated form and highlight its\npotential for next-generation instruments. Future efforts will focus on\nachieving even higher control loop frequencies (>3 kHz), optimizing the\ncalibration procedures, and testing its performance on fainter stars, where\nmore gain is expected for the unmodulated PWFS compared to its modulated\ncounterpart.\n", "date": "2025-03-20", "categories": []}
{"id": "2503.16930", "title": "Vision-Language Gradient Descent-driven All-in-One Deep Unfolding\n  Networks", "abstract": "  Dynamic image degradations, including noise, blur and lighting\ninconsistencies, pose significant challenges in image restoration, often due to\nsensor limitations or adverse environmental conditions. Existing Deep Unfolding\nNetworks (DUNs) offer stable restoration performance but require manual\nselection of degradation matrices for each degradation type, limiting their\nadaptability across diverse scenarios. To address this issue, we propose the\nVision-Language-guided Unfolding Network (VLU-Net), a unified DUN framework for\nhandling multiple degradation types simultaneously. VLU-Net leverages a\nVision-Language Model (VLM) refined on degraded image-text pairs to align image\nfeatures with degradation descriptions, selecting the appropriate transform for\ntarget degradation. By integrating an automatic VLM-based gradient estimation\nstrategy into the Proximal Gradient Descent (PGD) algorithm, VLU-Net\neffectively tackles complex multi-degradation restoration tasks while\nmaintaining interpretability. Furthermore, we design a hierarchical feature\nunfolding structure to enhance VLU-Net framework, efficiently synthesizing\ndegradation patterns across various levels. VLU-Net is the first all-in-one DUN\nframework and outperforms current leading one-by-one and all-in-one end-to-end\nmethods by 3.74 dB on the SOTS dehazing dataset and 1.70 dB on the Rain100L\nderaining dataset.\n", "date": "2025-03-21", "categories": []}
{"id": "2503.17269", "title": "Recovering Pulse Waves from Video Using Deep Unrolling and Deep\n  Equilibrium Models", "abstract": "  Camera-based monitoring of vital signs, also known as imaging\nphotoplethysmography (iPPG), has seen applications in driver-monitoring,\nperfusion assessment in surgical settings, affective computing, and more. iPPG\ninvolves sensing the underlying cardiac pulse from video of the skin and\nestimating vital signs such as the heart rate or a full pulse waveform. Some\nprevious iPPG methods impose model-based sparse priors on the pulse signals and\nuse iterative optimization for pulse wave recovery, while others use end-to-end\nblack-box deep learning methods. In contrast, we introduce methods that combine\nsignal processing and deep learning methods in an inverse problem framework.\nOur methods estimate the underlying pulse signal and heart rate from facial\nvideo by learning deep-network-based denoising operators that leverage deep\nalgorithm unfolding and deep equilibrium models. Experiments show that our\nmethods can denoise an acquired signal from the face and infer the correct\nunderlying pulse rate, achieving state-of-the-art heart rate estimation\nperformance on well-known benchmarks, all with less than one-fifth the number\nof learnable parameters as the closest competing method.\n", "date": "2025-03-21", "categories": []}
{"id": "2209.12075", "title": "S^2-Transformer for Mask-Aware Hyperspectral Image Reconstruction", "abstract": "  Snapshot compressive imaging (SCI) surges as a novel way of capturing\nhyperspectral images. It operates an optical encoder to compress the 3D data\ninto a 2D measurement and adopts a software decoder for the signal\nreconstruction. Recently, a representative SCI set-up of coded aperture\nsnapshot compressive imager (CASSI) with Transformer reconstruction backend\nremarks high-fidelity sensing performance. However, dominant spatial and\nspectral attention designs show limitations in hyperspectral modeling. The\nspatial attention values describe the inter-pixel correlation but overlook the\nacross-spectra variation within each pixel. The spectral attention size is\nunscalable to the token spatial size and thus bottlenecks information\nallocation. Besides, CASSI entangles the spatial and spectral information into\na 2D measurement, placing a barrier for information disentanglement and\nmodeling. In addition, CASSI blocks the light with a physical binary mask,\nyielding the masked data loss. To tackle above challenges, we propose a\nspatial-spectral (S2-) Transformer implemented by a paralleled attention design\nand a mask-aware learning strategy. Firstly, we systematically explore pros and\ncons of different spatial (-spectral) attention designs, based on which we find\nperforming both attentions in parallel well disentangles and models the blended\ninformation. Secondly, the masked pixels induce higher prediction difficulty\nand should be treated differently from unmasked ones. We adaptively prioritize\nthe loss penalty attributing to the mask structure by referring to the\nmask-encoded prediction as an uncertainty estimator. We theoretically discuss\nthe distinct convergence tendencies between masked/unmasked regions of the\nproposed learning strategy. Extensive experiments demonstrate that on average,\nthe results of the proposed method are superior over the state-of-the-art\nmethod.\n", "date": "2022-09-24", "categories": []}
{"id": "2306.01176", "title": "Cooperative Hardware-Prompt Learning for Snapshot Compressive Imaging", "abstract": "  Existing reconstruction models in snapshot compressive imaging systems (SCI)\nare trained with a single well-calibrated hardware instance, making their\nperformance vulnerable to hardware shifts and limited in adapting to multiple\nhardware configurations. To facilitate cross-hardware learning, previous\nefforts attempt to directly collect multi-hardware data and perform centralized\ntraining, which is impractical due to severe user data privacy concerns and\nhardware heterogeneity across different platforms/institutions. In this study,\nwe explicitly consider data privacy and heterogeneity in cooperatively\noptimizing SCI systems by proposing a Federated Hardware-Prompt learning\n(FedHP) framework. Rather than mitigating the client drift by rectifying the\ngradients, which only takes effect on the learning manifold but fails to solve\nthe heterogeneity rooted in the input data space, FedHP learns a\nhardware-conditioned prompter to align inconsistent data distribution across\nclients, serving as an indicator of the data inconsistency among different\nhardware (e.g., coded apertures). Extensive experimental results demonstrate\nthat the proposed FedHP coordinates the pre-trained model to multiple hardware\nconfigurations, outperforming prevalent FL frameworks for 0.35dB under\nchallenging heterogeneous settings. Moreover, a Snapshot Spectral Heterogeneous\nDataset has been built upon multiple practical SCI systems. Data and code are\naveilable at https://github.com/Jiamian-Wang/FedHP-Snapshot-Compressive-Imaging\n", "date": "2023-06-01", "categories": []}
{"id": "2503.14892", "title": "Degradation Alchemy: Self-Supervised Unknown-to-Known Transformation for\n  Blind Hyperspectral Image Fusion", "abstract": "  Hyperspectral image (HSI) fusion is an efficient technique that combines\nlow-resolution HSI (LR-HSI) and high-resolution multispectral images (HR-MSI)\nto generate high-resolution HSI (HR-HSI). Existing supervised learning methods\n(SLMs) can yield promising results when test data degradation matches the\ntraining ones, but they face challenges in generalizing to unknown\ndegradations. To unleash the potential and generalization ability of SLMs, we\npropose a novel self-supervised unknown-to-known degradation transformation\nframework (U2K) for blind HSI fusion, which adaptively transforms unknown\ndegradation into the same type of degradation as those handled by pre-trained\nSLMs. Specifically, the proposed U2K framework consists of: (1) spatial and\nspectral Degradation Wrapping (DW) modules that map HR-HSI to unknown degraded\nHR-MSI and LR-HSI, and (2) Degradation Transformation (DT) modules that convert\nthese wrapped data into predefined degradation patterns. The transformed HR-MSI\nand LR-HSI pairs are then processed by a pre-trained network to reconstruct the\ntarget HR-HSI. We train the U2K framework in a self-supervised manner using\nconsistency loss and greedy alternating optimization, significantly improving\nthe flexibility of blind HSI fusion. Extensive experiments confirm the\neffectiveness of our proposed U2K framework in boosting the adaptability of\nfive existing SLMs under various degradation settings and surpassing\nstate-of-the-art blind methods.\n", "date": "2025-03-19", "categories": []}
{"id": "2312.01061", "title": "Spectral-wise Implicit Neural Representation for Hyperspectral Image\n  Reconstruction", "abstract": "  Coded Aperture Snapshot Spectral Imaging (CASSI) reconstruction aims to\nrecover the 3D spatial-spectral signal from 2D measurement. Existing methods\nfor reconstructing Hyperspectral Image (HSI) typically involve learning\nmappings from a 2D compressed image to a predetermined set of discrete spectral\nbands. However, this approach overlooks the inherent continuity of the spectral\ninformation. In this study, we propose an innovative method called\nSpectral-wise Implicit Neural Representation (SINR) as a pioneering step toward\naddressing this limitation. SINR introduces a continuous spectral amplification\nprocess for HSI reconstruction, enabling spectral super-resolution with\ncustomizable magnification factors. To achieve this, we leverage the concept of\nimplicit neural representation. Specifically, our approach introduces a\nspectral-wise attention mechanism that treats individual channels as distinct\ntokens, thereby capturing global spectral dependencies. Additionally, our\napproach incorporates two components, namely a Fourier coordinate encoder and a\nspectral scale factor module. The Fourier coordinate encoder enhances the\nSINR's ability to emphasize high-frequency components, while the spectral scale\nfactor module guides the SINR to adapt to the variable number of spectral\nchannels. Notably, the SINR framework enhances the flexibility of CASSI\nreconstruction by accommodating an unlimited number of spectral bands in the\ndesired output. Extensive experiments demonstrate that our SINR outperforms\nbaseline methods. By enabling continuous reconstruction within the CASSI\nframework, we take the initial stride toward integrating implicit neural\nrepresentation into the field.\n", "date": "2023-12-02", "categories": []}
{"id": "2410.21301", "title": "Evaluating the Posterior Sampling Ability of Plug&Play Diffusion Methods\n  in Sparse-View CT", "abstract": "  Plug&Play (PnP) diffusion models are state-of-the-art methods in computed\ntomography (CT) reconstruction. Such methods usually consider applications\nwhere the sinogram contains a sufficient amount of information for the\nposterior distribution to be concentrated around a single mode, and\nconsequently are evaluated using image-to-image metrics such as PSNR/SSIM.\nInstead, we are interested in reconstructing compressible flow images from\nsinograms having a small number of projections, which results in a posterior\ndistribution no longer concentrated or even multimodal. Thus, in this paper, we\naim at evaluating the approximate posterior of PnP diffusion models and\nintroduce two posterior evaluation properties. We quantitatively evaluate three\nPnP diffusion methods on three different datasets for several numbers of\nprojections. We surprisingly find that, for each method, the approximate\nposterior deviates from the true posterior when the number of projections\ndecreases.\n", "date": "2024-10-21", "categories": []}
{"id": "2503.04843", "title": "Self-Supervised Z-Slice Augmentation for 3D Bio-Imaging via Knowledge\n  Distillation", "abstract": "  Three-dimensional biological microscopy has significantly advanced our\nunderstanding of complex biological structures. However, limitations due to\nmicroscopy techniques, sample properties or phototoxicity often result in poor\nz-resolution, hindering accurate cellular measurements. Here, we introduce\nZAugNet, a fast, accurate, and self-supervised deep learning method for\nenhancing z-resolution in biological images. By performing nonlinear\ninterpolation between consecutive slices, ZAugNet effectively doubles\nresolution with each iteration. Compared on several microscopy modalities and\nbiological objects, it outperforms competing methods on most metrics. Our\nmethod leverages a generative adversarial network (GAN) architecture combined\nwith knowledge distillation to maximize prediction speed without compromising\naccuracy. We also developed ZAugNet+, an extended version enabling continuous\ninterpolation at arbitrary distances, making it particularly useful for\ndatasets with nonuniform slice spacing. Both ZAugNet and ZAugNet+ provide\nhigh-performance, scalable z-slice augmentation solutions for large-scale 3D\nimaging. They are available as open-source frameworks in PyTorch, with an\nintuitive Colab notebook interface for easy access by the scientific community.\n", "date": "2025-03-05", "categories": []}
{"id": "2503.14037", "title": "Intra and Inter Parser-Prompted Transformers for Effective Image\n  Restoration", "abstract": "  We propose Intra and Inter Parser-Prompted Transformers (PPTformer) that\nexplore useful features from visual foundation models for image restoration.\nSpecifically, PPTformer contains two parts: an Image Restoration Network\n(IRNet) for restoring images from degraded observations and a Parser-Prompted\nFeature Generation Network (PPFGNet) for providing IRNet with reliable parser\ninformation to boost restoration. To enhance the integration of the parser\nwithin IRNet, we propose Intra Parser-Prompted Attention (IntraPPA) and Inter\nParser-Prompted Attention (InterPPA) to implicitly and explicitly learn useful\nparser features to facilitate restoration. The IntraPPA re-considers cross\nattention between parser and restoration features, enabling implicit perception\nof the parser from a long-range and intra-layer perspective. Conversely, the\nInterPPA initially fuses restoration features with those of the parser,\nfollowed by formulating these fused features within an attention mechanism to\nexplicitly perceive parser information. Further, we propose a parser-prompted\nfeed-forward network to guide restoration within pixel-wise gating modulation.\nExperimental results show that PPTformer achieves state-of-the-art performance\non image deraining, defocus deblurring, desnowing, and low-light enhancement.\n", "date": "2025-03-18", "categories": []}
{"id": "2503.14439", "title": "Graph-CNNs for RF Imaging: Learning the Electric Field Integral\n  Equations", "abstract": "  Radio-Frequency (RF) imaging concerns the digital recreation of the surfaces\nof scene objects based on the scattered field at distributed receivers. To\nsolve this difficult inverse scattering problems, data-driven methods are often\nemployed that extract patterns from similar training examples, while offering\nminimal latency. In this paper, we first provide an approximate yet fast\nelectromagnetic model, which is based on the electric field integral equations,\nfor data generation, and subsequently propose a Deep Neural Network (DNN)\narchitecture to learn the corresponding inverse model. A graph-attention\nbackbone allows for the system geometry to be passed to the DNN, where residual\nconvolutional layers extract features about the objects, while a UNet head\nperforms the final image reconstruction. Our quantitative and qualitative\nevaluations on two synthetic data sets of different characteristics showcase\nthe performance gains of thee proposed advanced architecture and its relative\nresilience to signal noise levels and various reception configurations.\n", "date": "2025-03-18", "categories": []}
{"id": "2503.14463", "title": "SIR-DIFF: Sparse Image Sets Restoration with Multi-View Diffusion Model", "abstract": "  The computer vision community has developed numerous techniques for digitally\nrestoring true scene information from single-view degraded photographs, an\nimportant yet extremely ill-posed task. In this work, we tackle image\nrestoration from a different perspective by jointly denoising multiple\nphotographs of the same scene. Our core hypothesis is that degraded images\ncapturing a shared scene contain complementary information that, when combined,\nbetter constrains the restoration problem. To this end, we implement a powerful\nmulti-view diffusion model that jointly generates uncorrupted views by\nextracting rich information from multi-view relationships. Our experiments show\nthat our multi-view approach outperforms existing single-view image and even\nvideo-based methods on image deblurring and super-resolution tasks. Critically,\nour model is trained to output 3D consistent images, making it a promising tool\nfor applications requiring robust multi-view integration, such as 3D\nreconstruction or pose estimation.\n", "date": "2025-03-18", "categories": []}
{"id": "2405.00027", "title": "Multidimensional Compressed Sensing for Spectral Light Field Imaging", "abstract": "  This paper considers a compressive multi-spectral light field camera model\nthat utilizes a one-hot spectralcoded mask and a microlens array to capture\nspatial, angular, and spectral information using a single monochrome sensor. We\npropose a model that employs compressed sensing techniques to reconstruct the\ncomplete multi-spectral light field from undersampled measurements. Unlike\nprevious work where a light field is vectorized to a 1D signal, our method\nemploys a 5D basis and a novel 5D measurement model, hence, matching the\nintrinsic dimensionality of multispectral light fields. We mathematically and\nempirically show the equivalence of 5D and 1D sensing models, and most\nimportantly that the 5D framework achieves orders of magnitude faster\nreconstruction while requiring a small fraction of the memory. Moreover, our\nnew multidimensional sensing model opens new research directions for designing\nefficient visual data acquisition algorithms and hardware.\n", "date": "2024-02-27", "categories": []}
{"id": "2412.13070", "title": "Learning of Patch-Based Smooth-Plus-Sparse Models for Image\n  Reconstruction", "abstract": "  We aim at the solution of inverse problems in imaging, by combining a\npenalized sparse representation of image patches with an unconstrained smooth\none. This allows for a straightforward interpretation of the reconstruction. We\nformulate the optimization as a bilevel problem. The inner problem deploys\nclassical algorithms while the outer problem optimizes the dictionary and the\nregularizer parameters through supervised learning. The process is carried out\nvia implicit differentiation and gradient-based optimization. We evaluate our\nmethod for denoising, super-resolution, and compressed-sensing\nmagnetic-resonance imaging. We compare it to other classical models as well as\ndeep-learning-based methods and show that it always outperforms the former and\nalso the latter in some instances.\n", "date": "2024-12-17", "categories": []}
{"id": "2503.09403", "title": "Multi-Agent Image Restoration", "abstract": "  Image restoration (IR) is challenging due to the complexity of real-world\ndegradations. While many specialized and all-in-one IR models have been\ndeveloped, they fail to effectively handle complex, mixed degradations. Recent\nagentic methods RestoreAgent and AgenticIR leverage intelligent, autonomous\nworkflows to alleviate this issue, yet they suffer from suboptimal results and\ninefficiency due to their resource-intensive finetunings, and ineffective\nsearches and tool execution trials for satisfactory outputs. In this paper, we\npropose MAIR, a novel Multi-Agent approach for complex IR problems. We\nintroduce a real-world degradation prior, categorizing degradations into three\ntypes: (1) scene, (2) imaging, and (3) compression, which are observed to occur\nsequentially in real world, and reverse them in the opposite order. Built upon\nthis three-stage restoration framework, MAIR emulates a team of collaborative\nhuman specialists, including a \"scheduler\" for overall planning and multiple\n\"experts\" dedicated to specific degradations. This design minimizes search\nspace and trial efforts, improving image quality while reducing inference\ncosts. In addition, a registry mechanism is introduced to enable easy\nintegration of new tools. Experiments on both synthetic and real-world datasets\nshow that proposed MAIR achieves competitive performance and improved\nefficiency over the previous agentic IR system. Code and models will be made\navailable.\n", "date": "2025-03-12", "categories": []}
{"id": "2503.12783", "title": "Mixed-granularity Implicit Representation for Continuous Hyperspectral\n  Compressive Reconstruction", "abstract": "  Hyperspectral Images (HSIs) are crucial across numerous fields but are\nhindered by the long acquisition times associated with traditional\nspectrometers. The Coded Aperture Snapshot Spectral Imaging (CASSI) system\nmitigates this issue through a compression technique that accelerates the\nacquisition process. However, reconstructing HSIs from compressed data presents\nchallenges due to fixed spatial and spectral resolution constraints. This study\nintroduces a novel method using implicit neural representation for continuous\nhyperspectral image reconstruction. We propose the Mixed Granularity Implicit\nRepresentation (MGIR) framework, which includes a Hierarchical Spectral-Spatial\nImplicit Encoder for efficient multi-scale implicit feature extraction. This is\ncomplemented by a Mixed-Granularity Local Feature Aggregator that adaptively\nintegrates local features across scales, combined with a decoder that merges\ncoordinate information for precise reconstruction. By leveraging implicit\nneural representations, the MGIR framework enables reconstruction at any\ndesired spatial-spectral resolution, significantly enhancing the flexibility\nand adaptability of the CASSI system. Extensive experimental evaluations\nconfirm that our model produces reconstructed images at arbitrary resolutions\nand matches state-of-the-art methods across varying spectral-spatial\ncompression ratios. The code will be released at https://github.com/chh11/MGIR.\n", "date": "2025-03-17", "categories": []}
{"id": "2503.13241", "title": "Sampling Innovation-Based Adaptive Compressive Sensing", "abstract": "  Scene-aware Adaptive Compressive Sensing (ACS) has attracted significant\ninterest due to its promising capability for efficient and high-fidelity\nacquisition of scene images. ACS typically prescribes adaptive sampling\nallocation (ASA) based on previous samples in the absence of ground truth.\nHowever, when confronting unknown scenes, existing ACS methods often lack\naccurate judgment and robust feedback mechanisms for ASA, thus limiting the\nhigh-fidelity sensing of the scene. In this paper, we introduce a Sampling\nInnovation-Based ACS (SIB-ACS) method that can effectively identify and\nallocate sampling to challenging image reconstruction areas, culminating in\nhigh-fidelity image reconstruction. An innovation criterion is proposed to\njudge ASA by predicting the decrease in image reconstruction error attributable\nto sampling increments, thereby directing more samples towards regions where\nthe reconstruction error diminishes significantly. A sampling innovation-guided\nmulti-stage adaptive sampling (AS) framework is proposed, which iteratively\nrefines the ASA through a multi-stage feedback process. For image\nreconstruction, we propose a Principal Component Compressed Domain Network\n(PCCD-Net), which efficiently and faithfully reconstructs images under AS\nscenarios. Extensive experiments demonstrate that the proposed SIB-ACS method\nsignificantly outperforms the state-of-the-art methods in terms of image\nreconstruction fidelity and visual effects. Codes are available at\nhttps://github.com/giant-pandada/SIB-ACS_CVPR2025.\n", "date": "2025-03-17", "categories": []}
{"id": "2503.13354", "title": "Parameter-free structure-texture image decomposition by unrolling", "abstract": "  In this work, we propose a parameter-free and efficient method to tackle the\nstructure-texture image decomposition problem. In particular, we present a\nneural network LPR-NET based on the unrolling of the Low Patch Rank model. On\nthe one hand, this allows us to automatically learn parameters from data, and\non the other hand to be computationally faster while obtaining qualitatively\nsimilar results compared to traditional iterative model-based methods.\nMoreover, despite being trained on synthetic images, numerical experiments show\nthe ability of our network to generalize well when applied to natural images.\n", "date": "2025-03-17", "categories": []}
{"id": "2409.15784", "title": "Deep-learning real-time phase retrieval of imperfect diffraction\n  patterns from X-ray free-electron lasers", "abstract": "  Machine learning is attracting surging interest across nearly all scientific\nareas by enabling the analysis of large datasets and the extraction of\nscientific information from incomplete data. Data-driven science is rapidly\ngrowing, especially in X-ray methodologies, where advanced light sources and\ndetection technologies accumulate vast amounts of data that exceed meticulous\nhuman inspection capabilities. Despite the increasing demands, the full\napplication of machine learning has been hindered by the need for data-specific\noptimizations. In this study, we introduce a new deep-learning-based phase\nretrieval method for imperfect diffraction data. This method provides robust\nphase retrieval for simulated data and performs well on weak-signal\nsingle-pulse diffraction data from X-ray free-electron lasers. Moreover, the\nmethod significantly reduces data processing time, facilitating real-time image\nreconstructions that are crucial for high-repetition-rate data acquisition.\nThus, this approach offers a reliable solution to the phase problem and is\nexpected to be widely adopted across various research areas.\n", "date": "2024-09-24", "categories": []}
{"id": "2503.11073", "title": "Perceive, Understand and Restore: Real-World Image Super-Resolution with\n  Autoregressive Multimodal Generative Models", "abstract": "  By leveraging the generative priors from pre-trained text-to-image diffusion\nmodels, significant progress has been made in real-world image super-resolution\n(Real-ISR). However, these methods tend to generate inaccurate and unnatural\nreconstructions in complex and/or heavily degraded scenes, primarily due to\ntheir limited perception and understanding capability of the input low-quality\nimage. To address these limitations, we propose, for the first time to our\nknowledge, to adapt the pre-trained autoregressive multimodal model such as\nLumina-mGPT into a robust Real-ISR model, namely PURE, which Perceives and\nUnderstands the input low-quality image, then REstores its high-quality\ncounterpart. Specifically, we implement instruction tuning on Lumina-mGPT to\nperceive the image degradation level and the relationships between previously\ngenerated image tokens and the next token, understand the image content by\ngenerating image semantic descriptions, and consequently restore the image by\ngenerating high-quality image tokens autoregressively with the collected\ninformation. In addition, we reveal that the image token entropy reflects the\nimage structure and present a entropy-based Top-k sampling strategy to optimize\nthe local structure of the image during inference. Experimental results\ndemonstrate that PURE preserves image content while generating realistic\ndetails, especially in complex scenes with multiple objects, showcasing the\npotential of autoregressive multimodal generative models for robust Real-ISR.\nThe model and code will be available at https://github.com/nonwhy/PURE.\n", "date": "2025-03-14", "categories": []}
{"id": "2503.11328", "title": "TransiT: Transient Transformer for Non-line-of-sight Videography", "abstract": "  High quality and high speed videography using Non-Line-of-Sight (NLOS)\nimaging benefit autonomous navigation, collision prevention, and post-disaster\nsearch and rescue tasks. Current solutions have to balance between the frame\nrate and image quality. High frame rates, for example, can be achieved by\nreducing either per-point scanning time or scanning density, but at the cost of\nlowering the information density at individual frames. Fast scanning process\nfurther reduces the signal-to-noise ratio and different scanning systems\nexhibit different distortion characteristics. In this work, we design and\nemploy a new Transient Transformer architecture called TransiT to achieve\nreal-time NLOS recovery under fast scans. TransiT directly compresses the\ntemporal dimension of input transients to extract features, reducing\ncomputation costs and meeting high frame rate requirements. It further adopts a\nfeature fusion mechanism as well as employs a spatial-temporal Transformer to\nhelp capture features of NLOS transient videos. Moreover, TransiT applies\ntransfer learning to bridge the gap between synthetic and real-measured data.\nIn real experiments, TransiT manages to reconstruct from sparse transients of\n$16 \\times 16$ measured at an exposure time of 0.4 ms per point to NLOS videos\nat a $64 \\times 64$ resolution at 10 frames per second. We will make our code\nand dataset available to the community.\n", "date": "2025-03-14", "categories": []}
{"id": "2503.11538", "title": "FLASH{\\mu}: Fast Localizing And Sizing of Holographic Microparticles", "abstract": "  Reconstructing the 3D location and size of microparticles from diffraction\nimages - holograms - is a computationally expensive inverse problem that has\ntraditionally been solved using physics-based reconstruction methods. More\nrecently, researchers have used machine learning methods to speed up the\nprocess. However, for small particles in large sample volumes the performance\nof these methods falls short of standard physics-based reconstruction methods.\nHere we designed a two-stage neural network architecture, FLASH$\\mu$, to detect\nsmall particles (6-100$\\mu$m) from holograms with large sample depths up to\n20cm. Trained only on synthetic data with added physical noise, our method\nreliably detects particles of at least 9$\\mu$m diameter in real holograms,\ncomparable to the standard reconstruction-based approaches while operating on\nsmaller crops, at quarter of the original resolution and providing roughly a\n600-fold speedup. In addition to introducing a novel approach to a non-local\nobject detection or signal demixing problem, our work could enable low-cost,\nreal-time holographic imaging setups.\n", "date": "2025-03-14", "categories": []}
{"id": "2503.11640", "title": "Enhancing Deep Learning Based Structured Illumination Microscopy\n  Reconstruction with Light Field Awareness", "abstract": "  Structured illumination microscopy (SIM) is a pivotal technique for dynamic\nsubcellular imaging in live cells. Conventional SIM reconstruction algorithms\ndepend on accurately estimating the illumination pattern and can introduce\nartefacts when this estimation is imprecise. Although recent deep\nlearning-based SIM reconstruction methods have improved speed, accuracy, and\nrobustness, they often struggle with out-of-distribution data. To address this\nlimitation, we propose an Awareness-of-Light-field SIM (AL-SIM) reconstruction\napproach that directly estimates the actual light field to correct for errors\narising from data distribution shifts. Through comprehensive experiments on\nboth simulated filament structures and live BSC1 cells, our method demonstrates\na 7% reduction in the normalized root mean square error (NRMSE) and\nsubstantially lowers reconstruction artefacts. By minimizing these artefacts\nand improving overall accuracy, AL-SIM broadens the applicability of SIM for\ncomplex biological systems.\n", "date": "2025-03-14", "categories": []}
{"id": "2503.10015", "title": "RSR-NF: Neural Field Regularization by Static Restoration Priors for\n  Dynamic Imaging", "abstract": "  Dynamic imaging involves the reconstruction of a spatio-temporal object at\nall times using its undersampled measurements. In particular, in dynamic\ncomputed tomography (dCT), only a single projection at one view angle is\navailable at a time, making the inverse problem very challenging. Moreover,\nground-truth dynamic data is usually either unavailable or too scarce to be\nused for supervised learning techniques. To tackle this problem, we propose\nRSR-NF, which uses a neural field (NF) to represent the dynamic object and,\nusing the Regularization-by-Denoising (RED) framework, incorporates an\nadditional static deep spatial prior into a variational formulation via a\nlearned restoration operator. We use an ADMM-based algorithm with variable\nsplitting to efficiently optimize the variational objective. We compare RSR-NF\nto three alternatives: NF with only temporal regularization; a recent method\ncombining a partially-separable low-rank representation with RED using a\ndenoiser pretrained on static data; and a deep-image prior-based model. The\nfirst comparison demonstrates the reconstruction improvements achieved by\ncombining the NF representation with static restoration priors, whereas the\nother two demonstrate the improvement over state-of-the art techniques for dCT.\n", "date": "2025-03-13", "categories": []}
{"id": "2412.02386", "title": "Single-Shot Metric Depth from Focused Plenoptic Cameras", "abstract": "  Metric depth estimation from visual sensors is crucial for robots to\nperceive, navigate, and interact with their environment. Traditional range\nimaging setups, such as stereo or structured light cameras, face hassles\nincluding calibration, occlusions, and hardware demands, with accuracy limited\nby the baseline between cameras. Single- and multi-view monocular depth offers\na more compact alternative, but is constrained by the unobservability of the\nmetric scale. Light field imaging provides a promising solution for estimating\nmetric depth by using a unique lens configuration through a single device.\nHowever, its application to single-view dense metric depth is under-addressed\nmainly due to the technology's high cost, the lack of public benchmarks, and\nproprietary geometrical models and software. Our work explores the potential of\nfocused plenoptic cameras for dense metric depth. We propose a novel pipeline\nthat predicts metric depth from a single plenoptic camera shot by first\ngenerating a sparse metric point cloud using machine learning, which is then\nused to scale and align a dense relative depth map regressed by a foundation\ndepth model, resulting in dense metric depth. To validate it, we curated the\nLight Field & Stereo Image Dataset (LFS) of real-world light field images with\nstereo depth labels, filling a current gap in existing resources. Experimental\nresults show that our pipeline produces accurate metric depth predictions,\nlaying a solid groundwork for future research in this field.\n", "date": "2024-12-03", "categories": []}
{"id": "2503.08710", "title": "Large model enhanced computational ghost imaging", "abstract": "  Ghost imaging (GI) achieves 2D image reconstruction through high-order\ncorrelation of 1D bucket signals and 2D light field information, particularly\ndemonstrating enhanced detection sensitivity and high-quality image\nreconstruction via efficient photon collection in scattering media. Recent\ninvestigations have established that deep learning (DL) can substantially\nenhance the ghost imaging reconstruction quality. Furthermore, with the\nemergence of large models like SDXL, GPT-4, etc., the constraints of\nconventional DL in parameters and architecture have been transcended, enabling\nmodels to comprehensively explore relationships among all distinct positions\nwithin feature sequences. This paradigm shift has significantly advanced the\ncapability of DL in restoring severely degraded and low-resolution imagery,\nmaking it particularly advantageous for noise-robust image reconstruction in GI\napplications. In this paper, we propose the first large imaging model with 1.4\nbillion parameters that incorporates the physical principles of GI (GILM). The\nproposed GILM implements a skip connection mechanism to mitigate gradient\nexplosion challenges inherent in deep architectures, ensuring sufficient\nparametric capacity to capture intricate correlations among object single-pixel\nmeasurements. Moreover, GILM leverages multi-head attention mechanism to learn\nspatial dependencies across pixel points during image reconstruction,\nfacilitating the extraction of comprehensive object information for subsequent\nreconstruction. We validated the effectiveness of GILM through a series of\nexperiments, including simulated object imaging, imaging objects in free space,\nand imaging object located 52 meters away in underwater environment. The\nexperimental results show that GILM effectively analyzes the fluctuation trends\nof the collected signals, thereby optimizing the recovery of the object's image\nfrom the acquired data.\n", "date": "2025-03-10", "categories": []}
{"id": "2311.14282", "title": "Image Super-Resolution with Text Prompt Diffusion", "abstract": "  Image super-resolution (SR) methods typically model degradation to improve\nreconstruction accuracy in complex and unknown degradation scenarios. However,\nextracting degradation information from low-resolution images is challenging,\nwhich limits the model performance. To boost image SR performance, one feasible\napproach is to introduce additional priors. Inspired by advancements in\nmulti-modal methods and text prompt image processing, we introduce text prompts\nto image SR to provide degradation priors. Specifically, we first design a\ntext-image generation pipeline to integrate text into the SR dataset through\nthe text degradation representation and degradation model. By adopting a\ndiscrete design, the text representation is flexible and user-friendly.\nMeanwhile, we propose the PromptSR to realize the text prompt SR. The PromptSR\nleverages the latest multi-modal large language model (MLLM) to generate\nprompts from low-resolution images. It also utilizes the pre-trained language\nmodel (e.g., T5 or CLIP) to enhance text comprehension. We train the PromptSR\non the text-image dataset. Extensive experiments indicate that introducing text\nprompts into SR, yields impressive results on both synthetic and real-world\nimages. Code: https://github.com/zhengchen1999/PromptSR.\n", "date": "2023-11-24", "categories": []}
{"id": "2407.02906", "title": "Single Image Rolling Shutter Removal with Diffusion Models", "abstract": "  We present RS-Diffusion, the first Diffusion Models-based method for\nsingle-frame Rolling Shutter (RS) correction. RS artifacts compromise visual\nquality of frames due to the row-wise exposure of CMOS sensors. Most previous\nmethods have focused on multi-frame approaches, using temporal information from\nconsecutive frames for the motion rectification. However, few approaches\naddress the more challenging but important single frame RS correction. In this\nwork, we present an ``image-to-motion\" framework via diffusion techniques, with\na designed patch-attention module. In addition, we present the RS-Real dataset,\ncomprised of captured RS frames alongside their corresponding Global Shutter\n(GS) ground-truth pairs. The GS frames are corrected from the RS ones, guided\nby the corresponding Inertial Measurement Unit (IMU) gyroscope data acquired\nduring capture. Experiments show that RS-Diffusion surpasses previous\nsingle-frame RS methods, demonstrates the potential of diffusion-based\napproaches, and provides a valuable dataset for further research.\n", "date": "2024-07-03", "categories": []}
{"id": "2411.15269", "title": "MambaIRv2: Attentive State Space Restoration", "abstract": "  The Mamba-based image restoration backbones have recently demonstrated\nsignificant potential in balancing global reception and computational\nefficiency. However, the inherent causal modeling limitation of Mamba, where\neach token depends solely on its predecessors in the scanned sequence,\nrestricts the full utilization of pixels across the image and thus presents new\nchallenges in image restoration. In this work, we propose MambaIRv2, which\nequips Mamba with the non-causal modeling ability similar to ViTs to reach the\nattentive state space restoration model. Specifically, the proposed attentive\nstate-space equation allows to attend beyond the scanned sequence and\nfacilitate image unfolding with just one single scan. Moreover, we further\nintroduce a semantic-guided neighboring mechanism to encourage interaction\nbetween distant but similar pixels. Extensive experiments show our MambaIRv2\noutperforms SRFormer by even 0.35dB PSNR for lightweight SR even with 9.3\\%\nless parameters and suppresses HAT on classic SR by up to 0.29dB. Code is\navailable at https://github.com/csguoh/MambaIR.\n", "date": "2024-11-22", "categories": []}
{"id": "2502.11382", "title": "A Physics-Informed Blur Learning Framework for Imaging Systems", "abstract": "  Accurate blur estimation is essential for high-performance imaging across\nvarious applications. Blur is typically represented by the point spread\nfunction (PSF). In this paper, we propose a physics-informed PSF learning\nframework for imaging systems, consisting of a simple calibration followed by a\nlearning process. Our framework could achieve both high accuracy and universal\napplicability. Inspired by the Seidel PSF model for representing spatially\nvarying PSF, we identify its limitations in optimization and introduce a novel\nwavefront-based PSF model accompanied by an optimization strategy, both\nreducing optimization complexity and improving estimation accuracy. Moreover,\nour wavefront-based PSF model is independent of lens parameters, eliminate the\nneed for prior knowledge of the lens. To validate our approach, we compare it\nwith recent PSF estimation methods (Degradation Transfer and Fast Two-step)\nthrough a deblurring task, where all the estimated PSFs are used to train\nstate-of-the-art deblurring algorithms. Our approach demonstrates improvements\nin image quality in simulation and also showcases noticeable visual quality\nimprovements on real captured images.\n", "date": "2025-02-17", "categories": []}
{"id": "2503.07748", "title": "AdaptSR: Low-Rank Adaptation for Efficient and Scalable Real-World\n  Super-Resolution", "abstract": "  Recovering high-frequency details and textures from low-resolution images\nremains a fundamental challenge in super-resolution (SR), especially when\nreal-world degradations are complex and unknown. While GAN-based methods\nenhance realism, they suffer from training instability and introduce unnatural\nartifacts. Diffusion models, though promising, demand excessive computational\nresources, often requiring multiple GPU days, even for single-step variants.\nRather than naively fine-tuning entire models or adopting unstable generative\napproaches, we introduce AdaptSR, a low-rank adaptation (LoRA) framework that\nefficiently repurposes bicubic-trained SR models for real-world tasks. AdaptSR\nleverages architecture-specific insights and selective layer updates to\noptimize real SR adaptation. By updating only lightweight LoRA layers while\nkeeping the pretrained backbone intact, it captures domain-specific adjustments\nwithout adding inference cost, as the adapted layers merge seamlessly\npost-training. This efficient adaptation not only reduces memory and compute\nrequirements but also makes real-world SR feasible on lightweight hardware. Our\nexperiments demonstrate that AdaptSR outperforms GAN and diffusion-based SR\nmethods by up to 4 dB in PSNR and 2% in perceptual scores on real SR\nbenchmarks. More impressively, it matches or exceeds full model fine-tuning\nwhile training 92% fewer parameters, enabling rapid adaptation to real SR tasks\nwithin minutes.\n", "date": "2025-03-10", "categories": []}
{"id": "2503.07814", "title": "Whiteness-based bilevel estimation of weighted TV parameter maps for\n  image denoising", "abstract": "  We consider a bilevel optimisation strategy based on normalised residual\nwhiteness loss for estimating the weighted total variation parameter maps for\ndenoising images corrupted by additive white Gaussian noise. Compared to\nsupervised and semi-supervised approaches relying on prior knowledge of\n(approximate) reference data and/or information on the noise magnitude, the\nproposal is fully unsupervised. To avoid noise overfitting an early stopping\nstrategy is used, relying on simple statistics of optimal performances on a set\nof natural images. Numerical results comparing the supervised/unsupervised\nprocedures for scalar/pixel-dependent \\mbox{parameter maps are shown.\n", "date": "2025-03-10", "categories": []}
{"id": "2503.08136", "title": "FlowDPS: Flow-Driven Posterior Sampling for Inverse Problems", "abstract": "  Flow matching is a recent state-of-the-art framework for generative modeling\nbased on ordinary differential equations (ODEs). While closely related to\ndiffusion models, it provides a more general perspective on generative\nmodeling. Although inverse problem solving has been extensively explored using\ndiffusion models, it has not been rigorously examined within the broader\ncontext of flow models. Therefore, here we extend the diffusion inverse solvers\n(DIS) - which perform posterior sampling by combining a denoising diffusion\nprior with an likelihood gradient - into the flow framework. Specifically, by\ndriving the flow-version of Tweedie's formula, we decompose the flow ODE into\ntwo components: one for clean image estimation and the other for noise\nestimation. By integrating the likelihood gradient and stochastic noise into\neach component, respectively, we demonstrate that posterior sampling for\ninverse problem solving can be effectively achieved using flows. Our proposed\nsolver, Flow-Driven Posterior Sampling (FlowDPS), can also be seamlessly\nintegrated into a latent flow model with a transformer architecture. Across\nfour linear inverse problems, we confirm that FlowDPS outperforms\nstate-of-the-art alternatives, all without requiring additional training.\n", "date": "2025-03-11", "categories": []}
{"id": "2503.08429", "title": "Using Powerful Prior Knowledge of Diffusion Model in Deep Unfolding\n  Networks for Image Compressive Sensing", "abstract": "  Recently, Deep Unfolding Networks (DUNs) have achieved impressive\nreconstruction quality in the field of image Compressive Sensing (CS) by\nunfolding iterative optimization algorithms into neural networks. The\nreconstruction quality of DUNs depends on the learned prior knowledge, so\nintroducing stronger prior knowledge can further improve reconstruction\nquality. On the other hand, pre-trained diffusion models contain powerful prior\nknowledge and have a solid theoretical foundation and strong scalability, but\nit requires a large number of iterative steps to achieve reconstruction. In\nthis paper, we propose to use the powerful prior knowledge of pre-trained\ndiffusion model in DUNs to achieve high-quality reconstruction with less steps\nfor image CS. Specifically, we first design an iterative optimization algorithm\nnamed Diffusion Message Passing (DMP), which embeds a pre-trained diffusion\nmodel into each iteration process of DMP. Then, we deeply unfold the DMP\nalgorithm into a neural network named DMP-DUN. The proposed DMP-DUN can use\nlightweight neural networks to achieve mapping from measurement data to the\nintermediate steps of the reverse diffusion process and directly approximate\nthe divergence of the diffusion model, thereby further improving reconstruction\nefficiency. Extensive experiments show that our proposed DMP-DUN achieves\nstate-of-the-art performance and requires at least only 2 steps to reconstruct\nthe image. Codes are available at\nhttps://github.com/FengodChen/DMP-DUN-CVPR2025.\n", "date": "2025-03-11", "categories": []}
{"id": "2404.03906", "title": "Deep Phase Coded Image Prior", "abstract": "  Phase-coded imaging is a computational imaging method designed to tackle\ntasks such as passive depth estimation and extended depth of field (EDOF) using\ndepth cues inserted during image capture. Most of the current deep\nlearning-based methods for depth estimation or all-in-focus imaging require a\ntraining dataset with high-quality depth maps and an optimal focus point at\ninfinity for all-in-focus images. Such datasets are difficult to create,\nusually synthetic, and require external graphic programs. We propose a new\nmethod named \"Deep Phase Coded Image Prior\" (DPCIP) for jointly recovering the\ndepth map and all-in-focus image from a coded-phase image using solely the\ncaptured image and the optical information of the imaging system. Our approach\ndoes not depend on any specific dataset and surpasses prior supervised\ntechniques utilizing the same imaging system. This improvement is achieved\nthrough the utilization of a problem formulation based on implicit neural\nrepresentation (INR) and deep image prior (DIP). Due to our zero-shot method,\nwe overcome the barrier of acquiring accurate ground-truth data of depth maps\nand all-in-focus images for each new phase-coded system introduced. This allows\nfocusing mainly on developing the imaging system, and not on ground-truth data\ncollection.\n", "date": "2024-04-05", "categories": []}
{"id": "2411.13383", "title": "Adversarial Diffusion Compression for Real-World Image Super-Resolution", "abstract": "  Real-world image super-resolution (Real-ISR) aims to reconstruct\nhigh-resolution images from low-resolution inputs degraded by complex, unknown\nprocesses. While many Stable Diffusion (SD)-based Real-ISR methods have\nachieved remarkable success, their slow, multi-step inference hinders practical\ndeployment. Recent SD-based one-step networks like OSEDiff and S3Diff alleviate\nthis issue but still incur high computational costs due to their reliance on\nlarge pretrained SD models. This paper proposes a novel Real-ISR method, AdcSR,\nby distilling the one-step diffusion network OSEDiff into a streamlined\ndiffusion-GAN model under our Adversarial Diffusion Compression (ADC)\nframework. We meticulously examine the modules of OSEDiff, categorizing them\ninto two types: (1) Removable (VAE encoder, prompt extractor, text encoder,\netc.) and (2) Prunable (denoising UNet and VAE decoder). Since direct removal\nand pruning can degrade the model's generation capability, we pretrain our\npruned VAE decoder to restore its ability to decode images and employ\nadversarial distillation to compensate for performance loss. This ADC-based\ndiffusion-GAN hybrid design effectively reduces complexity by 73% in inference\ntime, 78% in computation, and 74% in parameters, while preserving the model's\ngeneration capability. Experiments manifest that our proposed AdcSR achieves\ncompetitive recovery quality on both synthetic and real-world datasets,\noffering up to 9.3$\\times$ speedup over previous one-step diffusion-based\nmethods. Code and models are available at\nhttps://github.com/Guaishou74851/AdcSR.\n", "date": "2024-11-20", "categories": []}
{"id": "2502.19068", "title": "Dynamic Degradation Decomposition Network for All-in-One Image\n  Restoration", "abstract": "  Currently, restoring clean images from a variety of degradation types using a\nsingle model is still a challenging task. Existing all-in-one image restoration\napproaches struggle with addressing complex and ambiguously defined degradation\ntypes. In this paper, we introduce a dynamic degradation decomposition network\nfor all-in-one image restoration, named D$^3$Net. D$^3$Net achieves\ndegradation-adaptive image restoration with guided prompt through cross-domain\ninteraction and dynamic degradation decomposition. Concretely, in D$^3$Net, the\nproposed Cross-Domain Degradation Analyzer (CDDA) engages in deep interaction\nbetween frequency domain degradation characteristics and spatial domain image\nfeatures to identify and model variations of different degradation types on the\nimage manifold, generating degradation correction prompt and strategy prompt,\nwhich guide the following decomposition process. Furthermore, the prompt-based\nDynamic Decomposition Mechanism (DDM) for progressive degradation\ndecomposition, that encourages the network to adaptively select restoration\nstrategies utilizing the two-level prompt generated by CDDA. Thanks to the\nsynergistic cooperation between CDDA and DDM, D$^3$Net achieves superior\nflexibility and scalability in handling unknown degradation, while effectively\nreducing unnecessary computational overhead. Extensive experiments on multiple\nimage restoration tasks demonstrate that D$^3$Net significantly outperforms the\nstate-of-the-art approaches, especially improving PSNR by 5.47dB and 3.30dB on\nthe SOTS-Outdoor and GoPro datasets, respectively.\n", "date": "2025-02-26", "categories": []}
{"id": "2503.06107", "title": "Feature Fusion Attention Network with CycleGAN for Image Dehazing,\n  De-Snowing and De-Raining", "abstract": "  This paper presents a novel approach to image dehazing by combining Feature\nFusion Attention (FFA) networks with CycleGAN architecture. Our method\nleverages both supervised and unsupervised learning techniques to effectively\nremove haze from images while preserving crucial image details. The proposed\nhybrid architecture demonstrates significant improvements in image quality\nmetrics, achieving superior PSNR and SSIM scores compared to traditional\ndehazing methods. Through extensive experimentation on the RESIDE and DenseHaze\nCVPR 2019 dataset, we show that our approach effectively handles both synthetic\nand real-world hazy images. CycleGAN handles the unpaired nature of hazy and\nclean images effectively, enabling the model to learn mappings even without\npaired data.\n", "date": "2025-03-08", "categories": []}
{"id": "2503.06617", "title": "Pixel to Gaussian: Ultra-Fast Continuous Super-Resolution with 2D\n  Gaussian Modeling", "abstract": "  Arbitrary-scale super-resolution (ASSR) aims to reconstruct high-resolution\n(HR) images from low-resolution (LR) inputs with arbitrary upsampling factors\nusing a single model, addressing the limitations of traditional SR methods\nconstrained to fixed-scale factors (\\textit{e.g.}, $\\times$ 2). Recent advances\nleveraging implicit neural representation (INR) have achieved great progress by\nmodeling coordinate-to-pixel mappings. However, the efficiency of these methods\nmay suffer from repeated upsampling and decoding, while their reconstruction\nfidelity and quality are constrained by the intrinsic representational\nlimitations of coordinate-based functions. To address these challenges, we\npropose a novel ContinuousSR framework with a Pixel-to-Gaussian paradigm, which\nexplicitly reconstructs 2D continuous HR signals from LR images using Gaussian\nSplatting. This approach eliminates the need for time-consuming upsampling and\ndecoding, enabling extremely fast arbitrary-scale super-resolution. Once the\nGaussian field is built in a single pass, ContinuousSR can perform\narbitrary-scale rendering in just 1ms per scale. Our method introduces several\nkey innovations. Through statistical ana\n", "date": "2025-03-09", "categories": []}
{"id": "2503.07004", "title": "NukesFormers: Unpaired Hyperspectral Image Generation with Non-Uniform\n  Domain Alignment", "abstract": "  The inherent difficulty in acquiring accurately co-registered\nRGB-hyperspectral image (HSI) pairs has significantly impeded the practical\ndeployment of current data-driven Hyperspectral Image Generation (HIG) networks\nin engineering applications. Gleichzeitig, the ill-posed nature of the aligning\nconstraints, compounded with the complexities of mining cross-domain features,\nalso hinders the advancement of unpaired HIG (UnHIG) tasks. In this paper, we\nconquer these challenges by modeling the UnHIG to range space interaction and\ncompensations of null space through Range-Null Space Decomposition (RND)\nmethodology. Specifically, the introduced contrastive learning effectively\naligns the geometric and spectral distributions of unpaired data by building\nthe interaction of range space, considering the consistent feature in\ndegradation process. Following this, we map the frequency representations of\ndual-domain input and thoroughly mining the null space, like degraded and\nhigh-frequency components, through the proposed Non-uniform Kolmogorov-Arnold\nNetworks. Extensive comparative experiments demonstrate that it establishes a\nnew benchmark in UnHIG.\n", "date": "2025-03-10", "categories": []}
{"id": "2502.16532", "title": "Deep unrolling for learning optimal spatially varying regularisation\n  parameters for Total Generalised Variation", "abstract": "  We extend a recently introduced deep unrolling framework for learning\nspatially varying regularisation parameters in inverse imaging problems to the\ncase of Total Generalised Variation (TGV). The framework combines a deep\nconvolutional neural network (CNN) inferring the two spatially varying TGV\nparameters with an unrolled algorithmic scheme that solves the corresponding\nvariational problem. The two subnetworks are jointly trained end-to-end in a\nsupervised fashion and as such the CNN learns to compute those parameters that\ndrive the reconstructed images as close to the ground truth as possible.\nNumerical results in image denoising and MRI reconstruction show a significant\nqualitative and quantitative improvement compared to the best TGV scalar\nparameter case as well as to other approaches employing spatially varying\nparameters computed by unsupervised methods. We also observe that the inferred\nspatially varying parameter maps have a consistent structure near the image\nedges, asking for further theoretical investigations. In particular, the\nparameter that weighs the first-order TGV term has a triple-edge structure with\nalternating high-low-high values whereas the one that weighs the second-order\nterm attains small values in a large neighbourhood around the edges.\n", "date": "2025-02-23", "categories": []}
{"id": "2210.09604", "title": "Perceptual Multi-Exposure Fusion", "abstract": "  As an ever-increasing demand for high dynamic range (HDR) scene shooting,\nmulti-exposure image fusion (MEF) technology has abounded. In recent years,\nmulti-scale exposure fusion approaches based on detail-enhancement have led the\nway for improvement in highlight and shadow details. Most of such methods,\nhowever, are too computationally expensive to be deployed on mobile devices.\nThis paper presents a perceptual multi-exposure fusion method that not just\nensures fine shadow/highlight details but with lower complexity than\ndetailenhanced methods. We analyze the potential defects of three classical\nexposure measures in lieu of using detail-enhancement component and improve two\nof them, namely adaptive Wellexposedness (AWE) and the gradient of color images\n(3-D gradient). AWE designed in YCbCr color space considers the difference\nbetween varying exposure images. 3-D gradient is employed to extract fine\ndetails. We build a large-scale multiexposure benchmark dataset suitable for\nstatic scenes, which contains 167 image sequences all told. Experiments on the\nconstructed dataset demonstrate that the proposed method exceeds existing eight\nstate-of-the-art approaches in terms of visually and MEF-SSIM value. Moreover,\nour approach can achieve a better improvement for current image enhancement\ntechniques, ensuring fine detail in bright light.\n", "date": "2022-10-18", "categories": []}
{"id": "2503.02908", "title": "Hyperspectral Image Restoration and Super-resolution with Physics-Aware\n  Deep Learning for Biomedical Applications", "abstract": "  Hyperspectral imaging is a powerful bioimaging tool which can uncover novel\ninsights, thanks to its sensitivity to the intrinsic properties of materials.\nHowever, this enhanced contrast comes at the cost of system complexity,\nconstrained by an inherent trade-off between spatial resolution, spectral\nresolution, and imaging speed. To overcome this limitation, we present a deep\nlearning-based approach that restores and enhances pixel resolution\npost-acquisition without any a priori knowledge. Fine-tuned using metrics\naligned with the imaging model, our physics-aware method achieves a 16X pixel\nsuper-resolution enhancement and a 12X imaging speedup without the need of\nadditional training data for transfer learning. Applied to both synthetic and\nexperimental data from five different sample types, we demonstrate that the\nmodel preserves biological integrity, ensuring no features are lost or\nhallucinated. We also concretely demonstrate the model's ability to reveal\ndisease-associated metabolic changes in Downs syndrome that would otherwise\nremain undetectable. Furthermore, we provide physical insights into the inner\nworkings of the model, paving the way for future refinements that could\npotentially surpass instrumental limits in an explainable manner. All methods\nare available as open-source software on GitHub.\n", "date": "2025-03-03", "categories": []}
{"id": "2503.03115", "title": "NTR-Gaussian: Nighttime Dynamic Thermal Reconstruction with 4D Gaussian\n  Splatting Based on Thermodynamics", "abstract": "  Thermal infrared imaging offers the advantage of all-weather capability,\nenabling non-intrusive measurement of an object's surface temperature.\nConsequently, thermal infrared images are employed to reconstruct 3D models\nthat accurately reflect the temperature distribution of a scene, aiding in\napplications such as building monitoring and energy management. However,\nexisting approaches predominantly focus on static 3D reconstruction for a\nsingle time period, overlooking the impact of environmental factors on thermal\nradiation and failing to predict or analyze temperature variations over time.\nTo address these challenges, we propose the NTR-Gaussian method, which treats\ntemperature as a form of thermal radiation, incorporating elements like\nconvective heat transfer and radiative heat dissipation. Our approach utilizes\nneural networks to predict thermodynamic parameters such as emissivity,\nconvective heat transfer coefficient, and heat capacity. By integrating these\npredictions, we can accurately forecast thermal temperatures at various times\nthroughout a nighttime scene. Furthermore, we introduce a dynamic dataset\nspecifically for nighttime thermal imagery. Extensive experiments and\nevaluations demonstrate that NTR-Gaussian significantly outperforms comparison\nmethods in thermal reconstruction, achieving a predicted temperature error\nwithin 1 degree Celsius.\n", "date": "2025-03-05", "categories": []}
{"id": "2407.01574", "title": "cryoSPHERE: Single-particle heterogeneous reconstruction from cryo EM", "abstract": "  The three-dimensional structure of proteins plays a crucial role in\ndetermining their function. Protein structure prediction methods, like\nAlphaFold, offer rapid access to a protein structure. However, large protein\ncomplexes cannot be reliably predicted, and proteins are dynamic, making it\nimportant to resolve their full conformational distribution. Single-particle\ncryo-electron microscopy (cryo-EM) is a powerful tool for determining the\nstructures of large protein complexes. Importantly, the numerous images of a\ngiven protein contain underutilized information about conformational\nheterogeneity. These images are very noisy projections of the protein, and\ntraditional methods for cryo-EM reconstruction are limited to recovering only\none or a few consensus conformations. In this paper, we introduce cryoSPHERE,\nwhich is a deep learning method that uses a nominal protein structure (e.g.,\nfrom AlphaFold) as input, learns how to divide it into segments, and moves\nthese segments as approximately rigid bodies to fit the different conformations\npresent in the cryo-EM dataset. This approach provides enough constraints to\nenable meaningful reconstructions of single protein structural ensembles. We\ndemonstrate this with two synthetic datasets featuring varying levels of noise,\nas well as two real dataset. We show that cryoSPHERE is very resilient to the\nhigh levels of noise typically encountered in experiments, where we see\nconsistent improvements over the current state-of-the-art for heterogeneous\nreconstruction.\n", "date": "2024-05-29", "categories": []}
{"id": "2501.04477", "title": "Rethinking High-speed Image Reconstruction Framework with Spike Camera", "abstract": "  Spike cameras, as innovative neuromorphic devices, generate continuous spike\nstreams to capture high-speed scenes with lower bandwidth and higher dynamic\nrange than traditional RGB cameras. However, reconstructing high-quality images\nfrom the spike input under low-light conditions remains challenging.\nConventional learning-based methods often rely on the synthetic dataset as the\nsupervision for training. Still, these approaches falter when dealing with\nnoisy spikes fired under the low-light environment, leading to further\nperformance degradation in the real-world dataset. This phenomenon is primarily\ndue to inadequate noise modelling and the domain gap between synthetic and real\ndatasets, resulting in recovered images with unclear textures, excessive noise,\nand diminished brightness. To address these challenges, we introduce a novel\nspike-to-image reconstruction framework SpikeCLIP that goes beyond traditional\ntraining paradigms. Leveraging the CLIP model's powerful capability to align\ntext and images, we incorporate the textual description of the captured scene\nand unpaired high-quality datasets as the supervision. Our experiments on\nreal-world low-light datasets U-CALTECH and U-CIFAR demonstrate that SpikeCLIP\nsignificantly enhances texture details and the luminance balance of recovered\nimages. Furthermore, the reconstructed images are well-aligned with the broader\nvisual features needed for downstream tasks, ensuring more robust and versatile\nperformance in challenging environments.\n", "date": "2025-01-08", "categories": []}
{"id": "2502.19677", "title": "Towards Differential Handling of Various Blur Regions for Accurate Image\n  Deblurring", "abstract": "  Image deblurring aims to restore high-quality images by removing undesired\ndegradation. Although existing methods have yielded promising results, they\neither overlook the varying degrees of degradation across different regions of\nthe blurred image, or they approximate nonlinear function properties by\nstacking numerous nonlinear activation functions. In this paper, we propose a\ndifferential handling network (DHNet) to perform differential processing for\ndifferent blur regions. Specifically, we design a Volterra block (VBlock) to\nintegrate the nonlinear characteristics into the deblurring network, avoiding\nthe previous operation of stacking the number of nonlinear activation functions\nto map complex input-output relationships. To enable the model to adaptively\naddress varying degradation degrees in blurred regions, we devise the\ndegradation degree recognition expert module (DDRE). This module initially\nincorporates prior knowledge from a well-trained model to estimate spatially\nvariable blur information. Consequently, the router can map the learned\ndegradation representation and allocate weights to experts according to both\nthe degree of degradation and the size of the regions. Comprehensive\nexperimental results show that DHNet effectively surpasses state-of-the-art\n(SOTA) methods on both synthetic and real-world datasets.\n", "date": "2025-02-27", "categories": []}
{"id": "2503.01462", "title": "S-R2D2: a spherical extension of the R2D2 deep neural network series\n  paradigm for wide-field radio-interferometric imaging", "abstract": "  Recently, the R2D2 paradigm, standing for ''Residual-to-Residual DNN series\nfor high-Dynamic-range imaging'', was introduced for image formation in Radio\nInterferometry (RI) as a learned version of the traditional algorithm CLEAN.\nThe first incarnations of R2D2 are limited to planar imaging on small fields of\nview, failing to meet the spherical-imaging requirement of modern telescopes\nobserving wide fields. To address this limitation, we propose the\nspherical-imaging extension S-R2D2. Firstly, as R2D2, S-R2D2 encapsulates its\nminor cycles in existing 2D-Euclidean deep neural network (DNN) architectures,\nbut adapts its iterative scheme to incorporate the wide-field measurement model\nmapping a spherical image to visibility data. We implemented this model as the\ncomposition of an efficient Fourier-based interpolator mapping the spherical\nimage onto the equatorial plane, with the standard RI operator mapping the\nequatorial-plane image to visibility data. Importantly, the interpolation step\nmust inevitably be performed at a lower-than-optimal resolution on the plane,\nto meet the high-resolution requirement on the sphere of wide-field imaging\nwhile preserving scalability. Therefore, secondly, we design S-R2D2's DNN\ntraining loss to jointly learn to correct the interpolation approximations and\nidentify residual image structures on the sphere, ensuring consistency with the\nspherical ground truth using the adjoint plane-to-sphere interpolator. Finally,\nwe demonstrate through simulations S-R2D2's capability to perform fast and\naccurate reconstructions of spherical monochromatic intensity images, across\nhigh-resolution, high-dynamic-range settings.\n", "date": "2025-03-03", "categories": []}
{"id": "2503.01938", "title": "A Lightweight Deep Exclusion Unfolding Network for Single Image\n  Reflection Removal", "abstract": "  Single Image Reflection Removal (SIRR) is a canonical blind source separation\nproblem and refers to the issue of separating a reflection-contaminated image\ninto a transmission and a reflection image. The core challenge lies in\nminimizing the commonalities among different sources. Existing deep learning\napproaches either neglect the significance of feature interactions or rely on\nheuristically designed architectures. In this paper, we propose a novel Deep\nExclusion unfolding Network (DExNet), a lightweight, interpretable, and\neffective network architecture for SIRR. DExNet is principally constructed by\nunfolding and parameterizing a simple iterative Sparse and Auxiliary Feature\nUpdate (i-SAFU) algorithm, which is specifically designed to solve a new\nmodel-based SIRR optimization formulation incorporating a general exclusion\nprior. This general exclusion prior enables the unfolded SAFU module to\ninherently identify and penalize commonalities between the transmission and\nreflection features, ensuring more accurate separation. The principled design\nof DExNet not only enhances its interpretability but also significantly\nimproves its performance. Comprehensive experiments on four benchmark datasets\ndemonstrate that DExNet achieves state-of-the-art visual and quantitative\nresults while utilizing only approximately 8\\% of the parameters required by\nleading methods.\n", "date": "2025-03-03", "categories": []}
{"id": "2503.02242", "title": "$\\mathbf{\\Phi}$-GAN: Physics-Inspired GAN for Generating SAR Images\n  Under Limited Data", "abstract": "  Approaches for improving generative adversarial networks (GANs) training\nunder a few samples have been explored for natural images. However, these\nmethods have limited effectiveness for synthetic aperture radar (SAR) images,\nas they do not account for the unique electromagnetic scattering properties of\nSAR. To remedy this, we propose a physics-inspired regularization method dubbed\n$\\Phi$-GAN, which incorporates the ideal point scattering center (PSC) model of\nSAR with two physical consistency losses. The PSC model approximates SAR\ntargets using physical parameters, ensuring that $\\Phi$-GAN generates SAR\nimages consistent with real physical properties while preventing discriminator\noverfitting by focusing on PSC-based decision cues. To embed the PSC model into\nGANs for end-to-end training, we introduce a physics-inspired neural module\ncapable of estimating the physical parameters of SAR targets efficiently. This\nmodule retains the interpretability of the physical model and can be trained\nwith limited data. We propose two physical loss functions: one for the\ngenerator, guiding it to produce SAR images with physical parameters consistent\nwith real ones, and one for the discriminator, enhancing its robustness by\nbasing decisions on PSC attributes. We evaluate $\\Phi$-GAN across several\nconditional GAN (cGAN) models, demonstrating state-of-the-art performance in\ndata-scarce scenarios on three SAR image datasets.\n", "date": "2025-03-04", "categories": []}
{"id": "2503.02487", "title": "Scene-based nonuniformity correction with homography transformation", "abstract": "  Due to their affordable, low mass, and small dimensions, uncooled\nmicrobolometer-based thermal focal plane arrays (UC-FPAs) are useful for\nlong-wave infrared (LWIR)imaging applications. However, in outdoor conditions\ntypical in agricultural remote sensing, cameras based on UC-FPAs may suffer\nfrom drift in offset and gain. To tackle the persistent drift, the system\nrequires continuous calibration. Our goal in this study was to eliminate this\nrequirement via a computational schema. In a former study, we estimated unknown\ngain and offset values and thermographic images of an object from a sequence of\npairs of successive images taken at two different blur levels.In the current\nwork, we took on a similar problem using a sequence of shifted images, with\nrelative shifts caused by realistic drone hovering modeled by homography\ntransformation. This places our work in the realm of scene-based nonuniformity\ncorrection problems. We show that an object's thermographic values, as well as\ngain and offset, can be jointly estimated by relying on a few sets of shifted\nimages. We use a minimum likelihood estimator, which is found using alternating\nminimization. Registration is done using a generalized Lucas-Kanade method.\nSimulations show promising accuracy with mean Pearson correlation of more than\n0.9999998 between ground truth and restoration. Under ideal assumptions, this\nis equivalent to a mean restoration error of less than 0.01 Celsius degree.\n", "date": "2025-03-04", "categories": []}
{"id": "2503.00762", "title": "MR-EIT: Multi-Resolution Reconstruction for Electrical Impedance\n  Tomography via Data-Driven and Unsupervised Dual-Mode Neural Networks", "abstract": "  This paper presents a multi-resolution reconstruction method for Electrical\nImpedance Tomography (EIT), referred to as MR-EIT, which is capable of\noperating in both supervised and unsupervised learning modes. MR-EIT integrates\nan ordered feature extraction module and an unordered coordinate feature\nexpression module. The former achieves the mapping from voltage to\ntwo-dimensional conductivity features through pre-training, while the latter\nrealizes multi-resolution reconstruction independent of the order and size of\nthe input sequence by utilizing symmetric functions and local feature\nextraction mechanisms. In the data-driven mode, MR-EIT reconstructs\nhigh-resolution images from low-resolution data of finite element meshes\nthrough two stages of pre-training and joint training, and demonstrates\nexcellent performance in simulation experiments. In the unsupervised learning\nmode, MR-EIT does not require pre-training data and performs iterative\noptimization solely based on measured voltages to rapidly achieve image\nreconstruction from low to high resolution. It shows robustness to noise and\nefficient super-resolution reconstruction capabilities in both simulation and\nreal water tank experiments. Experimental results indicate that MR-EIT\noutperforms the comparison methods in terms of Structural Similarity (SSIM) and\nRelative Image Error (RIE), especially in the unsupervised learning mode, where\nit can significantly reduce the number of iterations and improve image\nreconstruction quality.\n", "date": "2025-03-02", "categories": []}
{"id": "2503.00908", "title": "Patient-Level Anatomy Meets Scanning-Level Physics: Personalized\n  Federated Low-Dose CT Denoising Empowered by Large Language Model", "abstract": "  Reducing radiation doses benefits patients, however, the resultant low-dose\ncomputed tomography (LDCT) images often suffer from clinically unacceptable\nnoise and artifacts. While deep learning (DL) shows promise in LDCT\nreconstruction, it requires large-scale data collection from multiple clients,\nraising privacy concerns. Federated learning (FL) has been introduced to\naddress these privacy concerns; however, current methods are typically tailored\nto specific scanning protocols, which limits their generalizability and makes\nthem less effective for unseen protocols. To address these issues, we propose\nSCAN-PhysFed, a novel SCanning- and ANatomy-level personalized Physics-Driven\nFederated learning paradigm for LDCT reconstruction. Since the noise\ndistribution in LDCT data is closely tied to scanning protocols and anatomical\nstructures being scanned, we design a dual-level physics-informed way to\naddress these challenges. Specifically, we incorporate physical and anatomical\nprompts into our physics-informed hypernetworks to capture scanning- and\nanatomy-specific information, enabling dual-level physics-driven\npersonalization of imaging features. These prompts are derived from the\nscanning protocol and the radiology report generated by a medical large\nlanguage model (MLLM), respectively. Subsequently, client-specific decoders\nproject these dual-level personalized imaging features back into the image\ndomain. Besides, to tackle the challenge of unseen data, we introduce a novel\nprotocol vector-quantization strategy (PVQS), which ensures consistent\nperformance across new clients by quantifying the unseen scanning code as one\nof the codes in the scanning codebook. Extensive experimental results\ndemonstrate the superior performance of SCAN-PhysFed on public datasets.\n", "date": "2025-03-02", "categories": []}
{"id": "2503.01136", "title": "Prior-guided Hierarchical Harmonization Network for Efficient Image\n  Dehazing", "abstract": "  Image dehazing is a crucial task that involves the enhancement of degraded\nimages to recover their sharpness and textures. While vision Transformers have\nexhibited impressive results in diverse dehazing tasks, their quadratic\ncomplexity and lack of dehazing priors pose significant drawbacks for\nreal-world applications.\n  In this paper, guided by triple priors, Bright Channel Prior (BCP), Dark\nChannel Prior (DCP), and Histogram Equalization (HE), we propose a\n\\textit{P}rior-\\textit{g}uided Hierarchical \\textit{H}armonization Network\n(PGH$^2$Net) for image dehazing. PGH$^2$Net is built upon the UNet-like\narchitecture with an efficient encoder and decoder, consisting of two module\ntypes: (1) Prior aggregation module that injects B/DCP and selects diverse\ncontexts with gating attention. (2) Feature harmonization modules that subtract\nlow-frequency components from spatial and channel aspects and learn more\ninformative feature distributions to equalize the feature maps.\n", "date": "2025-03-03", "categories": []}
{"id": "2503.01628", "title": "A General Purpose Spectral Foundational Model for Both Proximal and\n  Remote Sensing Spectral Imaging", "abstract": "  Spectral imaging data acquired via multispectral and hyperspectral cameras\ncan have hundreds of channels, where each channel records the reflectance at a\nspecific wavelength and bandwidth. Time and resource constraints limit our\nability to collect large spectral datasets, making it difficult to build and\ntrain predictive models from scratch. In the RGB domain, we can often alleviate\nsome of the limitations of smaller datasets by using pretrained foundational\nmodels as a starting point. However, most existing foundation models are\npretrained on large datasets of 3-channel RGB images, severely limiting their\neffectiveness when used with spectral imaging data. The few spectral foundation\nmodels that do exist usually have one of two limitations: (1) they are built\nand trained only on remote sensing data limiting their application in proximal\nspectral imaging, (2) they utilize the more widely available multispectral\nimaging datasets with less than 15 channels restricting their use with\nhundred-channel hyperspectral images. To alleviate these issues, we propose a\nlarge-scale foundational model and dataset built upon the masked autoencoder\narchitecture that takes advantage of spectral channel encoding,\nspatial-spectral masking and ImageNet pretraining for an adaptable and robust\nmodel for downstream spectral imaging tasks.\n", "date": "2025-03-03", "categories": []}
{"id": "2407.11748", "title": "Ubiquitous Metadata: Design and Fabrication of Embedded Markers for\n  Real-World Object Identification and Interaction", "abstract": "  The convergence of the physical and digital realms has ushered in a new era\nof immersive experiences and seamless interactions. As the boundaries between\nthe real world and virtual environments blur and result in a \"mixed reality,\"\nthere arises a need for robust and efficient methods to connect physical\nobjects with their virtual counterparts. In this thesis, we present a novel\napproach to bridging this gap through the design, fabrication, and detection of\nembedded machine-readable markers.\n  We categorize the proposed marking approaches into three distinct categories:\nnatural markers, structural markers, and internal markers. Natural markers,\nsuch as those used in SensiCut, are inherent fingerprints of objects repurposed\nas machine-readable identifiers, while structural markers, such as StructCode\nand G-ID, leverage the structural artifacts in objects that emerge during the\nfabrication process itself. Internal markers, such as InfraredTag and\nBrightMarker, are embedded inside fabricated objects using specialized\nmaterials. Leveraging a combination of methods from computer vision, machine\nlearning, computational imaging, and material science, the presented approaches\noffer robust and versatile solutions for object identification, tracking, and\ninteraction.\n  These markers, seamlessly integrated into real-world objects, effectively\ncommunicate an object's identity, origin, function, and interaction,\nfunctioning as gateways to \"ubiquitous metadata\" - a concept where metadata is\nembedded into physical objects, similar to metadata in digital files. Across\nthe different chapters, we demonstrate the applications of the presented\nmethods in diverse domains, including product design, manufacturing, retail,\nlogistics, education, entertainment, security, and sustainability.\n", "date": "2024-07-16", "categories": []}
{"id": "2502.01988", "title": "ReMiDi: Reconstruction of Microstructure Using a Differentiable\n  Diffusion MRI Simulator", "abstract": "  We propose ReMiDi, a novel method for inferring neuronal microstructure as\narbitrary 3D meshes using a differentiable diffusion Magnetic Resonance Imaging\n(dMRI) simulator. We first implemented in PyTorch a differentiable dMRI\nsimulator that simulates the forward diffusion process using a finite-element\nmethod on an input 3D microstructure mesh. To achieve significantly faster\nsimulations, we solve the differential equation semi-analytically using a\nmatrix formalism approach. Given a reference dMRI signal $S_{ref}$, we use the\ndifferentiable simulator to iteratively update the input mesh such that it\nmatches $S_{ref}$ using gradient-based learning. Since directly optimizing the\n3D coordinates of the vertices is challenging, particularly due to\nill-posedness of the inverse problem, we instead optimize a lower-dimensional\nlatent space representation of the mesh. The mesh is first encoded into\nspectral coefficients, which are further encoded into a latent $\\textbf{z}$\nusing an auto-encoder, and are then decoded back into the true mesh. We present\nan end-to-end differentiable pipeline that simulates signals that can be tuned\nto match a reference signal by iteratively updating the latent representation\n$\\textbf{z}$. We demonstrate the ability to reconstruct microstructures of\narbitrary shapes represented by finite-element meshes, with a focus on axonal\ngeometries found in the brain white matter, including bending, fanning and\nbeading fibers. Our source code is available online.\n", "date": "2025-02-04", "categories": []}
{"id": "2502.20877", "title": "Guiding Quantitative MRI Reconstruction with Phase-wise Uncertainty", "abstract": "  Quantitative magnetic resonance imaging (qMRI) requires multi-phase\nacqui-sition, often relying on reduced data sampling and reconstruction\nalgorithms to accelerate scans, which inherently poses an ill-posed inverse\nproblem. While many studies focus on measuring uncertainty during this process,\nfew explore how to leverage it to enhance reconstruction performance. In this\npaper, we in-troduce PUQ, a novel approach that pioneers the use of uncertainty\ninfor-mation for qMRI reconstruction. PUQ employs a two-stage reconstruction\nand parameter fitting framework, where phase-wise uncertainty is estimated\nduring reconstruction and utilized in the fitting stage. This design allows\nuncertainty to reflect the reliability of different phases and guide\ninformation integration during parameter fitting. We evaluated PUQ on in vivo\nT1 and T2 mapping datasets from healthy subjects. Compared to existing qMRI\nreconstruction methods, PUQ achieved the state-of-the-art performance in\nparameter map-pings, demonstrating the effectiveness of uncertainty guidance.\nOur code is available at https://anonymous.4open.science/r/PUQ-75B2/.\n", "date": "2025-02-28", "categories": []}
{"id": "2405.02497", "title": "Prediction techniques for dynamic imaging with online primal-dual\n  methods", "abstract": "  Online optimisation facilitates the solution of dynamic inverse problems,\nsuch as image stabilisation, fluid flow monitoring, and dynamic medical\nimaging. In this paper, we improve upon previous work on predictive online\nprimal-dual methods on two fronts. Firstly, we provide a more concise analysis\nthat symmetrises previously unsymmetric regret bounds, and relaxes previous\nrestrictive conditions on the dual predictor. Secondly, based on the latter, we\ndevelop several improved dual predictors. We numerically demonstrate their\nefficacy in image stabilisation and dynamic positron emission tomography.\n", "date": "2024-05-03", "categories": []}
{"id": "2409.18731", "title": "A Generalized Tensor Formulation for Hyperspectral Image\n  Super-Resolution Under General Spatial Blurring", "abstract": "  Hyperspectral super-resolution is commonly accomplished by the fusing of a\nhyperspectral imaging of low spatial resolution with a multispectral image of\nhigh spatial resolution, and many tensor-based approaches to this task have\nbeen recently proposed. Yet, it is assumed in such tensor-based methods that\nthe spatial-blurring operation that creates the observed hyperspectral image\nfrom the desired super-resolved image is separable into independent horizontal\nand vertical blurring. Recent work has argued that such separable spatial\ndegradation is ill-equipped to model the operation of real sensors which may\nexhibit, for example, anisotropic blurring. To accommodate this fact, a\ngeneralized tensor formulation based on a Kronecker decomposition is proposed\nto handle any general spatial-degradation matrix, including those that are not\nseparable as previously assumed. Analysis of the generalized formulation\nreveals conditions under which exact recovery of the desired super-resolved\nimage is guaranteed, and a practical algorithm for such recovery, driven by a\nblockwise-group-sparsity regularization, is proposed. Extensive experimental\nresults demonstrate that the proposed generalized tensor approach outperforms\nnot only traditional matrix-based techniques but also state-of-the-art\ntensor-based methods; the gains with respect to the latter are especially\nsignificant in cases of anisotropic spatial blurring.\n", "date": "2024-09-27", "categories": []}
{"id": "2502.19451", "title": "Multispectral to Hyperspectral using Pretrained Foundational model", "abstract": "  Hyperspectral imaging provides detailed spectral information, offering\nsignificant potential for monitoring greenhouse gases like CH4 and NO2.\nHowever, its application is constrained by limited spatial coverage and\ninfrequent revisit times. In contrast, multispectral imaging delivers broader\nspatial and temporal coverage but lacks the spectral granularity required for\nprecise GHG detection. To address these challenges, this study proposes\nSpectral and Spatial-Spectral transformer models that reconstruct hyperspectral\ndata from multispectral inputs. The models in this paper are pretrained on\nEnMAP and EMIT datasets and fine-tuned on spatio-temporally aligned\n(Sentinel-2, EnMAP) and (HLS-S30, EMIT) image pairs respectively. Our model has\nthe potential to enhance atmospheric monitoring by combining the strengths of\nhyperspectral and multispectral imaging systems.\n", "date": "2025-02-26", "categories": []}
{"id": "2502.19683", "title": "Dual-branch Graph Feature Learning for NLOS Imaging", "abstract": "  The domain of non-line-of-sight (NLOS) imaging is advancing rapidly, offering\nthe capability to reveal occluded scenes that are not directly visible.\nHowever, contemporary NLOS systems face several significant challenges: (1) The\ncomputational and storage requirements are profound due to the inherent\nthree-dimensional grid data structure, which restricts practical application.\n(2) The simultaneous reconstruction of albedo and depth information requires a\ndelicate balance using hyperparameters in the loss function, rendering the\nconcurrent reconstruction of texture and depth information difficult. This\npaper introduces the innovative methodology, \\xnet, which integrates an\nalbedo-focused reconstruction branch dedicated to albedo information recovery\nand a depth-focused reconstruction branch that extracts geometrical structure,\nto overcome these obstacles. The dual-branch framework segregates content\ndelivery to the respective reconstructions, thereby enhancing the quality of\nthe retrieved data. To our knowledge, we are the first to employ the GNN as a\nfundamental component to transform dense NLOS grid data into sparse structural\nfeatures for efficient reconstruction. Comprehensive experiments demonstrate\nthat our method attains the highest level of performance among existing methods\nacross synthetic and real data. https://github.com/Nicholassu/DG-NLOS.\n", "date": "2025-02-27", "categories": []}
{"id": "2502.19194", "title": "Self-supervised conformal prediction for uncertainty quantification in\n  Poisson imaging problems", "abstract": "  Image restoration problems are often ill-posed, leading to significant\nuncertainty in reconstructed images. Accurately quantifying this uncertainty is\nessential for the reliable interpretation of reconstructed images. However,\nimage restoration methods often lack uncertainty quantification capabilities.\nConformal prediction offers a rigorous framework to augment image restoration\nmethods with accurate uncertainty quantification estimates, but it typically\nrequires abundant ground truth data for calibration. This paper presents a\nself-supervised conformal prediction method for Poisson imaging problems which\nleverages Poisson Unbiased Risk Estimator to eliminate the need for ground\ntruth data. The resulting self-calibrating conformal prediction approach is\napplicable to any Poisson linear imaging problem that is ill-conditioned, and\nis particularly effective when combined with modern self-supervised image\nrestoration techniques trained directly on measurement data. The proposed\nmethod is demonstrated through numerical experiments on image denoising and\ndeblurring; its performance are comparable to supervised conformal prediction\nmethods relying on ground truth data.\n", "date": "2025-02-26", "categories": []}
{"id": "2502.19238", "title": "Arbitrary Volumetric Refocusing of Dense and Sparse Light Fields", "abstract": "  A four-dimensional light field (LF) captures both textural and geometrical\ninformation of a scene in contrast to a two-dimensional image that captures\nonly the textural information of a scene. Post-capture refocusing is an\nexciting application of LFs enabled by the geometric information captured.\nPreviously proposed LF refocusing methods are mostly limited to the refocusing\nof single planar or volumetric region of a scene corresponding to a depth range\nand cannot simultaneously generate in-focus and out-of-focus regions having the\nsame depth range. In this paper, we propose an end-to-end pipeline to\nsimultaneously refocus multiple arbitrary planar or volumetric regions of a\ndense or a sparse LF. We employ pixel-dependent shifts with the typical\nshift-and-sum method to refocus an LF. The pixel-dependent shifts enables to\nrefocus each pixel of an LF independently. For sparse LFs, the shift-and-sum\nmethod introduces ghosting artifacts due to the spatial undersampling. We\nemploy a deep learning model based on U-Net architecture to almost completely\neliminate the ghosting artifacts. The experimental results obtained with\nseveral LF datasets confirm the effectiveness of the proposed method. In\nparticular, sparse LFs refocused with the proposed method archive structural\nsimilarity index higher than 0.9 despite having only 20% of data compared to\ndense LFs.\n", "date": "2025-02-26", "categories": []}
{"id": "2502.15986", "title": "Improved Partial Differential Equation and Fast Approximation Algorithm\n  for Hazy/Underwater/Dust Storm Image Enhancement", "abstract": "  This paper presents an improved and modified partial differential equation\n(PDE)-based de-hazing algorithm. The proposed method combines logarithmic image\nprocessing models in a PDE formulation refined with linear filter-based\noperators in either spatial or frequency domain. Additionally, a fast,\nsimplified de-hazing function approximation of the hazy image formation model\nis developed in combination with fuzzy homomorphic refinement. The proposed\nalgorithm solves the problem of image darkening and over-enhancement of edges\nin addition to enhancement of dark image regions encountered in previous\nformulations. This is in addition to avoiding enhancement of sky regions in\nde-hazed images while avoiding halo effect. Furthermore, the proposed algorithm\nis utilized for underwater and dust storm image enhancement with the\nincorporation of a modified global contrast enhancement algorithm. Experimental\ncomparisons indicate that the proposed approach surpasses a majority of the\nalgorithms from the literature based on quantitative image quality metrics.\n", "date": "2025-02-21", "categories": []}
{"id": "2404.19201", "title": "Exploring Quasi-Global Solutions to Compound Lens Based Computational\n  Imaging Systems", "abstract": "  Recently, joint design approaches that simultaneously optimize optical\nsystems and downstream algorithms through data-driven learning have\ndemonstrated superior performance over traditional separate design approaches.\nHowever, current joint design approaches heavily rely on the manual\nidentification of initial lenses, posing challenges and limitations,\nparticularly for compound lens systems with multiple potential starting points.\nIn this work, we present Quasi-Global Search Optics (QGSO) to automatically\ndesign compound lens based computational imaging systems through two parts: (i)\nFused Optimization Method for Automatic Optical Design (OptiFusion), which\nsearches for diverse initial optical systems under certain design\nspecifications; and (ii) Efficient Physic-aware Joint Optimization (EPJO),\nwhich conducts parallel joint optimization of initial optical systems and image\nreconstruction networks with the consideration of physical constraints,\nculminating in the selection of the optimal solution in all search results.\nExtensive experimental results illustrate that QGSO serves as a transformative\nend-to-end lens design paradigm for superior global search ability, which\nautomatically provides compound lens based computational imaging systems with\nhigher imaging quality compared to existing paradigms. The source code will be\nmade publicly available at https://github.com/LiGpy/QGSO.\n", "date": "2024-04-30", "categories": []}
{"id": "2409.04679", "title": "Neural Augmentation Based Panoramic High Dynamic Range Stitching", "abstract": "  Due to saturated regions of inputting low dynamic range (LDR) images and\nlarge intensity changes among the LDR images caused by different exposures, it\nis challenging to produce an information enriched panoramic LDR image without\nvisual artifacts for a high dynamic range (HDR) scene through stitching\nmultiple geometrically synchronized LDR images with different exposures and\npairwise overlapping fields of views (OFOVs). Fortunately, the stitching of\nsuch images is innately a perfect scenario for the fusion of a physics-driven\napproach and a data-driven approach due to their OFOVs. Based on this new\ninsight, a novel neural augmentation based panoramic HDR stitching algorithm is\nproposed in this paper. The physics-driven approach is built up using the\nOFOVs. Different exposed images of each view are initially generated by using\nthe physics-driven approach, are then refined by a data-driven approach, and\nare finally used to produce panoramic LDR images with different exposures. All\nthe panoramic LDR images with different exposures are combined together via a\nmulti-scale exposure fusion algorithm to produce the final panoramic LDR image.\nExperimental results demonstrate the proposed algorithm outperforms existing\npanoramic stitching algorithms.\n", "date": "2024-09-07", "categories": []}
{"id": "2502.14920", "title": "Display Field-Of-View Agnostic Robust CT Kernel Synthesis Using\n  Model-Based Deep Learning", "abstract": "  In X-ray computed tomography (CT) imaging, the choice of reconstruction\nkernel is crucial as it significantly impacts the quality of clinical images.\nDifferent kernels influence spatial resolution, image noise, and contrast in\nvarious ways. Clinical applications involving lung imaging often require images\nreconstructed with both soft and sharp kernels. The reconstruction of images\nwith different kernels requires raw sinogram data and storing images for all\nkernels increases processing time and storage requirements. The Display\nField-of-View (DFOV) adds complexity to kernel synthesis, as data acquired at\ndifferent DFOVs exhibit varying levels of sharpness and details. This work\nintroduces an efficient, DFOV-agnostic solution for image-based kernel\nsynthesis using model-based deep learning. The proposed method explicitly\nintegrates CT kernel and DFOV characteristics into the forward model.\nExperimental results on clinical data, along with quantitative analysis of the\nestimated modulation transfer function using wire phantom data, clearly\ndemonstrate the utility of the proposed method in real-time. Additionally, a\ncomparative study with a direct learning network, that lacks forward model\ninformation, shows that the proposed method is more robust to DFOV variations.\n", "date": "2025-02-19", "categories": []}
{"id": "2502.15397", "title": "Super-Resolution for Interferometric Imaging: Model Comparisons and\n  Performance Analysis", "abstract": "  This study investigates the application of Super-Resolution techniques in\nholographic microscopy to enhance quantitative phase imaging. An off-axis\nMach-Zehnder interferometric setup was employed to capture interferograms. The\nstudy evaluates two Super-Resolution models, RCAN and Real-ESRGAN, for their\neffectiveness in reconstructing high-resolution interferograms from a\nmicroparticle-based dataset. The models were assessed using two primary\napproaches: image-based analysis for structural detail enhancement and\nmorphological evaluation for maintaining sample integrity and phase map\naccuracy. The results demonstrate that RCAN achieves superior numerical\nprecision, making it ideal for applications requiring highly accurate phase map\nreconstruction, while Real-ESRGAN enhances visual quality and structural\ncoherence, making it suitable for visualization-focused applications. This\nstudy highlights the potential of Super-Resolution models in overcoming\ndiffraction-imposed resolution limitations in holographic microscopy, opening\nthe way for improved imaging techniques in biomedical diagnostics, materials\nscience, and other high-precision fields.\n", "date": "2025-02-21", "categories": []}
{"id": "2210.06330", "title": "CoRRECT: A Deep Unfolding Framework for Motion-Corrected Quantitative\n  R2* Mapping", "abstract": "  Quantitative MRI (qMRI) refers to a class of MRI methods for quantifying the\nspatial distribution of biological tissue parameters. Traditional qMRI methods\nusually deal separately with artifacts arising from accelerated data\nacquisition, involuntary physical motion, and magnetic-field inhomogeneities,\nleading to suboptimal end-to-end performance. This paper presents CoRRECT, a\nunified deep unfolding (DU) framework for qMRI consisting of a model-based\nend-to-end neural network, a method for motion-artifact reduction, and a\nself-supervised learning scheme. The network is trained to produce R2* maps\nwhose k-space data matches the real data by also accounting for motion and\nfield inhomogeneities. When deployed, CoRRECT only uses the k-space data\nwithout any pre-computed parameters for motion or inhomogeneity correction. Our\nresults on experimentally collected multi-Gradient-Recalled Echo (mGRE) MRI\ndata show that CoRRECT recovers motion and inhomogeneity artifact-free R2* maps\nin highly accelerated acquisition settings. This work opens the door to DU\nmethods that can integrate physical measurement models, biophysical signal\nmodels, and learned prior models for high-quality qMRI.\n", "date": "2022-10-12", "categories": []}
{"id": "2502.14209", "title": "Spatial and Frequency Domain Adaptive Fusion Network for Image\n  Deblurring", "abstract": "  Image deblurring aims to reconstruct a latent sharp image from its\ncorresponding blurred one. Although existing methods have achieved good\nperformance, most of them operate exclusively in either the spatial domain or\nthe frequency domain, rarely exploring solutions that fuse both domains. In\nthis paper, we propose a spatial-frequency domain adaptive fusion network\n(SFAFNet) to address this limitation. Specifically, we design a gated\nspatial-frequency domain feature fusion block (GSFFBlock), which consists of\nthree key components: a spatial domain information module, a frequency domain\ninformation dynamic generation module (FDGM), and a gated fusion module (GFM).\nThe spatial domain information module employs the NAFBlock to integrate local\ninformation. Meanwhile, in the FDGM, we design a learnable low-pass filter that\ndynamically decomposes features into separate frequency subbands, capturing the\nimage-wide receptive field and enabling the adaptive exploration of global\ncontextual information. Additionally, to facilitate information flow and the\nlearning of complementary representations. In the GFM, we present a gating\nmechanism (GATE) to re-weight spatial and frequency domain features, which are\nthen fused through the cross-attention mechanism (CAM). Experimental results\ndemonstrate that our SFAFNet performs favorably compared to state-of-the-art\napproaches on commonly used benchmarks.\n", "date": "2025-02-20", "categories": []}
{"id": "2502.14462", "title": "Single-image Reflectance and Transmittance Estimation from Any Flatbed\n  Scanner", "abstract": "  Flatbed scanners have emerged as promising devices for high-resolution,\nsingle-image material capture. However, existing approaches assume very\nspecific conditions, such as uniform diffuse illumination, which are only\navailable in certain high-end devices, hindering their scalability and cost. In\ncontrast, in this work, we introduce a method inspired by intrinsic image\ndecomposition, which accurately removes both shading and specularity,\neffectively allowing captures with any flatbed scanner. Further, we extend\nprevious work on single-image material reflectance capture with the estimation\nof opacity and transmittance, critical components of full material appearance\n(SVBSDF), improving the results for any material captured with a flatbed\nscanner, at a very high resolution and accuracy\n", "date": "2025-02-20", "categories": []}
{"id": "2502.12293", "title": "Data-Efficient Limited-Angle CT Using Deep Priors and Regularization", "abstract": "  Reconstructing an image from its Radon transform is a fundamental computed\ntomography (CT) task arising in applications such as X-ray scans. In many\npractical scenarios, a full 180-degree scan is not feasible, or there is a\ndesire to reduce radiation exposure. In these limited-angle settings, the\nproblem becomes ill-posed, and methods designed for full-view data often leave\nsignificant artifacts. We propose a very low-data approach to reconstruct the\noriginal image from its Radon transform under severe angle limitations. Because\nthe inverse problem is ill-posed, we combine multiple regularization methods,\nincluding Total Variation, a sinogram filter, Deep Image Prior, and a\npatch-level autoencoder. We use a differentiable implementation of the Radon\ntransform, which allows us to use gradient-based techniques to solve the\ninverse problem. Our method is evaluated on a dataset from the Helsinki\nTomography Challenge 2022, where the goal is to reconstruct a binary disk from\nits limited-angle sinogram. We only use a total of 12 data points--eight for\nlearning a prior and four for hyperparameter selection--and achieve results\ncomparable to the best synthetic data-driven approaches.\n", "date": "2025-02-17", "categories": []}
{"id": "2302.00816", "title": "Dynamic Atomic Column Detection in Transmission Electron Microscopy\n  Videos via Ridge Estimation", "abstract": "  Ridge detection is a classical tool to extract curvilinear features in image\nprocessing. As such, it has great promise in applications to material science\nproblems; specifically, for trend filtering relatively stable atom-shaped\nobjects in image sequences, such as Transmission Electron Microscopy (TEM)\nvideos. Standard analysis of TEM videos is limited to frame-by-frame object\nrecognition. We instead harness temporal correlation across frames through\nsimultaneous analysis of long image sequences, specified as a spatio-temporal\nimage tensor. We define new ridge detection algorithms to non-parametrically\nestimate explicit trajectories of atomic-level object locations as a continuous\nfunction of time. Our approach is specially tailored to handle temporal\nanalysis of objects that seemingly stochastically disappear and subsequently\nreappear throughout a sequence. We demonstrate that the proposed method is\nhighly effective and efficient in simulation scenarios, and delivers notable\nperformance improvements in TEM experiments compared to other material science\nbenchmarks.\n", "date": "2023-02-02", "categories": []}
{"id": "2405.09873", "title": "IRSRMamba: Infrared Image Super-Resolution via Mamba-based Wavelet\n  Transform Feature Modulation Model", "abstract": "  Infrared image super-resolution demands long-range dependency modeling and\nmulti-scale feature extraction to address challenges such as homogeneous\nbackgrounds, weak edges, and sparse textures. While Mamba-based state-space\nmodels (SSMs) excel in global dependency modeling with linear complexity, their\nblock-wise processing disrupts spatial consistency, limiting their\neffectiveness for IR image reconstruction. We propose IRSRMamba, a novel\nframework integrating wavelet transform feature modulation for multi-scale\nadaptation and an SSMs-based semantic consistency loss to restore fragmented\ncontextual information. This design enhances global-local feature fusion,\nstructural coherence, and fine-detail preservation while mitigating\nblock-induced artifacts. Experiments on benchmark datasets demonstrate that\nIRSRMamba outperforms state-of-the-art methods in PSNR, SSIM, and perceptual\nquality. This work establishes Mamba-based architectures as a promising\ndirection for high-fidelity IR image enhancement. Code are available at\nhttps://github.com/yongsongH/IRSRMamba.\n", "date": "2024-05-16", "categories": []}
{"id": "2502.10714", "title": "Disentangle Nighttime Lens Flares: Self-supervised Generation-based Lens\n  Flare Removal", "abstract": "  Lens flares arise from light reflection and refraction within sensor arrays,\nwhose diverse types include glow, veiling glare, reflective flare and so on.\nExisting methods are specialized for one specific type only, and overlook the\nsimultaneous occurrence of multiple typed lens flares, which is common in the\nreal-world, e.g. coexistence of glow and displacement reflections from the same\nlight source. These co-occurring lens flares cannot be effectively resolved by\nthe simple combination of individual flare removal methods, since these\ncoexisting flares originates from the same light source and are generated\nsimultaneously within the same sensor array, exhibit a complex interdependence\nrather than simple additive relation. To model this interdependent flare\nrelationship, our Nighttime Lens Flare Formation model is the first attempt to\nlearn the intrinsic physical relationship between flares on the imaging plane.\nBuilding on this physical model, we introduce a solution to this joint flare\nremoval task named Self-supervised Generation-based Lens Flare Removal Network\n(SGLFR-Net), which is self-supervised without pre-training. Specifically, the\nnighttime glow is detangled in PSF Rendering Network(PSFR-Net) based on PSF\nRendering Prior, while the reflective flare is modelled in Texture Prior Based\nReflection Flare Removal Network (TPRR-Net). Empirical evaluations demonstrate\nthe effectiveness of the proposed method in both joint and individual glare\nremoval tasks.\n", "date": "2025-02-15", "categories": []}
{"id": "2502.10876", "title": "Super Resolution image reconstructs via total variation-based image\n  deconvolution: a majorization-minimization approach", "abstract": "  This work aims to reconstruct image sequences with Total Variation regularity\nin super-resolution. We consider, in particular, images of scenes for which the\npoint-to-point image transformation is a plane projective transformation. We\nfirst describe the super-resolution image's imaging observation model, an\ninterpolation and Fusion estimator, and Projection on Convex Sets. We explain\nmotion and compute the optical flow of a sequence of images using the\nHorn-Shunck algorithm to estimate motion. We then propose a Total Variation\nregulazer via a Majorization-Minimization approach to obtain a suitable result.\nSuper Resolution restoration from motion measurements is also discussed.\nFinally, the simulation's part demonstrates the power of the proposed\nmethodology. As expected, this model does not give real-time results, as seen\nin the numerical experiments section, but it is the cornerstone for future\napproaches. Finally, the simulation's part demonstrates the power of the\nproposed methodology. As expected, this model does not give real-time results,\nas seen in the numerical experiments section, but it is the cornerstone for\nfuture approaches.\n", "date": "2025-02-15", "categories": []}
{"id": "2502.11749", "title": "JotlasNet: Joint Tensor Low-Rank and Attention-based Sparse Unrolling\n  Network for Accelerating Dynamic MRI", "abstract": "  Joint low-rank and sparse unrolling networks have shown superior performance\nin dynamic MRI reconstruction. However, existing works mainly utilized matrix\nlow-rank priors, neglecting the tensor characteristics of dynamic MRI images,\nand only a global threshold is applied for the sparse constraint to the\nmulti-channel data, limiting the flexibility of the network. Additionally, most\nof them have inherently complex network structure, with intricate interactions\namong variables. In this paper, we propose a novel deep unrolling network,\nJotlasNet, for dynamic MRI reconstruction by jointly utilizing tensor low-rank\nand attention-based sparse priors. Specifically, we utilize tensor low-rank\nprior to exploit the structural correlations in high-dimensional data.\nConvolutional neural networks are used to adaptively learn the low-rank and\nsparse transform domains. A novel attention-based soft thresholding operator is\nproposed to assign a unique learnable threshold to each channel of the data in\nthe CNN-learned sparse domain. The network is unrolled from the elaborately\ndesigned composite splitting algorithm and thus features a simple yet efficient\nparallel structure. Extensive experiments on two datasets (OCMR, CMRxRecon)\ndemonstrate the superior performance of JotlasNet in dynamic MRI\nreconstruction.\n", "date": "2025-02-17", "categories": []}
{"id": "2502.11913", "title": "PreAdaptFWI: Pretrained-Based Adaptive Residual Learning for\n  Full-Waveform Inversion Without Dataset Dependency", "abstract": "  Full-waveform inversion (FWI) is a method that utilizes seismic data to\ninvert the physical parameters of subsurface media by minimizing the difference\nbetween simulated and observed waveforms. Due to its ill-posed nature, FWI is\nsusceptible to getting trapped in local minima. Consequently, various research\nefforts have attempted to combine neural networks with FWI to stabilize the\ninversion process. This study presents a simple yet effective training\nframework that is independent of dataset reliance and requires only moderate\npre-training on a simple initial model to stabilize network outputs. During the\ntransfer learning phase, the conventional FWI gradients will simultaneously\nupdate both the neural network and the proposed adaptive residual learning\nmodule, which learns the residual mapping of large-scale distribution features\nin the network's output, rather than directly fitting the target mapping.\nThrough this synergistic training paradigm, the proposed algorithm effectively\ninfers the physically-informed prior knowledge into a global representation of\nstratigraphic distribution, as well as capturing subtle variations in\ninter-layer velocities within local details, thereby escaping local optima.\nEvaluating the method on two benchmark models under various conditions,\nincluding absent low-frequency data, noise interference, and differing initial\nmodels, along with corresponding ablation experiments, consistently\ndemonstrates the superiority of the proposed approach.\n", "date": "2025-02-17", "categories": []}
{"id": "2408.16868", "title": "Characterization of point-source transient events with a rolling-shutter\n  compressed sensing system", "abstract": "  Point-source transient events (PSTEs) - optical events that are both\nextremely fast and extremely small - pose several challenges to an imaging\nsystem. Due to their speed, accurately characterizing such events often\nrequires detectors with very high frame rates. Due to their size, accurately\ndetecting such events requires maintaining coverage over an extended\nfield-of-view, often through the use of imaging focal plane arrays (FPA) with a\nglobal shutter readout. Traditional imaging systems that meet these\nrequirements are costly in terms of price, size, weight, power consumption, and\ndata bandwidth, and there is a need for cheaper solutions with adequate\ntemporal and spatial coverage. To address these issues, we develop a novel\ncompressed sensing algorithm adapted to the rolling shutter readout of an\nimaging system. This approach enables reconstruction of a PSTE signature at the\nsampling rate of the rolling shutter, offering a 1-2 order of magnitude\ntemporal speedup and a proportional reduction in data bandwidth. We present\nempirical results demonstrating accurate recovery of PSTEs using measurements\nthat are spatially undersampled by a factor of 25, and our simulations show\nthat, relative to other compressed sensing algorithms, our algorithm is both\nfaster and yields higher quality reconstructions. We also present theoretical\nresults characterizing our algorithm and corroborating simulations. The\npotential impact of our work includes the development of much faster, cheaper\nsensor solutions for PSTE detection and characterization.\n", "date": "2024-08-29", "categories": []}
{"id": "2409.09754", "title": "Towards Single-Lens Controllable Depth-of-Field Imaging via Depth-Aware\n  Point Spread Functions", "abstract": "  Controllable Depth-of-Field (DoF) imaging commonly produces amazing visual\neffects based on heavy and expensive high-end lenses. However, confronted with\nthe increasing demand for mobile scenarios, it is desirable to achieve a\nlightweight solution with Minimalist Optical Systems (MOS). This work centers\naround two major limitations of MOS, i.e., the severe optical aberrations and\nuncontrollable DoF, for achieving single-lens controllable DoF imaging via\ncomputational methods. A Depth-aware Controllable DoF Imaging (DCDI) framework\nis proposed equipped with All-in-Focus (AiF) aberration correction and\nmonocular depth estimation, where the recovered image and corresponding depth\nmap are utilized to produce imaging results under diverse DoFs of any high-end\nlens via patch-wise convolution. To address the depth-varying optical\ndegradation, we introduce a Depth-aware Degradation-adaptive Training (DA2T)\nscheme. At the dataset level, a Depth-aware Aberration MOS (DAMOS) dataset is\nestablished based on the simulation of Point Spread Functions (PSFs) under\ndifferent object distances. Additionally, we design two plug-and-play\ndepth-aware mechanisms to embed depth information into the aberration image\nrecovery for better tackling depth-aware degradation. Furthermore, we propose a\nstorage-efficient Omni-Lens-Field model to represent the 4D PSF library of\nvarious lenses. With the predicted depth map, recovered image, and depth-aware\nPSF map inferred by Omni-Lens-Field, single-lens controllable DoF imaging is\nachieved. Comprehensive experimental results demonstrate that the proposed\nframework enhances the recovery performance, and attains impressive single-lens\ncontrollable DoF imaging results, providing a seminal baseline for this field.\nThe source code and the established dataset will be publicly available at\nhttps://github.com/XiaolongQian/DCDI.\n", "date": "2024-09-15", "categories": []}
{"id": "2502.06894", "title": "AI-Driven HSI: Multimodality, Fusion, Challenges, and the Deep Learning\n  Revolution", "abstract": "  Hyperspectral imaging (HSI) captures spatial and spectral data, enabling\nanalysis of features invisible to conventional systems. The technology is vital\nin fields such as weather monitoring, food quality control, counterfeit\ndetection, healthcare diagnostics, and extending into defense, agriculture, and\nindustrial automation at the same time. HSI has advanced with improvements in\nspectral resolution, miniaturization, and computational methods. This study\nprovides an overview of the HSI, its applications, challenges in data fusion\nand the role of deep learning models in processing HSI data. We discuss how\nintegration of multimodal HSI with AI, particularly with deep learning,\nimproves classification accuracy and operational efficiency. Deep learning\nenhances HSI analysis in areas like feature extraction, change detection,\ndenoising unmixing, dimensionality reduction, landcover mapping, data\naugmentation, spectral construction and super resolution. An emerging focus is\nthe fusion of hyperspectral cameras with large language models (LLMs), referred\nas highbrain LLMs, enabling the development of advanced applications such as\nlow visibility crash detection and face antispoofing. We also highlight key\nplayers in HSI industry, its compound annual growth rate and the growing\nindustrial significance. The purpose is to offer insight to both technical and\nnon-technical audience, covering HSI's images, trends, and future directions,\nwhile providing valuable information on HSI datasets and software libraries.\n", "date": "2025-02-09", "categories": []}
{"id": "2502.07403", "title": "Extended monocular 3D imaging", "abstract": "  3D vision is of paramount importance for numerous applications ranging from\nmachine intelligence to precision metrology. Despite much recent progress, the\nmajority of 3D imaging hardware remains bulky and complicated and provides much\nlower image resolution compared to their 2D counterparts. Moreover, there are\nmany well-known scenarios that existing 3D imaging solutions frequently fail.\nHere, we introduce an extended monocular 3D imaging (EM3D) framework that fully\nexploits the vectorial wave nature of light. Via the multi-stage fusion of\ndiffraction- and polarization-based depth cues, using a compact monocular\ncamera equipped with a diffractive-refractive hybrid lens, we experimentally\ndemonstrate the snapshot acquisition of a million-pixel and accurate 3D point\ncloud for extended scenes that are traditionally challenging, including those\nwith low texture, being highly reflective, or nearly transparent, without a\ndata prior. Furthermore, we discover that the combination of depth and\npolarization information can unlock unique new opportunities in material\nidentification, which may further expand machine intelligence for applications\nlike target recognition and face anti-spoofing. The straightforward yet\npowerful architecture thus opens up a new path for a higher-dimensional machine\nvision in a minimal form factor, facilitating the deployment of monocular\ncameras for applications in much more diverse scenarios.\n", "date": "2025-02-11", "categories": []}
{"id": "2502.06338", "title": "Zero-shot Depth Completion via Test-time Alignment with Affine-invariant\n  Depth Prior", "abstract": "  Depth completion, predicting dense depth maps from sparse depth measurements,\nis an ill-posed problem requiring prior knowledge. Recent methods adopt\nlearning-based approaches to implicitly capture priors, but the priors\nprimarily fit in-domain data and do not generalize well to out-of-domain\nscenarios. To address this, we propose a zero-shot depth completion method\ncomposed of an affine-invariant depth diffusion model and test-time alignment.\nWe use pre-trained depth diffusion models as depth prior knowledge, which\nimplicitly understand how to fill in depth for scenes. Our approach aligns the\naffine-invariant depth prior with metric-scale sparse measurements, enforcing\nthem as hard constraints via an optimization loop at test-time. Our zero-shot\ndepth completion method demonstrates generalization across various domain\ndatasets, achieving up to a 21\\% average performance improvement over the\nprevious state-of-the-art methods while enhancing spatial understanding by\nsharpening scene details. We demonstrate that aligning a monocular\naffine-invariant depth prior with sparse metric measurements is a proven\nstrategy to achieve domain-generalizable depth completion without relying on\nextensive training data. Project page:\nhttps://hyoseok1223.github.io/zero-shot-depth-completion/.\n", "date": "2025-02-10", "categories": []}
{"id": "2502.06354", "title": "Guidance-base Diffusion Models for Improving Photoacoustic Image Quality", "abstract": "  Photoacoustic(PA) imaging is a non-destructive and non-invasive technology\nfor visualizing minute blood vessel structures in the body using ultrasonic\nsensors. In PA imaging, the image quality of a single-shot image is poor, and\nit is necessary to improve the image quality by averaging many single-shot\nimages. Therefore, imaging the entire subject requires high imaging costs. In\nour study, we propose a method to improve the quality of PA images using\ndiffusion models. In our method, we improve the reverse diffusion process using\nsensor information of PA imaging and introduce a guidance method using imaging\ncondition information to generate high-quality images.\n", "date": "2025-02-10", "categories": []}
{"id": "2502.06452", "title": "SparseFocus: Learning-based One-shot Autofocus for Microscopy with\n  Sparse Content", "abstract": "  Autofocus is necessary for high-throughput and real-time scanning in\nmicroscopic imaging. Traditional methods rely on complex hardware or iterative\nhill-climbing algorithms. Recent learning-based approaches have demonstrated\nremarkable efficacy in a one-shot setting, avoiding hardware modifications or\niterative mechanical lens adjustments. However, in this paper, we highlight a\nsignificant challenge that the richness of image content can significantly\naffect autofocus performance. When the image content is sparse, previous\nautofocus methods, whether traditional climbing-hill or learning-based, tend to\nfail. To tackle this, we propose a content-importance-based solution, named\nSparseFocus, featuring a novel two-stage pipeline. The first stage measures the\nimportance of regions within the image, while the second stage calculates the\ndefocus distance from selected important regions. To validate our approach and\nbenefit the research community, we collect a large-scale dataset comprising\nmillions of labelled defocused images, encompassing both dense, sparse and\nextremely sparse scenarios. Experimental results show that SparseFocus\nsurpasses existing methods, effectively handling all levels of content\nsparsity. Moreover, we integrate SparseFocus into our Whole Slide Imaging (WSI)\nsystem that performs well in real-world applications. The code and dataset will\nbe made available upon the publication of this paper.\n", "date": "2025-02-10", "categories": []}
{"id": "2501.18401", "title": "MatIR: A Hybrid Mamba-Transformer Image Restoration Model", "abstract": "  In recent years, Transformers-based models have made significant progress in\nthe field of image restoration by leveraging their inherent ability to capture\ncomplex contextual features. Recently, Mamba models have made a splash in the\nfield of computer vision due to their ability to handle long-range dependencies\nand their significant computational efficiency compared to Transformers.\nHowever, Mamba currently lags behind Transformers in contextual learning\ncapabilities. To overcome the limitations of these two models, we propose a\nMamba-Transformer hybrid image restoration model called MatIR. Specifically,\nMatIR cross-cycles the blocks of the Transformer layer and the Mamba layer to\nextract features, thereby taking full advantage of the advantages of the two\narchitectures. In the Mamba module, we introduce the Image Inpainting State\nSpace (IRSS) module, which traverses along four scan paths to achieve efficient\nprocessing of long sequence data. In the Transformer module, we combine\ntriangular window-based local attention with channel-based global attention to\neffectively activate the attention mechanism over a wider range of image\npixels. Extensive experimental results and ablation studies demonstrate the\neffectiveness of our approach.\n", "date": "2025-01-30", "categories": []}
{"id": "2502.04713", "title": "Leveraging band diversity for feature selection in EO data", "abstract": "  Hyperspectral imaging (HSI) is a powerful earth observation technology that\ncaptures and processes information across a wide spectrum of wavelengths.\nHyperspectral imaging provides comprehensive and detailed spectral data that is\ninvaluable for a wide range of reconstruction problems. However due to\ncomplexity in analysis it often becomes difficult to handle this data. To\naddress the challenge of handling large number of bands in reconstructing high\nquality HSI, we propose to form groups of bands. In this position paper we\npropose a method of selecting diverse bands using determinantal point processes\nin correlated bands. To address the issue of overlapping bands that may arise\nfrom grouping, we use spectral angle mapper analysis. This analysis can be fed\nto any Machine learning model to enable detailed analysis and monitoring with\nhigh precision and accuracy.\n", "date": "2025-02-07", "categories": []}
{"id": "2502.04719", "title": "Tolerance-Aware Deep Optics", "abstract": "  Deep optics has emerged as a promising approach by co-designing optical\nelements with deep learning algorithms. However, current research typically\noverlooks the analysis and optimization of manufacturing and assembly\ntolerances. This oversight creates a significant performance gap between\ndesigned and fabricated optical systems. To address this challenge, we present\nthe first end-to-end tolerance-aware optimization framework that incorporates\nmultiple tolerance types into the deep optics design pipeline. Our method\ncombines physics-informed modelling with data-driven training to enhance\noptical design by accounting for and compensating for structural deviations in\nmanufacturing and assembly. We validate our approach through computational\nimaging applications, demonstrating results in both simulations and real-world\nexperiments. We further examine how our proposed solution improves the\nrobustness of optical systems and vision algorithms against tolerances through\nqualitative and quantitative analyses. Code and additional visual results are\navailable at openimaginglab.github.io/LensTolerance.\n", "date": "2025-02-07", "categories": []}
{"id": "2502.05055", "title": "Differentiable Mobile Display Photometric Stereo", "abstract": "  Display photometric stereo uses a display as a programmable light source to\nilluminate a scene with diverse illumination conditions. Recently,\ndifferentiable display photometric stereo (DDPS) demonstrated improved normal\nreconstruction accuracy by using learned display patterns. However, DDPS faced\nlimitations in practicality, requiring a fixed desktop imaging setup using a\npolarization camera and a desktop-scale monitor. In this paper, we propose a\nmore practical physics-based photometric stereo, differentiable mobile display\nphotometric stereo (DMDPS), that leverages a mobile phone consisting of a\ndisplay and a camera. We overcome the limitations of using a mobile device by\ndeveloping a mobile app and method that simultaneously displays patterns and\ncaptures high-quality HDR images. Using this technique, we capture real-world\n3D-printed objects and learn display patterns via a differentiable learning\nprocess. We demonstrate the effectiveness of DMDPS on both a 3D printed dataset\nand a first dataset of fallen leaves. The leaf dataset contains reconstructed\nsurface normals and albedos of fallen leaves that may enable future research\nbeyond computer graphics and vision. We believe that DMDPS takes a step forward\nfor practical physics-based photometric stereo.\n", "date": "2025-02-07", "categories": []}
{"id": "2408.11323", "title": "Optimizing Transmit Field Inhomogeneity of Parallel RF Transmit Design\n  in 7T MRI using Deep Learning", "abstract": "  Ultrahigh field (UHF) Magnetic Resonance Imaging (MRI) provides a higher\nsignal-to-noise ratio and, thereby, higher spatial resolution. However, UHF MRI\nintroduces challenges such as transmit radiofrequency (RF) field (B1+)\ninhomogeneities, leading to uneven flip angles and image intensity anomalies.\nThese issues can significantly degrade imaging quality and its medical\napplications. This study addresses B1+ field homogeneity through a novel deep\nlearning-based strategy. Traditional methods like Magnitude Least Squares (MLS)\noptimization have been effective but are time-consuming and dependent on the\npatient's presence. Recent machine learning approaches, such as RF Shim\nPrediction by Iteratively Projected Ridge Regression and deep learning\nframeworks, have shown promise but face limitations like extensive training\ntimes and oversimplified architectures. We propose a two-step deep learning\nstrategy. First, we obtain the desired reference RF shimming weights from\nmulti-channel B1+ fields using random-initialized Adaptive Moment Estimation.\nThen, we employ Residual Networks (ResNets) to train a model that maps B1+\nfields to target RF shimming outputs. Our approach does not rely on\npre-calculated reference optimizations for the testing process and efficiently\nlearns residual functions. Comparative studies with traditional MLS\noptimization demonstrate our method's advantages in terms of speed and\naccuracy. The proposed strategy achieves a faster and more efficient RF\nshimming design, significantly improving imaging quality at UHF. This\nadvancement holds potential for broader applications in medical imaging and\ndiagnostics.\n", "date": "2024-08-21", "categories": []}
{"id": "2502.03505", "title": "Enhancing Free-hand 3D Photoacoustic and Ultrasound Reconstruction using\n  Deep Learning", "abstract": "  This study introduces a motion-based learning network with a global-local\nself-attention module (MoGLo-Net) to enhance 3D reconstruction in handheld\nphotoacoustic and ultrasound (PAUS) imaging. Standard PAUS imaging is often\nlimited by a narrow field of view and the inability to effectively visualize\ncomplex 3D structures. The 3D freehand technique, which aligns sequential 2D\nimages for 3D reconstruction, faces significant challenges in accurate motion\nestimation without relying on external positional sensors. MoGLo-Net addresses\nthese limitations through an innovative adaptation of the self-attention\nmechanism, which effectively exploits the critical regions, such as\nfully-developed speckle area or high-echogenic tissue area within successive\nultrasound images to accurately estimate motion parameters. This facilitates\nthe extraction of intricate features from individual frames. Additionally, we\ndesigned a patch-wise correlation operation to generate a correlation volume\nthat is highly correlated with the scanning motion. A custom loss function was\nalso developed to ensure robust learning with minimized bias, leveraging the\ncharacteristics of the motion parameters. Experimental evaluations demonstrated\nthat MoGLo-Net surpasses current state-of-the-art methods in both quantitative\nand qualitative performance metrics. Furthermore, we expanded the application\nof 3D reconstruction technology beyond simple B-mode ultrasound volumes to\nincorporate Doppler ultrasound and photoacoustic imaging, enabling 3D\nvisualization of vasculature. The source code for this study is publicly\navailable at: https://github.com/guhong3648/US3D\n", "date": "2025-02-05", "categories": []}
{"id": "2403.17006", "title": "Invertible Diffusion Models for Compressed Sensing", "abstract": "  While deep neural networks (NN) significantly advance image compressed\nsensing (CS) by improving reconstruction quality, the necessity of training\ncurrent CS NNs from scratch constrains their effectiveness and hampers rapid\ndeployment. Although recent methods utilize pre-trained diffusion models for\nimage reconstruction, they struggle with slow inference and restricted\nadaptability to CS. To tackle these challenges, this paper proposes Invertible\nDiffusion Models (IDM), a novel efficient, end-to-end diffusion-based CS\nmethod. IDM repurposes a large-scale diffusion sampling process as a\nreconstruction model, and fine-tunes it end-to-end to recover original images\ndirectly from CS measurements, moving beyond the traditional paradigm of\none-step noise estimation learning. To enable such memory-intensive end-to-end\nfine-tuning, we propose a novel two-level invertible design to transform both\n(1) multi-step sampling process and (2) noise estimation U-Net in each step\ninto invertible networks. As a result, most intermediate features are cleared\nduring training to reduce up to 93.8% GPU memory. In addition, we develop a set\nof lightweight modules to inject measurements into noise estimator to further\nfacilitate reconstruction. Experiments demonstrate that IDM outperforms\nexisting state-of-the-art CS networks by up to 2.64dB in PSNR. Compared to the\nrecent diffusion-based approach DDNM, our IDM achieves up to 10.09dB PSNR gain\nand 14.54 times faster inference. Code is available at\nhttps://github.com/Guaishou74851/IDM.\n", "date": "2024-03-25", "categories": []}
{"id": "2405.18731", "title": "VBIM-Net: Variational Born Iterative Network for Inverse Scattering\n  Problems", "abstract": "  Recently, studies have shown the potential of integrating field-type\niterative methods with deep learning (DL) techniques in solving inverse\nscattering problems (ISPs). In this article, we propose a novel Variational\nBorn Iterative Network, namely, VBIM-Net, to solve the full-wave ISPs with\nsignificantly improved structural rationality and inversion quality. The\nproposed VBIM-Net emulates the alternating updates of the total electric field\nand the contrast in the variational Born iterative method (VBIM) by multiple\nlayers of subnetworks. We embed the analytical calculation of the contrast\nvariation into each subnetwork, converting the scattered field residual into an\napproximate contrast variation and then enhancing it by a U-Net, thus avoiding\nthe requirement of matched measurement dimension and grid resolution as in\nexisting approaches. The total field and contrast of each layer's output is\nsupervised in the loss function of VBIM-Net, imposing soft physical constraints\non the variables in the subnetworks, which benefits the model's performance. In\naddition, we design a training scheme with extra noise to enhance the model's\nstability. Extensive numerical results on synthetic and experimental data both\nverify the inversion quality, generalization ability, and robustness of the\nproposed VBIM-Net. This work may provide some new inspiration for the design of\nefficient field-type DL schemes.\n", "date": "2024-05-29", "categories": []}
{"id": "2502.01102", "title": "Towards Robust and Generalizable Lensless Imaging with Modular Learned\n  Reconstruction", "abstract": "  Lensless cameras disregard the conventional design that imaging should mimic\nthe human eye. This is done by replacing the lens with a thin mask, and moving\nimage formation to the digital post-processing. State-of-the-art lensless\nimaging techniques use learned approaches that combine physical modeling and\nneural networks. However, these approaches make simplifying modeling\nassumptions for ease of calibration and computation. Moreover, the\ngeneralizability of learned approaches to lensless measurements of new masks\nhas not been studied. To this end, we utilize a modular learned reconstruction\nin which a key component is a pre-processor prior to image recovery. We\ntheoretically demonstrate the pre-processor's necessity for standard image\nrecovery techniques (Wiener filtering and iterative algorithms), and through\nextensive experiments show its effectiveness for multiple lensless imaging\napproaches and across datasets of different mask types (amplitude and phase).\nWe also perform the first generalization benchmark across mask types to\nevaluate how well reconstructions trained with one system generalize to others.\nOur modular reconstruction enables us to use pre-trained components and\ntransfer learning on new systems to cut down weeks of tedious measurements and\ntraining. As part of our work, we open-source four datasets, and software for\nmeasuring datasets and for training our modular reconstruction.\n", "date": "2025-02-03", "categories": []}
{"id": "2502.01338", "title": "PtyGenography: using generative models for regularization of the phase\n  retrieval problem", "abstract": "  In phase retrieval and similar inverse problems, the stability of solutions\nacross different noise levels is crucial for applications. One approach to\npromote it is using signal priors in a form of a generative model as a\nregularization, at the expense of introducing a bias in the reconstruction. In\nthis paper, we explore and compare the reconstruction properties of classical\nand generative inverse problem formulations. We propose a new unified\nreconstruction approach that mitigates overfitting to the generative model for\nvarying noise levels.\n", "date": "2025-02-03", "categories": []}
{"id": "2501.19129", "title": "RGB-Event ISP: The Dataset and Benchmark", "abstract": "  Event-guided imaging has received significant attention due to its potential\nto revolutionize instant imaging systems. However, the prior methods primarily\nfocus on enhancing RGB images in a post-processing manner, neglecting the\nchallenges of image signal processor (ISP) dealing with event sensor and the\nbenefits events provide for reforming the ISP process. To achieve this, we\nconduct the first research on event-guided ISP. First, we present a new\nevent-RAW paired dataset, collected with a novel but still confidential sensor\nthat records pixel-level aligned events and RAW images. This dataset includes\n3373 RAW images with 2248 x 3264 resolution and their corresponding events,\nspanning 24 scenes with 3 exposure modes and 3 lenses. Second, we propose a\nconventional ISP pipeline to generate good RGB frames as reference. This\nconventional ISP pipleline performs basic ISP operations, e.g.demosaicing,\nwhite balancing, denoising and color space transforming, with a ColorChecker as\nreference. Third, we classify the existing learnable ISP methods into 3\nclasses, and select multiple methods to train and evaluate on our new dataset.\nLastly, since there is no prior work for reference, we propose a simple\nevent-guided ISP method and test it on our dataset. We further put forward key\ntechnical challenges and future directions in RGB-Event ISP. In summary, to the\nbest of our knowledge, this is the very first research focusing on event-guided\nISP, and we hope it will inspire the community. The code and dataset are\navailable at: https://github.com/yunfanLu/RGB-Event-ISP.\n", "date": "2025-01-31", "categories": []}
{"id": "2501.17898", "title": "Distilling Knowledge for Designing Computational Imaging Systems", "abstract": "  Designing the physical encoder is crucial for accurate image reconstruction\nin computational imaging (CI) systems. Currently, these systems are designed\nvia end-to-end (E2E) optimization, where the encoder is modeled as a neural\nnetwork layer and is jointly optimized with the decoder. However, the\nperformance of E2E optimization is significantly reduced by the physical\nconstraints imposed on the encoder. Also, since the E2E learns the parameters\nof the encoder by backpropagating the reconstruction error, it does not promote\noptimal intermediate outputs and suffers from gradient vanishing. To address\nthese limitations, we reinterpret the concept of knowledge distillation (KD)\nfor designing a physically constrained CI system by transferring the knowledge\nof a pretrained, less-constrained CI system. Our approach involves three steps:\n(1) Given the original CI system (student), a teacher system is created by\nrelaxing the constraints on the student's encoder. (2) The teacher is optimized\nto solve a less-constrained version of the student's problem. (3) The teacher\nguides the training of the student through two proposed knowledge transfer\nfunctions, targeting both the encoder and the decoder feature space. The\nproposed method can be employed to any imaging modality since the relaxation\nscheme and the loss functions can be adapted according to the physical\nacquisition and the employed decoder. This approach was validated on three\nrepresentative CI modalities: magnetic resonance, single-pixel, and compressive\nspectral imaging. Simulations show that a teacher system with an encoder that\nhas a structure similar to that of the student encoder provides effective\nguidance. Our approach achieves significantly improved reconstruction\nperformance and encoder design, outperforming both E2E optimization and\ntraditional non-data-driven encoder designs.\n", "date": "2025-01-29", "categories": []}
{"id": "2501.18219", "title": "Revisiting $\\Psi$DONet: microlocally inspired filters for\n  incomplete-data tomographic reconstructions", "abstract": "  In this paper, we revisit a supervised learning approach based on unrolling,\nknown as $\\Psi$DONet, by providing a deeper microlocal interpretation for its\ntheoretical analysis, and extending its study to the case of sparse-angle\ntomography. Furthermore, we refine the implementation of the original\n$\\Psi$DONet considering special filters whose structure is specifically\ninspired by the streak artifact singularities characterizing tomographic\nreconstructions from incomplete data. This allows to considerably lower the\nnumber of (learnable) parameters while preserving (or even slightly improving)\nthe same quality for the reconstructions from limited-angle data and providing\na proof-of-concept for the case of sparse-angle tomographic data.\n", "date": "2025-01-30", "categories": []}
{"id": "2501.18517", "title": "Integrating Spatial and Frequency Information for Under-Display Camera\n  Image Restoration", "abstract": "  Under-Display Camera (UDC) houses a digital camera lens under a display\npanel. However, UDC introduces complex degradations such as noise, blur,\ndecrease in transmittance, and flare. Despite the remarkable progress, previous\nresearch on UDC mainly focuses on eliminating diffraction in the spatial domain\nand rarely explores its potential in the frequency domain. It is essential to\nconsider both the spatial and frequency domains effectively. For example,\ndegradations, such as noise and blur, can be addressed by local information\n(e.g., CNN kernels in the spatial domain). At the same time, tackling flares\nmay require leveraging global information (e.g., the frequency domain). In this\npaper, we revisit the UDC degradations in the Fourier space and figure out\nintrinsic frequency priors that imply the presence of the flares. Based on this\nobservation, we propose a novel multi-level DNN architecture called SFIM. It\nefficiently restores UDC-distorted images by integrating local and global (the\ncollective contribution of all points in the image) information. The\narchitecture exploits CNNs to capture local information and FFT-based models to\ncapture global information. SFIM comprises a spatial domain block (SDB), a\nFrequency Domain Block (FDB), and an Attention-based Multi-level Integration\nBlock (AMIB). Specifically, SDB focuses more on detailed textures such as noise\nand blur, FDB emphasizes irregular texture loss in extensive areas such as\nflare, and AMIB enables effective cross-domain interaction. SFIM's superior\nperformance over state-of-the-art approaches is demonstrated through rigorous\nquantitative and qualitative assessments across three UDC benchmarks.\n", "date": "2025-01-30", "categories": []}
{"id": "2501.18595", "title": "ROSA: Reconstructing Object Shape and Appearance Textures by Adaptive\n  Detail Transfer", "abstract": "  Reconstructing an object's shape and appearance in terms of a mesh textured\nby a spatially-varying bidirectional reflectance distribution function (SVBRDF)\nfrom a limited set of images captured under collocated light is an ill-posed\nproblem. Previous state-of-the-art approaches either aim to reconstruct the\nappearance directly on the geometry or additionally use texture normals as part\nof the appearance features. However, this requires detailed but inefficiently\nlarge meshes, that would have to be simplified in a post-processing step, or\nsuffers from well-known limitations of normal maps such as missing shadows or\nincorrect silhouettes. Another limiting factor is the fixed and typically low\nresolution of the texture estimation resulting in loss of important surface\ndetails. To overcome these problems, we present ROSA, an inverse rendering\nmethod that directly optimizes mesh geometry with spatially adaptive mesh\nresolution solely based on the image data. In particular, we refine the mesh\nand locally condition the surface smoothness based on the estimated normal\ntexture and mesh curvature. In addition, we enable the reconstruction of fine\nappearance details in high-resolution textures through a pioneering tile-based\nmethod that operates on a single pre-trained decoder network but is not limited\nby the network output resolution.\n", "date": "2025-01-30", "categories": []}
{"id": "2501.17211", "title": "MR imaging in the low-field: Leveraging the power of machine learning", "abstract": "  Recent innovations in Magnetic Resonance Imaging (MRI) hardware and software\nhave reignited interest in low-field ($<1\\,\\mathrm{T}$) and ultra-low-field MRI\n($<0.1\\,\\mathrm{T}$). These technologies offer advantages such as lower power\nconsumption, reduced specific absorption rate, reduced field-inhomogeneities,\nand cost-effectiveness, presenting a promising alternative for resource-limited\nand point-of-care settings. However, low-field MRI faces inherent challenges\nlike reduced signal-to-noise ratio and therefore, potentially lower spatial\nresolution or longer scan times.\n  This chapter examines the challenges and opportunities of low-field and\nultra-low-field MRI, with a focus on the role of machine learning (ML) in\novercoming these limitations. We provide an overview of deep neural networks\nand their application in enhancing low-field and ultra-low-field MRI\nperformance. Specific ML-based solutions, including advanced image\nreconstruction, denoising, and super-resolution algorithms, are discussed. The\nchapter concludes by exploring how integrating ML with low-field MRI could\nexpand its clinical applications and improve accessibility, potentially\nrevolutionizing its use in diverse healthcare settings.\n", "date": "2025-01-28", "categories": []}
{"id": "2501.17152", "title": "Three-Dimensional Diffusion-Weighted Multi-Slab MRI With Slice Profile\n  Compensation Using Deep Energy Model", "abstract": "  Three-dimensional (3D) multi-slab acquisition is a technique frequently\nemployed in high-resolution diffusion-weighted MRI in order to achieve the best\nsignal-to-noise ratio (SNR) efficiency. However, this technique is limited by\nslab boundary artifacts that cause intensity fluctuations and aliasing between\nslabs which reduces the accuracy of anatomical imaging. Addressing this issue\nis crucial for advancing diffusion MRI quality and making high-resolution\nimaging more feasible for clinical and research applications. In this work, we\npropose a regularized slab profile encoding (PEN) method within a Plug-and-Play\nADMM framework, incorporating multi-scale energy (MuSE) regularization to\neffectively improve the slab combined reconstruction. Experimental results\ndemonstrate that the proposed method significantly improves image quality\ncompared to non-regularized and TV-regularized PEN approaches. The regularized\nPEN framework provides a more robust and efficient solution for high-resolution\n3D diffusion MRI, potentially enabling clearer, more reliable anatomical\nimaging across various applications.\n", "date": "2025-01-28", "categories": []}
{"id": "2501.13961", "title": "A Fast, Scalable, and Robust Deep Learning-based Iterative\n  Reconstruction Framework for Accelerated Industrial Cone-beam X-ray Computed\n  Tomography", "abstract": "  Cone-beam X-ray Computed Tomography (XCT) with large detectors and\ncorresponding large-scale 3D reconstruction plays a pivotal role in\nmicron-scale characterization of materials and parts across various industries.\nIn this work, we present a novel deep neural network-based iterative algorithm\nthat integrates an artifact reduction-trained CNN as a prior model with\nautomated regularization parameter selection, tailored for large-scale\nindustrial cone-beam XCT data. Our method achieves high-quality 3D\nreconstructions even for extremely dense thick metal parts - which\ntraditionally pose challenges to industrial CT images - in just a few\niterations. Furthermore, we show the generalizability of our approach to\nout-of-distribution scans obtained under diverse scanning conditions. Our\nmethod effectively handles significant noise and streak artifacts, surpassing\nstate-of-the-art supervised learning methods trained on the same data.\n", "date": "2025-01-21", "categories": []}
{"id": "2501.14709", "title": "Enhanced Confocal Laser Scanning Microscopy with Adaptive Physics\n  Informed Deep Autoencoders", "abstract": "  We present a physics-informed deep learning framework to address common\nlimitations in Confocal Laser Scanning Microscopy (CLSM), such as diffraction\nlimited resolution, noise, and undersampling due to low laser power conditions.\nThe optical system's point spread function (PSF) and common CLSM image\ndegradation mechanisms namely photon shot noise, dark current noise, motion\nblur, speckle noise, and undersampling were modeled and were directly included\ninto model architecture. The model reconstructs high fidelity images from\nheavily noisy inputs by using convolutional and transposed convolutional\nlayers. Following the advances in compressed sensing, our approach\nsignificantly reduces data acquisition requirements without compromising image\nresolution. The proposed method was extensively evaluated on simulated CLSM\nimages of diverse structures, including lipid droplets, neuronal networks, and\nfibrillar systems. Comparisons with traditional deconvolution algorithms such\nas Richardson-Lucy (RL), non-negative least squares (NNLS), and other methods\nlike Total Variation (TV) regularization, Wiener filtering, and Wavelet\ndenoising demonstrate the superiority of the network in restoring fine\nstructural details with high fidelity. Assessment metrics like Structural\nSimilarity Index (SSIM) and Peak Signal to Noise Ratio (PSNR), underlines that\nthe AdaptivePhysicsAutoencoder achieved robust image enhancement across diverse\nCLSM conditions, helping faster acquisition, reduced photodamage, and reliable\nperformance in low light and sparse sampling scenarios holding promise for\napplications in live cell imaging, dynamic biological studies, and high\nthroughput material characterization.\n", "date": "2025-01-24", "categories": []}
{"id": "2403.05245", "title": "Noise Level Adaptive Diffusion Model for Robust Reconstruction of\n  Accelerated MRI", "abstract": "  In general, diffusion model-based MRI reconstruction methods incrementally\nremove artificially added noise while imposing data consistency to reconstruct\nthe underlying images. However, real-world MRI acquisitions already contain\ninherent noise due to thermal fluctuations. This phenomenon is particularly\nnotable when using ultra-fast, high-resolution imaging sequences for advanced\nresearch, or using low-field systems favored by low- and middle-income\ncountries. These common scenarios can lead to sub-optimal performance or\ncomplete failure of existing diffusion model-based reconstruction techniques.\nSpecifically, as the artificially added noise is gradually removed, the\ninherent MRI noise becomes increasingly pronounced, making the actual noise\nlevel inconsistent with the predefined denoising schedule and consequently\ninaccurate image reconstruction. To tackle this problem, we propose a posterior\nsampling strategy with a novel NoIse Level Adaptive Data Consistency (Nila-DC)\noperation. Extensive experiments are conducted on two public datasets and an\nin-house clinical dataset with field strength ranging from 0.3T to 3T, showing\nthat our method surpasses the state-of-the-art MRI reconstruction methods, and\nis highly robust against various noise levels. The code for Nila is available\nat https://github.com/Solor-pikachu/Nila.\n", "date": "2024-03-08", "categories": []}
{"id": "2501.12705", "title": "The Marginal Importance of Distortions and Alignment in CASSI systems", "abstract": "  This paper introduces a differentiable ray-tracing based model that\nincorporates aberrations and distortions to render realistic coded\nhyperspectral acquisitions using Coded-Aperture Spectral Snapshot Imagers\n(CASSI). CASSI systems can now be optimized in order to fulfill simultaneously\nseveral optical design constraints as well as processing constraints. Four\ncomparable CASSI systems with varying degree of optical aberrations have been\ndesigned and modeled. The resulting rendered hyperspectral acquisitions from\neach of these systems are combined with five state-of-the-art hyperspectral\ncube reconstruction processes. These reconstruction processes encompass a\nmapping function created from each system's propagation model to account for\ndistortions and aberrations during the reconstruction process. Our analyses\nshow that if properly modeled, the effects of geometric distortions of the\nsystem and misalignments of the dispersive elements have a marginal impact on\nthe overall quality of the reconstructed hyperspectral data cubes. Therefore,\nrelaxing traditional constraints on measurement conformity and fidelity to the\nscene enables the development of novel imaging instruments, guided by\nperformance metrics applied to the design or the processing of acquisitions. By\nproviding a complete framework for design, simulation and evaluation, this work\ncontributes to the optimization and exploration of new CASSI systems, and more\ngenerally to the computational imaging community.\n", "date": "2025-01-22", "categories": []}
{"id": "2501.12832", "title": "FDG-Diff: Frequency-Domain-Guided Diffusion Framework for Compressed\n  Hazy Image Restoration", "abstract": "  In this study, we reveal that the interaction between haze degradation and\nJPEG compression introduces complex joint loss effects, which significantly\ncomplicate image restoration. Existing dehazing models often neglect\ncompression effects, which limits their effectiveness in practical\napplications. To address these challenges, we introduce three key\ncontributions. First, we design FDG-Diff, a novel frequency-domain-guided\ndehazing framework that improves JPEG image restoration by leveraging\nfrequency-domain information. Second, we introduce the High-Frequency\nCompensation Module (HFCM), which enhances spatial-domain detail restoration by\nincorporating frequency-domain augmentation techniques into a diffusion-based\nrestoration framework. Lastly, the introduction of the Degradation-Aware\nDenoising Timestep Predictor (DADTP) module further enhances restoration\nquality by enabling adaptive region-specific restoration, effectively\naddressing regional degradation inconsistencies in compressed hazy images.\nExperimental results across multiple compressed dehazing datasets demonstrate\nthat our method consistently outperforms the latest state-of-the-art\napproaches. Code be available at https://github.com/SYSUzrc/FDG-Diff.\n", "date": "2025-01-22", "categories": []}
{"id": "2409.16766", "title": "Let There Be Light: Robust Lensless Imaging Under External Illumination\n  With Deep Learning", "abstract": "  Lensless cameras relax the design constraints of traditional cameras by\nshifting image formation from analog optics to digital post-processing. While\nnew camera designs and applications can be enabled, lensless imaging is very\nsensitive to unwanted interference (other sources, noise, etc.). In this work,\nwe address a prevalent noise source that has not been studied for lensless\nimaging: external illumination e.g. from ambient and direct lighting. Being\nrobust to a variety of lighting conditions would increase the practicality and\nadoption of lensless imaging. To this end, we propose multiple recovery\napproaches that account for external illumination by incorporating its estimate\ninto the image recovery process. At the core is a physics-based reconstruction\nthat combines learnable image recovery and denoisers, all of whose parameters\nare trained using experimentally gathered data. Compared to standard\nreconstruction methods, our approach yields significant qualitative and\nquantitative improvements. We open-source our implementations and a 25K dataset\nof measurements under multiple lighting conditions.\n", "date": "2024-09-25", "categories": []}
{"id": "2501.11586", "title": "Compressibility Analysis for the differentiable shift-variant Filtered\n  Backprojection Model", "abstract": "  The differentiable shift-variant filtered backprojection (FBP) model enables\nthe reconstruction of cone-beam computed tomography (CBCT) data for any\nnon-circular trajectories. This method employs deep learning technique to\nestimate the redundancy weights required for reconstruction, given knowledge of\nthe specific trajectory at optimization time. However, computing the redundancy\nweight for each projection remains computationally intensive. This paper\npresents a novel approach to compress and optimize the differentiable\nshift-variant FBP model based on Principal Component Analysis (PCA). We apply\nPCA to the redundancy weights learned from sinusoidal trajectory projection\ndata, revealing significant parameter redundancy in the original model. By\nintegrating PCA directly into the differentiable shift-variant FBP\nreconstruction pipeline, we develop a method that decomposes the redundancy\nweight layer parameters into a trainable eigenvector matrix, compressed\nweights, and a mean vector. This innovative technique achieves a remarkable\n97.25% reduction in trainable parameters without compromising reconstruction\naccuracy. As a result, our algorithm significantly decreases the complexity of\nthe differentiable shift-variant FBP model and greatly improves training speed.\nThese improvements make the model substantially more practical for real-world\napplications.\n", "date": "2025-01-20", "categories": []}
{"id": "2308.12864", "title": "Auto-weighted Bayesian Physics-Informed Neural Networks and robust\n  estimations for multitask inverse problems in pore-scale imaging of\n  dissolution", "abstract": "  In this article, we present a novel data assimilation strategy in pore-scale\nimaging and demonstrate that this makes it possible to robustly address\nreactive inverse problems incorporating Uncertainty Quantification (UQ).\nPore-scale modeling of reactive flow offers a valuable opportunity to\ninvestigate the evolution of macro-scale properties subject to dynamic\nprocesses. Yet, they suffer from imaging limitations arising from the\nassociated X-ray microtomography (X-ray microCT) process, which induces\ndiscrepancies in the properties estimates. Assessment of the kinetic parameters\nalso raises challenges, as reactive coefficients are critical parameters that\ncan cover a wide range of values. We account for these two issues and ensure\nreliable calibration of pore-scale modeling, based on dynamical microCT images,\nby integrating uncertainty quantification in the workflow.\n  The present method is based on a multitasking formulation of reactive inverse\nproblems combining data-driven and physics-informed techniques in calcite\ndissolution. This allows quantifying morphological uncertainties on the\nporosity field and estimating reactive parameter ranges through prescribed PDE\nmodels with a latent concentration field and dynamical microCT. The data\nassimilation strategy relies on sequential reinforcement incorporating\nsuccessively additional PDE constraints. We guarantee robust and unbiased\nuncertainty quantification by straightforward adaptive weighting of Bayesian\nPhysics-Informed Neural Networks (BPINNs), ensuring reliable micro-porosity\nchanges during geochemical transformations. We demonstrate successful Bayesian\nInference in 1D+Time and 2D+Time calcite dissolution based on synthetic microCT\nimages with meaningful posterior distribution on the reactive parameters and\ndimensionless numbers.\n", "date": "2023-08-24", "categories": []}
{"id": "2411.18967", "title": "Deep Plug-and-Play HIO Approach for Phase Retrieval", "abstract": "  In the phase retrieval problem, the aim is the recovery of an unknown image\nfrom intensity-only measurements such as Fourier intensity. Although there are\nseveral solution approaches, solving this problem is challenging due to its\nnonlinear and ill-posed nature. Recently, learning-based approaches have\nemerged as powerful alternatives to the analytical methods for several inverse\nproblems. In the context of phase retrieval, a novel plug-and-play approach\nthat exploits learning-based prior and efficient update steps has been\npresented at the Computational Optical Sensing and Imaging topical meeting,\nwith demonstrated state-of-the-art performance. The key idea was to incorporate\nlearning-based prior to the Gerchberg-Saxton type algorithms through\nplug-and-play regularization. In this paper, we present the mathematical\ndevelopment of the method including the derivation of its analytical update\nsteps based on half-quadratic splitting and comparatively evaluate its\nperformance through extensive simulations on a large test dataset. The results\nshow the effectiveness of the method in terms of both image quality,\ncomputational efficiency, and robustness to initialization and noise.\n", "date": "2024-11-28", "categories": []}
{"id": "2501.05777", "title": "StructSR: Refuse Spurious Details in Real-World Image Super-Resolution", "abstract": "  Diffusion-based models have shown great promise in real-world image\nsuper-resolution (Real-ISR), but often generate content with structural errors\nand spurious texture details due to the empirical priors and illusions of these\nmodels. To address this issue, we introduce StructSR, a simple, effective, and\nplug-and-play method that enhances structural fidelity and suppresses spurious\ndetails for diffusion-based Real-ISR. StructSR operates without the need for\nadditional fine-tuning, external model priors, or high-level semantic\nknowledge. At its core is the Structure-Aware Screening (SAS) mechanism, which\nidentifies the image with the highest structural similarity to the\nlow-resolution (LR) input in the early inference stage, allowing us to leverage\nit as a historical structure knowledge to suppress the generation of spurious\ndetails. By intervening in the diffusion inference process, StructSR seamlessly\nintegrates with existing diffusion-based Real-ISR models. Our experimental\nresults demonstrate that StructSR significantly improves the fidelity of\nstructure and texture, improving the PSNR and SSIM metrics by an average of\n5.27% and 9.36% on a synthetic dataset (DIV2K-Val) and 4.13% and 8.64% on two\nreal-world datasets (RealSR and DRealSR) when integrated with four\nstate-of-the-art diffusion-based Real-ISR methods.\n", "date": "2025-01-10", "categories": []}
{"id": "2501.09259", "title": "OpticFusion: Multi-Modal Neural Implicit 3D Reconstruction of\n  Microstructures by Fusing White Light Interferometry and Optical Microscopy", "abstract": "  White Light Interferometry (WLI) is a precise optical tool for measuring the\n3D topography of microstructures. However, conventional WLI cannot capture the\nnatural color of a sample's surface, which is essential for many microscale\nresearch applications that require both 3D geometry and color information.\nPrevious methods have attempted to overcome this limitation by modifying WLI\nhardware and analysis software, but these solutions are often costly. In this\nwork, we address this challenge from a computer vision multi-modal\nreconstruction perspective for the first time. We introduce OpticFusion, a\nnovel approach that uses an additional digital optical microscope (OM) to\nachieve 3D reconstruction with natural color textures using multi-view WLI and\nOM images. Our method employs a two-step data association process to obtain the\nposes of WLI and OM data. By leveraging the neural implicit representation, we\nfuse multi-modal data and apply color decomposition technology to extract the\nsample's natural color. Tested on our multi-modal dataset of various microscale\nsamples, OpticFusion achieves detailed 3D reconstructions with color textures.\nOur method provides an effective tool for practical applications across\nnumerous microscale research fields. The source code and our real-world dataset\nare available at https://github.com/zju3dv/OpticFusion.\n", "date": "2025-01-16", "categories": []}
{"id": "2501.09403", "title": "PISCO: Self-Supervised k-Space Regularization for Improved Neural\n  Implicit k-Space Representations of Dynamic MRI", "abstract": "  Neural implicit k-space representations (NIK) have shown promising results\nfor dynamic magnetic resonance imaging (MRI) at high temporal resolutions. Yet,\nreducing acquisition time, and thereby available training data, results in\nsevere performance drops due to overfitting. To address this, we introduce a\nnovel self-supervised k-space loss function $\\mathcal{L}_\\mathrm{PISCO}$,\napplicable for regularization of NIK-based reconstructions. The proposed loss\nfunction is based on the concept of parallel imaging-inspired self-consistency\n(PISCO), enforcing a consistent global k-space neighborhood relationship\nwithout requiring additional data. Quantitative and qualitative evaluations on\nstatic and dynamic MR reconstructions show that integrating PISCO significantly\nimproves NIK representations. Particularly for high acceleration factors\n(R$\\geq$54), NIK with PISCO achieves superior spatio-temporal reconstruction\nquality compared to state-of-the-art methods. Furthermore, an extensive\nanalysis of the loss assumptions and stability shows PISCO's potential as\nversatile self-supervised k-space loss function for further applications and\narchitectures. Code is available at:\nhttps://github.com/compai-lab/2025-pisco-spieker\n", "date": "2025-01-16", "categories": []}
{"id": "2403.00394", "title": "List-Mode PET Image Reconstruction Using Dykstra-Like Splitting", "abstract": "  Convergence of the block iterative method in image reconstruction for\npositron emission tomography (PET) requires careful control of relaxation\nparameters, which is a challenging task. The automatic determination of\nrelaxation parameters for list-mode reconstructions also remains challenging.\nTherefore, a different approach would be desirable. In this study, we propose a\nlist-mode maximum likelihood Dykstra-like splitting PET reconstruction\n(LM-MLDS). LM-MLDS converges the list-mode block iterative method by adding the\ndistance from an initial image as a penalty term into an objective function.\nLM-MLDS takes a two-step approach because its performance depends on the\nquality of the initial image. The first step uses a uniform image as the\ninitial image, and then the second step uses a reconstructed image after one\nmain iteration as the initial image. In a simulation study, LM-MLDS provided a\nbetter tradeoff curve between noise and contrast than the other methods. In a\nclinical study, LM-MLDS removed the false hotspots at the edge of the axial\nfield of view and improved the image quality of slices covering the top of the\nhead to the cerebellum. List-mode proximal splitting reconstruction is useful\nnot only for optimizing nondifferential functions but also for converging block\niterative methods without controlling relaxation parameters.\n", "date": "2024-03-01", "categories": []}
{"id": "2501.04665", "title": "HyFusion: Enhanced Reception Field Transformer for Hyperspectral Image\n  Fusion", "abstract": "  Hyperspectral image (HSI) fusion addresses the challenge of reconstructing\nHigh-Resolution HSIs (HR-HSIs) from High-Resolution Multispectral images\n(HR-MSIs) and Low-Resolution HSIs (LR-HSIs), a critical task given the high\ncosts and hardware limitations associated with acquiring high-quality HSIs.\nWhile existing methods leverage spatial and spectral relationships, they often\nsuffer from limited receptive fields and insufficient feature utilization,\nleading to suboptimal performance. Furthermore, the scarcity of high-quality\nHSI data highlights the importance of efficient data utilization to maximize\nreconstruction quality. To address these issues, we propose HyFusion, a novel\nDual-Coupled Network (DCN) framework designed to enhance cross-domain feature\nextraction and enable effective feature map reusing. The framework first\nprocesses HR-MSI and LR-HSI inputs through specialized subnetworks that\nmutually enhance each other during feature extraction, preserving complementary\nspatial and spectral details. At its core, HyFusion utilizes an Enhanced\nReception Field Block (ERFB), which combines shifting-window attention and\ndense connections to expand the receptive field, effectively capturing\nlong-range dependencies while minimizing information loss. Extensive\nexperiments demonstrate that HyFusion achieves state-of-the-art performance in\nHR-MSI/LR-HSI fusion, significantly improving reconstruction quality while\nmaintaining a compact model size and computational efficiency. By integrating\nenhanced receptive fields and feature map reusing into a coupled network\narchitecture, HyFusion provides a practical and effective solution for HSI\nfusion in resource-constrained scenarios, setting a new benchmark in\nhyperspectral imaging. Our code will be publicly available.\n", "date": "2025-01-08", "categories": []}
{"id": "2412.15890", "title": "NeuroPump: Simultaneous Geometric and Color Rectification for Underwater\n  Images", "abstract": "  Underwater image restoration aims to remove geometric and color distortions\ndue to water refraction, absorption and scattering. Previous studies focus on\nrestoring either color or the geometry, but to our best knowledge, not both.\nHowever, in practice it may be cumbersome to address the two rectifications\none-by-one. In this paper, we propose NeuroPump, a self-supervised method to\nsimultaneously optimize and rectify underwater geometry and color as if water\nwere pumped out. The key idea is to explicitly model refraction, absorption and\nscattering in Neural Radiance Field (NeRF) pipeline, such that it not only\nperforms simultaneous geometric and color rectification, but also enables to\nsynthesize novel views and optical effects by controlling the decoupled\nparameters. In addition, to address issue of lack of real paired ground truth\nimages, we propose an underwater 360 benchmark dataset that has real paired\n(i.e., with and without water) images. Our method clearly outperforms other\nbaselines both quantitatively and qualitatively. Our project page is available\nat: https://ygswu.github.io/NeuroPump.github.io/.\n", "date": "2024-12-20", "categories": []}
{"id": "2501.05244", "title": "Optimized Sampling for Non-Line-of-Sight Imaging Using Modified Fast\n  Fourier Transforms", "abstract": "  Non-line-of-Sight (NLOS) imaging systems collect light at a diffuse relay\nsurface and input this measurement into computational algorithms that output a\n3D volumetric reconstruction. These algorithms utilize the Fast Fourier\nTransform (FFT) to accelerate the reconstruction process but require both input\nand output to be sampled spatially with uniform grids. However, the geometry of\nNLOS imaging inherently results in non-uniform sampling on the relay surface\nwhen using multi-pixel detector arrays, even though such arrays significantly\nreduce acquisition times. Furthermore, using these arrays increases the data\nrate required for sensor readout, posing challenges for real-world deployment.\nIn this work, we utilize the phasor field framework to demonstrate that\nexisting NLOS imaging setups typically oversample the relay surface spatially,\nexplaining why the measurement can be compressed without significantly\nsacrificing reconstruction quality. This enables us to utilize the Non-Uniform\nFast Fourier Transform (NUFFT) to reconstruct from sparse measurements acquired\nfrom irregularly sampled relay surfaces of arbitrary shapes. Furthermore, we\nutilize the NUFFT to reconstruct at arbitrary locations in the hidden volume,\nensuring flexible sampling schemes for both the input and output. Finally, we\nutilize the Scaled Fast Fourier Transform (SFFT) to reconstruct larger volumes\nwithout increasing the number of samples stored in memory. All algorithms\nintroduced in this paper preserve the computational complexity of FFT-based\nmethods, ensuring scalability for practical NLOS imaging applications.\n", "date": "2025-01-09", "categories": []}
{"id": "2501.05769", "title": "Conditional Diffusion Model for Electrical Impedance Tomography", "abstract": "  Electrical impedance tomography (EIT) is a non-invasive imaging technique,\nwhich has been widely used in the fields of industrial inspection, medical\nmonitoring and tactile sensing. However, due to the inherent non-linearity and\nill-conditioned nature of the EIT inverse problem, the reconstructed image is\nhighly sensitive to the measured data, and random noise artifacts often appear\nin the reconstructed image, which greatly limits the application of EIT. To\naddress this issue, a conditional diffusion model with voltage consistency\n(CDMVC) is proposed in this study. The method consists of a pre-imaging module,\na conditional diffusion model for reconstruction, a forward voltage constraint\nnetwork and a scheme of voltage consistency constraint during sampling process.\nThe pre-imaging module is employed to generate the initial reconstruction. This\nserves as a condition for training the conditional diffusion model. Finally,\nbased on the forward voltage constraint network, a voltage consistency\nconstraint is implemented in the sampling phase to incorporate forward\ninformation of EIT, thereby enhancing imaging quality. A more complete dataset,\nincluding both common and complex concave shapes, is generated. The proposed\nmethod is validated using both simulation and physical experiments.\nExperimental results demonstrate that our method can significantly improves the\nquality of reconstructed images. In addition, experimental results also\ndemonstrate that our method has good robustness and generalization performance.\n", "date": "2025-01-10", "categories": []}
{"id": "2501.04308", "title": "FSC-loss: A Frequency-domain Structure Consistency Learning Approach for\n  Signal Data Recovery and Reconstruction", "abstract": "  A core challenge for signal data recovery is to model the distribution of\nsignal matrix (SM) data based on measured low-quality data in biomedical\nengineering of magnetic particle imaging (MPI). For acquiring the\nhigh-resolution (high-quality) SM, the number of meticulous measurements at\nnumerous positions in the field-of-view proves time-consuming (measurement of a\n37x37x37 SM takes about 32 hours). To improve reconstructed signal quality and\nshorten SM measurement time, existing methods explore to generating\nhigh-resolution SM based on time-saving measured low-resolution SM (a 9x9x9 SM\njust takes about 0.5 hours). However, previous methods show poor performance\nfor high-frequency signal recovery in SM. To achieve a high-resolution SM\nrecovery and shorten its acquisition time, we propose a frequency-domain\nstructure consistency loss function and data component embedding strategy to\nmodel global and local structural information of SM. We adopt a\ntransformer-based network to evaluate this function and the strategy. We\nevaluate our methods and state-of-the-art (SOTA) methods on the two simulation\ndatasets and four public measured SMs in Open MPI Data. The results show that\nour method outperforms the SOTA methods in high-frequency structural signal\nrecovery. Additionally, our method can recover a high-resolution SM with clear\nhigh-frequency structure based on a down-sampling factor of 16 less than 15\nseconds, which accelerates the acquisition time over 60 times faster than the\nmeasurement-based HR SM with the minimum error (nRMSE=0.041). Moreover, our\nmethod is applied in our three in-house MPI systems, and boost their\nperformance for signal reconstruction.\n", "date": "2025-01-08", "categories": []}
{"id": "2405.03732", "title": "Deep Learning-based Accelerated MR Cholangiopancreatography without\n  Fully-sampled Data", "abstract": "  The purpose of this study was to accelerate MR cholangiopancreatography\n(MRCP) acquisitions using deep learning-based (DL) reconstruction at 3T and\n0.55T. A total of 35 healthy volunteers underwent conventional two-fold\naccelerated MRCP scans at field strengths of 3T and 0.55T. We trained DL\nreconstructions using two different training strategies, supervised (SV) and\nself-supervised (SSV), with retrospectively six-fold undersampled data obtained\nat 3T. We then evaluated the DL reconstructions against standard techniques,\nparallel imaging (PI) and compressed sensing (CS), focusing on peak\nsignal-to-noise ratio (PSNR) and structural similarity (SSIM) as metrics. We\nalso tested DL reconstructions with prospectively accelerated acquisitions and\nevaluated their robustness when changing fields strengths from 3T to 0.55T. DL\nreconstructions demonstrated a reduction in average acquisition time from\n599/542 to 255/180 seconds for MRCP at 3T/0.55T. In both retrospective and\nprospective undersampling, PSNR and SSIM of DL reconstructions were higher than\nthose of PI and CS. At the same time, DL reconstructions preserved the image\nquality of undersampled data, including sharpness and the visibility of\nhepatobiliary ducts. In addition, both DL approaches produced high-quality\nreconstructions at 0.55T. In summary, DL reconstructions trained for highly\naccelerated MRCP enabled a reduction in acquisition time by a factor of 2.4/3.0\nat 3T/0.55T while maintaining the image quality of conventional acquisitions.\n", "date": "2024-05-06", "categories": []}
{"id": "2501.03825", "title": "Deep Sylvester Posterior Inference for Adaptive Compressed Sensing in\n  Ultrasound Imaging", "abstract": "  Ultrasound images are commonly formed by sequential acquisition of\nbeam-steered scan-lines. Minimizing the number of required scan-lines can\nsignificantly enhance frame rate, field of view, energy efficiency, and data\ntransfer speeds. Existing approaches typically use static subsampling schemes\nin combination with sparsity-based or, more recently, deep-learning-based\nrecovery. In this work, we introduce an adaptive subsampling method that\nmaximizes intrinsic information gain in-situ, employing a Sylvester Normalizing\nFlow encoder to infer an approximate Bayesian posterior under partial\nobservation in real-time. Using the Bayesian posterior and a deep generative\nmodel for future observations, we determine the subsampling scheme that\nmaximizes the mutual information between the subsampled observations, and the\nnext frame of the video. We evaluate our approach using the EchoNet cardiac\nultrasound video dataset and demonstrate that our active sampling method\noutperforms competitive baselines, including uniform and variable-density\nrandom sampling, as well as equidistantly spaced scan-lines, improving mean\nabsolute reconstruction error by 15%. Moreover, posterior inference and the\nsampling scheme generation are performed in just 0.015 seconds (66Hz), making\nit fast enough for real-time 2D ultrasound imaging applications.\n", "date": "2025-01-07", "categories": []}
{"id": "2206.00775", "title": "Adaptive Local Neighborhood-based Neural Networks for MR Image\n  Reconstruction from Undersampled Data", "abstract": "  Recent medical image reconstruction techniques focus on generating\nhigh-quality medical images suitable for clinical use at the lowest possible\ncost and with the fewest possible adverse effects on patients. Recent works\nhave shown significant promise for reconstructing MR images from sparsely\nsampled k-space data using deep learning. In this work, we propose a technique\nthat rapidly estimates deep neural networks directly at reconstruction time by\nfitting them on small adaptively estimated neighborhoods of a training set. In\nbrief, our algorithm alternates between searching for neighbors in a data set\nthat are similar to the test reconstruction, and training a local network on\nthese neighbors followed by updating the test reconstruction. Because our\nreconstruction model is learned on a dataset that is in some sense similar to\nthe image being reconstructed rather than being fit on a large, diverse\ntraining set, it is more adaptive to new scans. It can also handle changes in\ntraining sets and flexible scan settings, while being relatively fast. Our\napproach, dubbed LONDN-MRI, was validated on multiple data sets using deep\nunrolled reconstruction networks. Reconstructions were performed at four fold\nand eight fold undersampling of k-space with 1D variable-density random\nphase-encode undersampling masks. Our results demonstrate that our proposed\nlocally-trained method produces higher-quality reconstructions compared to\nmodels trained globally on larger datasets as well as other scan-adaptive\nmethods.\n", "date": "2022-06-01", "categories": []}
{"id": "2406.12289", "title": "Stability of Data-Dependent Ridge-Regularization for Inverse Problems", "abstract": "  Theoretical guarantees for the robust solution of inverse problems have\nimportant implications for applications. To achieve both guarantees and high\nreconstruction quality, we propose learning a pixel-based ridge regularizer\nwith a data-dependent and spatially varying regularization strength. For this\narchitecture, we establish the existence of solutions to the associated\nvariational problem and the stability of its solution operator. Further, we\nprove that the reconstruction forms a maximum-a-posteriori approach.\nSimulations for biomedical imaging and material sciences demonstrate that the\napproach yields high-quality reconstructions even if only a small\ninstance-specific training set is available.\n", "date": "2024-06-18", "categories": []}
{"id": "2408.05705", "title": "TC-KANRecon: High-Quality and Accelerated MRI Reconstruction via\n  Adaptive KAN Mechanisms and Intelligent Feature Scaling", "abstract": "  Magnetic Resonance Imaging (MRI) has become essential in clinical diagnosis\ndue to its high resolution and multiple contrast mechanisms. However, the\nrelatively long acquisition time limits its broader application. To address\nthis issue, this study presents an innovative conditional guided diffusion\nmodel, named as TC-KANRecon, which incorporates the Multi-Free U-KAN (MF-UKAN)\nmodule and a dynamic clipping strategy. TC-KANRecon model aims to accelerate\nthe MRI reconstruction process through deep learning methods while maintaining\nthe quality of the reconstructed images. The MF-UKAN module can effectively\nbalance the tradeoff between image denoising and structure preservation.\nSpecifically, it presents the multi-head attention mechanisms and scalar\nmodulation factors, which significantly enhances the model's robustness and\nstructure preservation capabilities in complex noise environments. Moreover,\nthe dynamic clipping strategy in TC-KANRecon adjusts the cropping interval\naccording to the sampling steps, thereby mitigating image detail loss\ntypicalching the visual features of the images. Furthermore, the MC-Model\nincorporates full-sampling k-space information, realizing efficient fusion of\nconditional information, enhancing the model's ability to process complex data,\nand improving the realism and detail richness of reconstructed images.\nExperimental results demonstrate that the proposed method outperforms other MRI\nreconstruction methods in both qualitative and quantitative evaluations.\nNotably, TC-KANRecon method exhibits excellent reconstruction results when\nprocessing high-noise, low-sampling-rate MRI data. Our source code is available\nat https://github.com/lcbkmm/TC-KANRecon.\n", "date": "2024-08-11", "categories": []}
{"id": "2501.02473", "title": "IRIS: A Bayesian Approach for Image Reconstruction in Radio\n  Interferometry with expressive Score-Based priors", "abstract": "  Inferring sky surface brightness distributions from noisy interferometric\ndata in a principled statistical framework has been a key challenge in radio\nastronomy. In this work, we introduce Imaging for Radio Interferometry with\nScore-based models (IRIS). We use score-based models trained on optical images\nof galaxies as an expressive prior in combination with a Gaussian likelihood in\nthe uv-space to infer images of protoplanetary disks from visibility data of\nthe DSHARP survey conducted by ALMA. We demonstrate the advantages of this\nframework compared with traditional radio interferometry imaging algorithms,\nshowing that it produces plausible posterior samples despite the use of a\nmisspecified galaxy prior. Through coverage testing on simulations, we\nempirically evaluate the accuracy of this approach to generate calibrated\nposterior samples.\n", "date": "2025-01-05", "categories": []}
{"id": "2501.02872", "title": "Two-Dimensional Unknown View Tomography from Unknown Angle Distributions", "abstract": "  This study presents a technique for 2D tomography under unknown viewing\nangles when the distribution of the viewing angles is also unknown. Unknown\nview tomography (UVT) is a problem encountered in cryo-electron microscopy and\nin the geometric calibration of CT systems. There exists a moderate-sized\nliterature on the 2D UVT problem, but most existing 2D UVT algorithms assume\nknowledge of the angle distribution which is not available usually. Our\nproposed methodology formulates the problem as an optimization task based on\ncross-validation error, to estimate the angle distribution jointly with the\nunderlying 2D structure in an alternating fashion. We explore the algorithm's\ncapabilities for the case of two probability distribution models: a\nsemi-parametric mixture of von Mises densities and a probability mass function\nmodel. We evaluate our algorithm's performance under noisy projections using a\nPCA-based denoising technique and Graph Laplacian Tomography (GLT) driven by\norder statistics of the estimated distribution, to ensure near-perfect\nordering, and compare our algorithm to intuitive baselines.\n", "date": "2025-01-06", "categories": []}
{"id": "2501.03030", "title": "DDRM-PR: Fourier Phase Retrieval using Denoising Diffusion Restoration\n  Models", "abstract": "  Diffusion models have demonstrated their utility as learned priors for\nsolving various inverse problems. However, most existing approaches are limited\nto linear inverse problems. This paper exploits the efficient and unsupervised\nposterior sampling framework of Denoising Diffusion Restoration Models (DDRM)\nfor the solution of nonlinear phase retrieval problem, which requires\nreconstructing an image from its noisy intensity-only measurements such as\nFourier intensity. The approach combines the model-based alternating-projection\nmethods with the DDRM to utilize pretrained unconditional diffusion priors for\nphase retrieval. The performance is demonstrated through both simulations and\nexperimental data. Results demonstrate the potential of this approach for\nimproving the alternating-projection methods as well as its limitations.\n", "date": "2025-01-06", "categories": []}
{"id": "2405.18816", "title": "Flow Priors for Linear Inverse Problems via Iterative Corrupted\n  Trajectory Matching", "abstract": "  Generative models based on flow matching have attracted significant attention\nfor their simplicity and superior performance in high-resolution image\nsynthesis. By leveraging the instantaneous change-of-variables formula, one can\ndirectly compute image likelihoods from a learned flow, making them enticing\ncandidates as priors for downstream tasks such as inverse problems. In\nparticular, a natural approach would be to incorporate such image probabilities\nin a maximum-a-posteriori (MAP) estimation problem. A major obstacle, however,\nlies in the slow computation of the log-likelihood, as it requires\nbackpropagating through an ODE solver, which can be prohibitively slow for\nhigh-dimensional problems. In this work, we propose an iterative algorithm to\napproximate the MAP estimator efficiently to solve a variety of linear inverse\nproblems. Our algorithm is mathematically justified by the observation that the\nMAP objective can be approximated by a sum of $N$ ``local MAP'' objectives,\nwhere $N$ is the number of function evaluations. By leveraging Tweedie's\nformula, we show that we can perform gradient steps to sequentially optimize\nthese objectives. We validate our approach for various linear inverse problems,\nsuch as super-resolution, deblurring, inpainting, and compressed sensing, and\ndemonstrate that we can outperform other methods based on flow matching. Code\nis available at https://github.com/YasminZhang/ICTM.\n", "date": "2024-05-29", "categories": []}
{"id": "2501.01464", "title": "Estimation of 3T MR images from 1.5T images regularized with Physics\n  based Constraint", "abstract": "  Limited accessibility to high field MRI scanners (such as 7T, 11T) has\nmotivated the development of post-processing methods to improve low field\nimages. Several existing post-processing methods have shown the feasibility to\nimprove 3T images to produce 7T-like images [3,18]. It has been observed that\nimproving lower field (LF, <=1.5T) images comes with additional challenges due\nto poor image quality such as the function mapping 1.5T and higher field (HF,\n3T) images is more complex than the function relating 3T and 7T images [10].\nExcept for [10], no method has been addressed to improve <=1.5T MRI images.\nFurther, most of the existing methods [3,18] including [10] require example\nimages, and also often rely on pixel to pixel correspondences between LF and HF\nimages which are usually inaccurate for <=1.5T images. The focus of this paper\nis to address the unsupervised framework for quality improvement of 1.5T images\nand avoid the expensive requirements of example images and associated image\nregistration. The LF and HF images are assumed to be related by a linear\ntransformation (LT). The unknown HF image and unknown LT are estimated in\nalternate minimization framework. Further, a physics based constraint is\nproposed that provides an additional non-linear function relating LF and HF\nimages in order to achieve the desired high contrast in estimated HF image. The\nexperimental results demonstrate that the proposed approach provides processed\n1.5T images, i.e., estimated 3T-like images with improved image quality, and is\ncomparably better than the existing methods addressing similar problems. The\nimprovement in image quality is also shown to provide better tissue\nsegmentation and volume quantification as compared to scanner acquired 1.5T\nimages.\n", "date": "2024-12-31", "categories": []}
{"id": "2501.01912", "title": "Exoplanet Detection via Differentiable Rendering", "abstract": "  Direct imaging of exoplanets is crucial for advancing our understanding of\nplanetary systems beyond our solar system, but it faces significant challenges\ndue to the high contrast between host stars and their planets. Wavefront\naberrations introduce speckles in the telescope science images, which are\npatterns of diffracted starlight that can mimic the appearance of planets,\ncomplicating the detection of faint exoplanet signals. Traditional\npost-processing methods, operating primarily in the image intensity domain, do\nnot integrate wavefront sensing data. These data, measured mainly for adaptive\noptics corrections, have been overlooked as a potential resource for\npost-processing, partly due to the challenge of the evolving nature of\nwavefront aberrations. In this paper, we present a differentiable rendering\napproach that leverages these wavefront sensing data to improve exoplanet\ndetection. Our differentiable renderer models wave-based light propagation\nthrough a coronagraphic telescope system, allowing gradient-based optimization\nto significantly improve starlight subtraction and increase sensitivity to\nfaint exoplanets. Simulation experiments based on the James Webb Space\nTelescope configuration demonstrate the effectiveness of our approach,\nachieving substantial improvements in contrast and planet detection limits. Our\nresults showcase how the computational advancements enabled by differentiable\nrendering can revitalize previously underexploited wavefront data, opening new\navenues for enhancing exoplanet imaging and characterization.\n", "date": "2025-01-03", "categories": []}
{"id": "2501.01924", "title": "Transformer-Driven Inverse Problem Transform for Fast Blind\n  Hyperspectral Image Dehazing", "abstract": "  Hyperspectral dehazing (HyDHZ) has become a crucial signal processing\ntechnology to facilitate the subsequent identification and classification\ntasks, as the airborne visible/infrared imaging spectrometer (AVIRIS) data\nportal reports a massive portion of haze-corrupted areas in typical\nhyperspectral remote sensing images. The idea of inverse problem transform\n(IPT) has been proposed in recent remote sensing literature in order to\nreformulate a hardly tractable inverse problem (e.g., HyDHZ) into a relatively\nsimple one. Considering the emerging spectral super-resolution (SSR) technique,\nwhich spectrally upsamples multispectral data to hyperspectral data, we aim to\nsolve the challenging HyDHZ problem by reformulating it as an SSR problem.\nRoughly speaking, the proposed algorithm first automatically selects some\nuncorrupted/informative spectral bands, from which SSR is applied to spectrally\nupsample the selected bands in the feature space, thereby obtaining a clean\nhyperspectral image (HSI). The clean HSI is then further refined by a deep\ntransformer network to obtain the final dehazed HSI, where a global attention\nmechanism is designed to capture nonlocal information. There are very few HyDHZ\nworks in existing literature, and this article introduces the powerful\nspatial-spectral transformer into HyDHZ for the first time. Remarkably, the\nproposed transformer-driven IPT-based HyDHZ (T2HyDHZ) is a blind algorithm\nwithout requiring the user to manually select the corrupted region. Extensive\nexperiments demonstrate the superiority of T2HyDHZ with less color distortion.\n", "date": "2025-01-03", "categories": []}
{"id": "2312.16014", "title": "Passive Non-Line-of-Sight Imaging with Light Transport Modulation", "abstract": "  Passive non-line-of-sight (NLOS) imaging has witnessed rapid development in\nrecent years, due to its ability to image objects that are out of sight. The\nlight transport condition plays an important role in this task since changing\nthe conditions will lead to different imaging models. Existing learning-based\nNLOS methods usually train independent models for different light transport\nconditions, which is computationally inefficient and impairs the practicality\nof the models. In this work, we propose NLOS-LTM, a novel passive NLOS imaging\nmethod that effectively handles multiple light transport conditions with a\nsingle network. We achieve this by inferring a latent light transport\nrepresentation from the projection image and using this representation to\nmodulate the network that reconstructs the hidden image from the projection\nimage. We train a light transport encoder together with a vector quantizer to\nobtain the light transport representation. To further regulate this\nrepresentation, we jointly learn both the reconstruction network and the\nreprojection network during training. A set of light transport modulation\nblocks is used to modulate the two jointly trained networks in a multi-scale\nway. Extensive experiments on a large-scale passive NLOS dataset demonstrate\nthe superiority of the proposed method. The code is available at\nhttps://github.com/JerryOctopus/NLOS-LTM.\n", "date": "2023-12-26", "categories": []}
{"id": "2305.05651", "title": "SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions", "abstract": "  Self-supervised image denoising implies restoring the signal from a noisy\nimage without access to the ground truth. State-of-the-art solutions for this\ntask rely on predicting masked pixels with a fully-convolutional neural\nnetwork. This most often requires multiple forward passes, information about\nthe noise model, or intricate regularization functions. In this paper, we\npropose a Swin Transformer-based Image Autoencoder (SwinIA), the first\nfully-transformer architecture for self-supervised denoising. The flexibility\nof the attention mechanism helps to fulfill the blind-spot property that\nconvolutional counterparts normally approximate. SwinIA can be trained\nend-to-end with a simple mean squared error loss without masking and does not\nrequire any prior knowledge about clean data or noise distribution. Simple to\nuse, SwinIA establishes the state of the art on several common benchmarks.\n", "date": "2023-05-09", "categories": []}
{"id": "2409.04050", "title": "EigenSR: Eigenimage-Bridged Pre-Trained RGB Learners for Single\n  Hyperspectral Image Super-Resolution", "abstract": "  Single hyperspectral image super-resolution (single-HSI-SR) aims to improve\nthe resolution of a single input low-resolution HSI. Due to the bottleneck of\ndata scarcity, the development of single-HSI-SR lags far behind that of RGB\nnatural images. In recent years, research on RGB SR has shown that models\npre-trained on large-scale benchmark datasets can greatly improve performance\non unseen data, which may stand as a remedy for HSI. But how can we transfer\nthe pre-trained RGB model to HSI, to overcome the data-scarcity bottleneck?\nBecause of the significant difference in the channels between the pre-trained\nRGB model and the HSI, the model cannot focus on the correlation along the\nspectral dimension, thus limiting its ability to utilize on HSI. Inspired by\nthe HSI spatial-spectral decoupling, we propose a new framework that first\nfine-tunes the pre-trained model with the spatial components (known as\neigenimages), and then infers on unseen HSI using an iterative spectral\nregularization (ISR) to maintain the spectral correlation. The advantages of\nour method lie in: 1) we effectively inject the spatial texture processing\ncapabilities of the pre-trained RGB model into HSI while keeping spectral\nfidelity, 2) learning in the spectral-decorrelated domain can improve the\ngeneralizability to spectral-agnostic data, and 3) our inference in the\neigenimage domain naturally exploits the spectral low-rank property of HSI,\nthereby reducing the complexity. This work bridges the gap between pre-trained\nRGB models and HSI via eigenimages, addressing the issue of limited HSI\ntraining data, hence the name EigenSR. Extensive experiments show that EigenSR\noutperforms the state-of-the-art (SOTA) methods in both spatial and spectral\nmetrics.\n", "date": "2024-09-06", "categories": []}
{"id": "2401.16928", "title": "Dynamic MRI reconstruction using low-rank plus sparse decomposition with\n  smoothness regularization", "abstract": "  The low-rank plus sparse (L+S) decomposition model has enabled better\nreconstruction of dynamic magnetic resonance imaging (dMRI) with separation\ninto background (L) and dynamic (S) component. However, use of low-rank prior\nalone may not fully explain the slow variations or smoothness of the background\npart at the local scale. In this paper, we propose a smoothness-regularized L+S\n(SR-L+S) model for dMRI reconstruction from highly undersampled k-t-space data.\nWe exploit joint low-rank and smooth priors on the background component of dMRI\nto better capture both its global and local temporal correlated structures.\nExtending the L+S formulation, the low-rank property is encoded by the nuclear\nnorm, while the smoothness by a general \\ell_{p}-norm penalty on the local\ndifferences of the columns of L. The additional smoothness regularizer can\npromote piecewise local consistency between neighboring frames. By smoothing\nout the noise and dynamic activities, it allows accurate recovery of the\nbackground part, and subsequently more robust dMRI reconstruction. Extensive\nexperiments on multi-coil cardiac and synthetic data shows that the SR-L+S\nmodel outp\n", "date": "2024-01-30", "categories": []}
{"id": "2412.14873", "title": "Zero-Shot Artifact2Artifact: Self-incentive artifact removal for\n  photoacoustic imaging without any data", "abstract": "  Photoacoustic imaging (PAI) uniquely combines optical contrast with the\npenetration depth of ultrasound, making it critical for clinical applications.\nHowever, the quality of 3D PAI is often degraded due to reconstruction\nartifacts caused by the sparse and angle-limited configuration of detector\narrays. Existing iterative or deep learning-based methods are either\ntime-consuming or require large training datasets, significantly limiting their\npractical application. Here, we propose Zero-Shot Artifact2Artifact (ZS-A2A), a\nzero-shot self-supervised artifact removal method based on a super-lightweight\nnetwork, which leverages the fact that reconstruction artifacts are sensitive\nto irregularities caused by data loss. By introducing random perturbations to\nthe acquired PA data, it spontaneously generates subset data, which in turn\nstimulates the network to learn the artifact patterns in the reconstruction\nresults, thus enabling zero-shot artifact removal. This approach requires\nneither training data nor prior knowledge of the artifacts, and is capable of\nartifact removal for 3D PAI. For maximum amplitude projection (MAP) images or\nslice images in 3D PAI acquired with arbitrarily sparse or angle-limited\ndetector arrays, ZS-A2A employs a self-incentive strategy to complete artifact\nremoval and improves the Contrast-to-Noise Ratio (CNR). We validated ZS-A2A in\nboth simulation study and $ in\\ vivo $ animal experiments. Results demonstrate\nthat ZS-A2A achieves state-of-the-art (SOTA) performance compared to existing\nzero-shot methods, and for the $ in\\ vivo $ rat liver, ZS-A2A improves CNR from\n17.48 to 43.46 in just 8 seconds. The project for ZS-A2A will be available in\nthe following GitHub repository: https://github.com/JaegerCQ/ZS-A2A.\n", "date": "2024-12-19", "categories": []}
{"id": "2412.19459", "title": "A Prototype Unit for Image De-raining using Time-Lapse Data", "abstract": "  We address the challenge of single-image de-raining, a task that involves\nrecovering rain-free background information from a single rain image. While\nrecent advancements have utilized real-world time-lapse data for training,\nenabling the estimation of consistent backgrounds and realistic rain streaks,\nthese methods often suffer from computational and memory consumption, limiting\ntheir applicability in real-world scenarios. In this paper, we introduce a\nnovel solution: the Rain Streak Prototype Unit (RsPU). The RsPU efficiently\nencodes rain streak-relevant features as real-time prototypes derived from\ntime-lapse data, eliminating the need for excessive memory resources. Our\nde-raining network combines encoder-decoder networks with the RsPU, allowing us\nto learn and encapsulate diverse rain streak-relevant features as concise\nprototypes, employing an attention-based approach. To ensure the effectiveness\nof our approach, we propose a feature prototype loss encompassing cohesion and\ndivergence components. This loss function captures both the compactness and\ndiversity aspects of the prototypical rain streak features within the RsPU. Our\nmethod evaluates various de-raining benchmarks, accompanied by comprehensive\nablation studies. We show that it can achieve competitive results in various\nrain images compared to state-of-the-art methods.\n", "date": "2024-12-27", "categories": []}
{"id": "2412.19483", "title": "Learning Radiance Fields from a Single Snapshot Compressive Image", "abstract": "  In this paper, we explore the potential of Snapshot Compressive Imaging (SCI)\ntechnique for recovering the underlying 3D scene structure from a single\ntemporal compressed image. SCI is a cost-effective method that enables the\nrecording of high-dimensional data, such as hyperspectral or temporal\ninformation, into a single image using low-cost 2D imaging sensors. To achieve\nthis, a series of specially designed 2D masks are usually employed, reducing\nstorage and transmission requirements and offering potential privacy\nprotection. Inspired by this, we take one step further to recover the encoded\n3D scene information leveraging powerful 3D scene representation capabilities\nof neural radiance fields (NeRF). Specifically, we propose SCINeRF, in which we\nformulate the physical imaging process of SCI as part of the training of NeRF,\nallowing us to exploit its impressive performance in capturing complex scene\nstructures. In addition, we further integrate the popular 3D Gaussian Splatting\n(3DGS) framework and propose SCISplat to improve 3D scene reconstruction\nquality and training/rendering speed by explicitly optimizing point clouds into\n3D Gaussian representations. To assess the effectiveness of our method, we\nconduct extensive evaluations using both synthetic data and real data captured\nby our SCI system. Experimental results demonstrate that our proposed approach\nsurpasses the state-of-the-art methods in terms of image reconstruction and\nnovel view synthesis. Moreover, our method also exhibits the ability to render\nhigh frame-rate multi-view consistent images in real time by leveraging SCI and\nthe rendering capabilities of 3DGS. Codes will be available at:\nhttps://github.com/WU- CVGL/SCISplat.\n", "date": "2024-12-27", "categories": []}
{"id": "2412.14603", "title": "Successive optimization of optics and post-processing with\n  differentiable coherent PSF operator and field information", "abstract": "  Recently, the joint design of optical systems and downstream algorithms is\nshowing significant potential. However, existing rays-described methods are\nlimited to optimizing geometric degradation, making it difficult to fully\nrepresent the optical characteristics of complex, miniaturized lenses\nconstrained by wavefront aberration or diffraction effects. In this work, we\nintroduce a precise optical simulation model, and every operation in pipeline\nis differentiable. This model employs a novel initial value strategy to enhance\nthe reliability of intersection calculation on high aspherics. Moreover, it\nutilizes a differential operator to reduce memory consumption during coherent\npoint spread function calculations. To efficiently address various degradation,\nwe design a joint optimization procedure that leverages field information.\nGuided by a general restoration network, the proposed method not only enhances\nthe image quality, but also successively improves the optical performance\nacross multiple lenses that are already in professional level. This joint\noptimization pipeline offers innovative insights into the practical design of\nsophisticated optical systems and post-processing algorithms. The source code\nwill be made publicly available at\nhttps://github.com/Zrr-ZJU/Successive-optimization\n", "date": "2024-12-19", "categories": []}
{"id": "2412.16828", "title": "Technical Report: Towards Spatial Feature Regularization in\n  Deep-Learning-Based Array-SAR Reconstruction", "abstract": "  Array synthetic aperture radar (Array-SAR), also known as tomographic SAR\n(TomoSAR), has demonstrated significant potential for high-quality 3D mapping,\nparticularly in urban areas.While deep learning (DL) methods have recently\nshown strengths in reconstruction, most studies rely on pixel-by-pixel\nreconstruction, neglecting spatial features like building structures, leading\nto artifacts such as holes and fragmented edges. Spatial feature\nregularization, effective in traditional methods, remains underexplored in\nDL-based approaches. Our study integrates spatial feature regularization into\nDL-based Array-SAR reconstruction, addressing key questions: What spatial\nfeatures are relevant in urban-area mapping? How can these features be\neffectively described, modeled, regularized, and incorporated into DL networks?\nThe study comprises five phases: spatial feature description and modeling,\nregularization, feature-enhanced network design, evaluation, and discussions.\nSharp edges and geometric shapes in urban scenes are analyzed as key features.\nAn intra-slice and inter-slice strategy is proposed, using 2D slices as\nreconstruction units and fusing them into 3D scenes through parallel and serial\nfusion. Two computational frameworks-iterative reconstruction with enhancement\nand light reconstruction with enhancement-are designed, incorporating spatial\nfeature modules into DL networks, leading to four specialized reconstruction\nnetworks. Using our urban building simulation dataset and two public datasets,\nsix tests evaluate close-point resolution, structural integrity, and robustness\nin urban scenarios. Results show that spatial feature regularization\nsignificantly improves reconstruction accuracy, retrieves more complete\nbuilding structures, and enhances robustness by reducing noise and outliers.\n", "date": "2024-12-22", "categories": []}
{"id": "2402.06063", "title": "3D-2D Neural Nets for Phase Retrieval in Noisy Interferometric Imaging", "abstract": "  In recent years, neural networks have been used to solve phase retrieval\nproblems in imaging with superior accuracy and speed than traditional\ntechniques, especially in the presence of noise. However, in the context of\ninterferometric imaging, phase noise has been largely unaddressed by existing\nneural network architectures. Such noise arises naturally in an interferometer\ndue to mechanical instabilities or atmospheric turbulence, limiting measurement\nacquisition times and posing a challenge in scenarios with limited light\nintensity, such as remote sensing. Here, we introduce a 3D-2D Phase Retrieval\nU-Net (PRUNe) that takes noisy and randomly phase-shifted interferograms as\ninputs, and outputs a single 2D phase image. A 3D downsampling convolutional\nencoder captures correlations within and between frames to produce a 2D latent\nspace, which is upsampled by a 2D decoder into a phase image. We test our model\nagainst a state-of-the-art singular value decomposition algorithm and find\nPRUNe reconstructions consistently show more accurate and smooth\nreconstructions, with a x2.5 - 4 lower mean squared error at multiple\nsignal-to-noise ratios for interferograms with low (< 1 photon/pixel) and high\n(~100 photons/pixel) signal intensity. Our model presents a faster and more\naccurate approach to perform phase retrieval in extremely low light intensity\ninterferometry in presence of phase noise, and will find application in other\nmulti-frame noisy imaging techniques.\n", "date": "2024-02-08", "categories": []}
{"id": "2412.14925", "title": "Automatic Spectral Calibration of Hyperspectral Images:Method, Dataset\n  and Benchmark", "abstract": "  Hyperspectral image (HSI) densely samples the world in both the space and\nfrequency domain and therefore is more distinctive than RGB images. Usually,\nHSI needs to be calibrated to minimize the impact of various illumination\nconditions. The traditional way to calibrate HSI utilizes a physical reference,\nwhich involves manual operations, occlusions, and/or limits camera mobility.\nThese limitations inspire this paper to automatically calibrate HSIs using a\nlearning-based method. Towards this goal, a large-scale HSI calibration dataset\nis created, which has 765 high-quality HSI pairs covering diversified natural\nscenes and illuminations. The dataset is further expanded to 7650 pairs by\ncombining with 10 different physically measured illuminations. A spectral\nillumination transformer (SIT) together with an illumination attention module\nis proposed. Extensive benchmarks demonstrate the SoTA performance of the\nproposed SIT. The benchmarks also indicate that low-light conditions are more\nchallenging than normal conditions. The dataset and codes are available\nonline:https://github.com/duranze/Automatic-spectral-calibration-of-HSI\n", "date": "2024-12-19", "categories": []}
{"id": "2401.05055", "title": "Deep learning in motion deblurring: current status, benchmarks and\n  future prospects", "abstract": "  Motion deblurring is one of the fundamental problems of computer vision and\nhas received continuous attention. The variability in blur, both within and\nacross images, imposes limitations on non-blind deblurring techniques that rely\non estimating the blur kernel. As a response, blind motion deblurring has\nemerged, aiming to restore clear and detailed images without prior knowledge of\nthe blur type, fueled by the advancements in deep learning methodologies.\nDespite strides in this field, a comprehensive synthesis of recent progress in\ndeep learning-based blind motion deblurring is notably absent. This paper fills\nthat gap by providing an exhaustive overview of the role of deep learning in\nblind motion deblurring, encompassing datasets, evaluation metrics, and methods\ndeveloped over the last six years. Specifically, we first introduce the types\nof motion blur and the fundamental principles of deblurring. Next, we outline\nthe shortcomings of traditional non-blind deblurring algorithms, emphasizing\nthe advantages of employing deep learning techniques for deblurring tasks.\nFollowing this, we categorize and summarize existing blind motion deblurring\nmethods based on different backbone networks, including convolutional neural\nnetworks, generative adversarial networks, recurrent neural networks, and\nTransformer networks. Subsequently, we elaborate not only on the fundamental\nprinciples of these different categories but also provide a comprehensive\nsummary and comparison of their advantages and limitations. Qualitative and\nquantitative experimental results conducted on four widely used datasets\nfurther compare the performance of SOTA methods. Finally, an analysis of\npresent challenges and future pathways. All collected models, benchmark\ndatasets, source code links, and codes for evaluation have been made publicly\navailable at https://github.com/VisionVerse/Blind-Motion-Deblurring-Survey\n", "date": "2024-01-10", "categories": []}
{"id": "2412.14220", "title": "Distilled Pooling Transformer Encoder for Efficient Realistic Image\n  Dehazing", "abstract": "  This paper proposes a lightweight neural network designed for realistic image\ndehazing, utilizing a Distilled Pooling Transformer Encoder, named DPTE-Net.\nRecently, while vision transformers (ViTs) have achieved great success in\nvarious vision tasks, their self-attention (SA) module's complexity scales\nquadratically with image resolution, hindering their applicability on\nresource-constrained devices. To overcome this, the proposed DPTE-Net\nsubstitutes traditional SA modules with efficient pooling mechanisms,\nsignificantly reducing computational demands while preserving ViTs' learning\ncapabilities. To further enhance semantic feature learning, a\ndistillation-based training process is implemented which transfers rich\nknowledge from a larger teacher network to DPTE-Net. Additionally, DPTE-Net is\ntrained within a generative adversarial network (GAN) framework, leveraging the\nstrong generalization of GAN in image restoration, and employs a\ntransmission-aware loss function to dynamically adapt to varying haze\ndensities. Experimental results on various benchmark datasets have shown that\nthe proposed DPTE-Net can achieve competitive dehazing performance when\ncompared to state-of-the-art methods while maintaining low computational\ncomplexity, making it a promising solution for resource-limited applications.\nThe code of this work is available at https://github.com/tranleanh/dpte-net.\n", "date": "2024-12-18", "categories": []}
{"id": "2412.14599", "title": "Fast inverse lithography based on a model-driven block stacking\n  convolutional neural network", "abstract": "  In the realm of lithography, Optical Proximity Correction (OPC) is a crucial\nresolution enhancement technique that optimizes the transmission function of\nphotomasks on a pixel-based to effectively counter Optical Proximity Effects\n(OPE). However, conventional pixel-based OPC methods often generate patterns\nthat pose manufacturing challenges, thereby leading to the increased cost in\npractical scenarios. This paper presents a novel inverse lithographic approach\nto OPC, employing a model-driven, block stacking deep learning framework that\nexpedites the generation of masks conducive to manufacturing. This method is\nfounded on vector lithography modelling and streamlines the training process by\neliminating the requirement for extensive labeled datasets. Furthermore,\ndiversity of mask patterns is enhanced by employing a wave function collapse\nalgorithm, which facilitates the random generation of a multitude of target\npatterns, therefore significantly expanding the range of mask paradigm.\nNumerical experiments have substantiated the efficacy of the proposed\nend-to-end approach, highlighting its superior capability to manage mask\ncomplexity within the context of advanced OPC lithography. This advancement is\nanticipated to enhance the feasibility and economic viability of OPC technology\nwithin actual manufacturing environments.\n", "date": "2024-12-19", "categories": []}
{"id": "2412.14705", "title": "Event-assisted 12-stop HDR Imaging of Dynamic Scene", "abstract": "  High dynamic range (HDR) imaging is a crucial task in computational\nphotography, which captures details across diverse lighting conditions.\nTraditional HDR fusion methods face limitations in dynamic scenes with extreme\nexposure differences, as aligning low dynamic range (LDR) frames becomes\nchallenging due to motion and brightness variation. In this work, we propose a\nnovel 12-stop HDR imaging approach for dynamic scenes, leveraging a dual-camera\nsystem with an event camera and an RGB camera. The event camera provides\ntemporally dense, high dynamic range signals that improve alignment between LDR\nframes with large exposure differences, reducing ghosting artifacts caused by\nmotion. Also, a real-world finetuning strategy is proposed to increase the\ngeneralization of alignment module on real-world events. Additionally, we\nintroduce a diffusion-based fusion module that incorporates image priors from\npre-trained diffusion models to address artifacts in high-contrast regions and\nminimize errors from the alignment process. To support this work, we developed\nthe ESHDR dataset, the first dataset for 12-stop HDR imaging with synchronized\nevent signals, and validated our approach on both simulated and real-world\ndata. Extensive experiments demonstrate that our method achieves\nstate-of-the-art performance, successfully extending HDR imaging to 12 stops in\ndynamic scenes.\n", "date": "2024-12-19", "categories": []}
{"id": "2403.13680", "title": "Step-Calibrated Diffusion for Biomedical Optical Image Restoration", "abstract": "  High-quality, high-resolution medical imaging is essential for clinical care.\nRaman-based biomedical optical imaging uses non-ionizing infrared radiation to\nevaluate human tissues in real time and is used for early cancer detection,\nbrain tumor diagnosis, and intraoperative tissue analysis. Unfortunately,\noptical imaging is vulnerable to image degradation due to laser scattering and\nabsorption, which can result in diagnostic errors and misguided treatment.\nRestoration of optical images is a challenging computer vision task because the\nsources of image degradation are multi-factorial, stochastic, and\ntissue-dependent, preventing a straightforward method to obtain paired\nlow-quality/high-quality data. Here, we present Restorative Step-Calibrated\nDiffusion (RSCD), an unpaired diffusion-based image restoration method that\nuses a step calibrator model to dynamically determine the number of steps\nrequired to complete the reverse diffusion process for image restoration. RSCD\noutperforms other widely used unpaired image restoration methods on both image\nquality and perceptual evaluation metrics for restoring optical images. Medical\nimaging experts consistently prefer images restored using RSCD in blinded\ncomparison experiments and report minimal to no hallucinations. Finally, we\nshow that RSCD improves performance on downstream clinical imaging tasks,\nincluding automated brain tumor diagnosis and deep tissue imaging. Our code is\navailable at\nhttps://github.com/MLNeurosurg/restorative_step-calibrated_diffusion.\n", "date": "2024-03-20", "categories": []}
{"id": "2408.02367", "title": "StoDIP: Efficient 3D MRF image reconstruction with deep image priors and\n  stochastic iterations", "abstract": "  Magnetic Resonance Fingerprinting (MRF) is a time-efficient approach to\nquantitative MRI for multiparametric tissue mapping. The reconstruction of\nquantitative maps requires tailored algorithms for removing aliasing artefacts\nfrom the compressed sampled MRF acquisitions. Within approaches found in the\nliterature, many focus solely on two-dimensional (2D) image reconstruction,\nneglecting the extension to volumetric (3D) scans despite their higher\nrelevance and clinical value. A reason for this is that transitioning to 3D\nimaging without appropriate mitigations presents significant challenges,\nincluding increased computational cost and storage requirements, and the need\nfor large amount of ground-truth (artefact-free) data for training. To address\nthese issues, we introduce StoDIP, a new algorithm that extends the\nground-truth-free Deep Image Prior (DIP) reconstruction to 3D MRF imaging.\nStoDIP employs memory-efficient stochastic updates across the multicoil MRF\ndata, a carefully selected neural network architecture, as well as faster\nnonuniform FFT (NUFFT) transformations. This enables a faster convergence\ncompared against a conventional DIP implementation without these features.\nTested on a dataset of whole-brain scans from healthy volunteers, StoDIP\ndemonstrated superior performance over the ground-truth-free reconstruction\nbaselines, both quantitatively and qualitatively.\n", "date": "2024-08-05", "categories": []}
{"id": "2412.07527", "title": "Deep Joint Unrolling for Deblurring and Low-Light Image Enhancement\n  (JUDE)", "abstract": "  Low-light and blurring issues are prevalent when capturing photos at night,\noften due to the use of long exposure to address dim environments. Addressing\nthese joint problems can be challenging and error-prone if an end-to-end model\nis trained without incorporating an appropriate physical model. In this paper,\nwe introduce JUDE, a Deep Joint Unrolling for Deblurring and Low-Light Image\nEnhancement, inspired by the image physical model. Based on Retinex theory and\nthe blurring model, the low-light blurry input is iteratively deblurred and\ndecomposed, producing sharp low-light reflectance and illuminance through an\nunrolling mechanism. Additionally, we incorporate various modules to estimate\nthe initial blur kernel, enhance brightness, and eliminate noise in the final\nimage. Comprehensive experiments on LOL-Blur and Real-LOL-Blur demonstrate that\nour method outperforms existing techniques both quantitatively and\nqualitatively.\n", "date": "2024-12-10", "categories": []}
{"id": "2412.10882", "title": "Integrating Generative and Physics-Based Models for Ptychographic\n  Imaging with Uncertainty Quantification", "abstract": "  Ptychography is a scanning coherent diffractive imaging technique that\nenables imaging nanometer-scale features in extended samples. One main\nchallenge is that widely used iterative image reconstruction methods often\nrequire significant amount of overlap between adjacent scan locations, leading\nto large data volumes and prolonged acquisition times. To address this key\nlimitation, this paper proposes a Bayesian inversion method for ptychography\nthat performs effectively even with less overlap between neighboring scan\nlocations. Furthermore, the proposed method can quantify the inherent\nuncertainty on the ptychographic object, which is created by the ill-posed\nnature of the ptychographic inverse problem. At a high level, the proposed\nmethod first utilizes a deep generative model to learn the prior distribution\nof the object and then generates samples from the posterior distribution of the\nobject by using a Markov Chain Monte Carlo algorithm. Our results from\nsimulated ptychography experiments show that the proposed framework can\nconsistently outperform a widely used iterative reconstruction algorithm in\ncases of reduced overlap. Moreover, the proposed framework can provide\nuncertainty estimates that closely correlate with the true error, which is not\navailable in practice. The project website is available here.\n", "date": "2024-12-14", "categories": []}
{"id": "2412.11866", "title": "Event-based Motion Deblurring via Multi-Temporal Granularity Fusion", "abstract": "  Conventional frame-based cameras inevitably produce blurry effects due to\nmotion occurring during the exposure time. Event camera, a bio-inspired sensor\noffering continuous visual information could enhance the deblurring\nperformance. Effectively utilizing the high-temporal-resolution event data is\ncrucial for extracting precise motion information and enhancing deblurring\nperformance. However, existing event-based image deblurring methods usually\nutilize voxel-based event representations, losing the fine-grained temporal\ndetails that are mathematically essential for fast motion deblurring. In this\npaper, we first introduce point cloud-based event representation into the image\ndeblurring task and propose a Multi-Temporal Granularity Network (MTGNet). It\ncombines the spatially dense but temporally coarse-grained voxel-based event\nrepresentation and the temporally fine-grained but spatially sparse point\ncloud-based event. To seamlessly integrate such complementary representations,\nwe design a Fine-grained Point Branch. An Aggregation and Mapping Module (AMM)\nis proposed to align the low-level point-based features with frame-based\nfeatures and an Adaptive Feature Diffusion Module (AFDM) is designed to manage\nthe resolution discrepancies between event data and image data by enriching the\nsparse point feature. Extensive subjective and objective evaluations\ndemonstrate that our method outperforms current state-of-the-art approaches on\nboth synthetic and real-world datasets.\n", "date": "2024-12-16", "categories": []}
{"id": "2408.10283", "title": "Perception-based multiplicative noise removal using SDEs", "abstract": "  Multiplicative noise, also known as speckle or pepper noise, commonly affects\nimages produced by synthetic aperture radar (SAR), lasers, or optical lenses.\nUnlike additive noise, which typically arises from thermal processes or\nexternal factors, multiplicative noise is inherent to the system, originating\nfrom the fluctuation in diffuse reflections. These fluctuations result in\nmultiple copies of the same signal with varying magnitudes being combined.\nConsequently, despeckling, or removing multiplicative noise, necessitates\ndifferent techniques compared to those used for additive noise removal.\n  In this paper, we propose a novel approach using Stochastic Differential\nEquations based diffusion models to address multiplicative noise. We\ndemonstrate that multiplicative noise can be effectively modeled as a Geometric\nBrownian Motion process in the logarithmic domain. Utilizing the Fokker-Planck\nequation, we derive the corresponding reverse process for image denoising. To\nvalidate our method, we conduct extensive experiments on two different\ndatasets, comparing our approach to both classical signal processing techniques\nand contemporary CNN-based noise removal models. Our results indicate that the\nproposed method significantly outperforms existing methods on perception-based\nmetrics such as FID and LPIPS, while maintaining competitive performance on\ntraditional metrics like PSNR and SSIM.\n", "date": "2024-08-19", "categories": []}
{"id": "2412.09927", "title": "Neural Vector Tomography for Reconstructing a Magnetization Vector Field", "abstract": "  Discretized techniques for vector tomographic reconstructions are prone to\nproducing artifacts in the reconstructions. The quality of these\nreconstructions may further deteriorate as the amount of noise increases. In\nthis work, we instead model the underlying vector fields using smooth neural\nfields. Owing to the fact that the activation functions in the neural network\nmay be chosen to be smooth and the domain is no longer pixelated, the model\nresults in high-quality reconstructions, even under presence of noise. In the\ncase where we have underlying global continuous symmetry, we find that the\nneural network substantially improves the accuracy of the reconstruction over\nthe existing techniques.\n", "date": "2024-12-13", "categories": []}
{"id": "2412.10300", "title": "Iterating the Transient Light Transport Matrix for Non-Line-of-Sight\n  Imaging", "abstract": "  Active imaging systems sample the Transient Light Transport Matrix (TLTM) for\na scene by sequentially illuminating various positions in this scene using a\ncontrollable light source, and then measuring the resulting spatiotemporal\nlight transport with time of flight (ToF) sensors. Time-resolved\nNon-line-of-sight (NLOS) imaging employs an active imaging system that measures\npart of the TLTM of an intermediary relay surface, and uses the indirect\nreflections of light encoded within this TLTM to \"see around corners\". Such\nimaging systems have applications in diverse areas such as disaster response,\nremote surveillance, and autonomous navigation. While existing NLOS imaging\nsystems usually measure a subset of the full TLTM, development of customized\ngated Single Photon Avalanche Diode (SPAD) arrays\n\\cite{riccardo_fast-gated_2022} has made it feasible to probe the full\nmeasurement space. In this work, we demonstrate that the full TLTM on the relay\nsurface can be processed with efficient algorithms to computationally focus and\ndetect our illumination in different parts of the hidden scene, turning the\nrelay surface into a second-order active imaging system. These algorithms allow\nus to iterate on the measured, first-order TLTM, and extract a \\textbf{second\norder TLTM for surfaces in the hidden scene}. We showcase three applications of\nTLTMs in NLOS imaging: (1) Scene Relighting with novel illumination, (2)\nSeparation of direct and indirect components of light transport in the hidden\nscene, and (3) Dual Photography. Additionally, we empirically demonstrate that\nSPAD arrays enable parallel acquisition of photons, effectively mitigating long\nacquisition times.\n", "date": "2024-12-13", "categories": []}
{"id": "2412.09427", "title": "A Plug-and-Play Algorithm for 3D Video Super-Resolution of Single-Photon\n  LiDAR data", "abstract": "  Single-photon avalanche diodes (SPADs) are advanced sensors capable of\ndetecting individual photons and recording their arrival times with picosecond\nresolution using time-correlated Single-Photon Counting detection techniques.\nThey are used in various applications, such as LiDAR, and can capture\nhigh-speed sequences of binary single-photon images, offering great potential\nfor reconstructing 3D environments with high motion dynamics. To complement\nsingle-photon data, they are often paired with conventional passive cameras,\nwhich capture high-resolution (HR) intensity images at a lower frame rate.\nHowever, 3D reconstruction from SPAD data faces challenges. Aggregating\nmultiple binary measurements improves precision and reduces noise but can cause\nmotion blur in dynamic scenes. Additionally, SPAD arrays often have lower\nresolution than passive cameras. To address these issues, we propose a novel\ncomputational imaging algorithm to improve the 3D reconstruction of moving\nscenes from SPAD data by addressing the motion blur and increasing the native\nspatial resolution. We adopt a plug-and-play approach within an optimization\nscheme alternating between guided video super-resolution of the 3D scene, and\nprecise image realignment using optical flow. Experiments on synthetic data\nshow significantly improved image resolutions across various signal-to-noise\nratios and photon levels. We validate our method using real-world SPAD\nmeasurements on three practical situations with dynamic objects. First on\nfast-moving scenes in laboratory conditions at short range; second very low\nresolution imaging of people with a consumer-grade SPAD sensor from\nSTMicroelectronics; and finally, HR imaging of people walking outdoors in\ndaylight at a range of 325 meters under eye-safe illumination conditions using\na short-wave infrared SPAD camera. These results demonstrate the robustness and\nversatility of our approach.\n", "date": "2024-12-12", "categories": []}
{"id": "2412.07149", "title": "RAP-SR: RestorAtion Prior Enhancement in Diffusion Models for Realistic\n  Image Super-Resolution", "abstract": "  Benefiting from their powerful generative capabilities, pretrained diffusion\nmodels have garnered significant attention for real-world image\nsuper-resolution (Real-SR). Existing diffusion-based SR approaches typically\nutilize semantic information from degraded images and restoration prompts to\nactivate prior for producing realistic high-resolution images. However,\ngeneral-purpose pretrained diffusion models, not designed for restoration\ntasks, often have suboptimal prior, and manually defined prompts may fail to\nfully exploit the generated potential. To address these limitations, we\nintroduce RAP-SR, a novel restoration prior enhancement approach in pretrained\ndiffusion models for Real-SR. First, we develop the High-Fidelity Aesthetic\nImage Dataset (HFAID), curated through a Quality-Driven Aesthetic Image\nSelection Pipeline (QDAISP). Our dataset not only surpasses existing ones in\nfidelity but also excels in aesthetic quality. Second, we propose the\nRestoration Priors Enhancement Framework, which includes Restoration Priors\nRefinement (RPR) and Restoration-Oriented Prompt Optimization (ROPO) modules.\nRPR refines the restoration prior using the HFAID, while ROPO optimizes the\nunique restoration identifier, improving the quality of the resulting images.\nRAP-SR effectively bridges the gap between general-purpose models and the\ndemands of Real-SR by enhancing restoration prior. Leveraging the plug-and-play\nnature of RAP-SR, our approach can be seamlessly integrated into existing\ndiffusion-based SR methods, boosting their performance. Extensive experiments\ndemonstrate its broad applicability and state-of-the-art results. Codes and\ndatasets will be available upon acceptance.\n", "date": "2024-12-10", "categories": []}
{"id": "2412.07262", "title": "Deep Lidar-guided Image Deblurring", "abstract": "  The rise of portable Lidar instruments, including their adoption in\nsmartphones, opens the door to novel computational imaging techniques. Being an\nactive sensing instrument, Lidar can provide complementary data to passive\noptical sensors, particularly in situations like low-light imaging where motion\nblur can affect photos. In this paper, we study if the depth information\nprovided by mobile Lidar sensors is useful for the task of image deblurring and\nhow to integrate it with a general approach that transforms any\nstate-of-the-art neural deblurring model into a depth-aware one. To achieve\nthis, we developed a universal adapter structure that efficiently preprocesses\nthe depth information to modulate image features with depth features.\nAdditionally, we applied a continual learning strategy to pretrained\nencoder-decoder models, enabling them to incorporate depth information as an\nadditional input with minimal extra data requirements. We demonstrate that\nutilizing true depth information can significantly boost the effectiveness of\ndeblurring algorithms, as validated on a dataset with real-world depth data\ncaptured by a smartphone Lidar.\n", "date": "2024-12-10", "categories": []}
{"id": "2411.01893", "title": "A Global Depth-Range-Free Multi-View Stereo Transformer Network with\n  Pose Embedding", "abstract": "  In this paper, we propose a novel multi-view stereo (MVS) framework that gets\nrid of the depth range prior. Unlike recent prior-free MVS methods that work in\na pair-wise manner, our method simultaneously considers all the source images.\nSpecifically, we introduce a Multi-view Disparity Attention (MDA) module to\naggregate long-range context information within and across multi-view images.\nConsidering the asymmetry of the epipolar disparity flow, the key to our method\nlies in accurately modeling multi-view geometric constraints. We integrate pose\nembedding to encapsulate information such as multi-view camera poses, providing\nimplicit geometric constraints for multi-view disparity feature fusion\ndominated by attention. Additionally, we construct corresponding hidden states\nfor each source image due to significant differences in the observation quality\nof the same pixel in the reference frame across multiple source frames. We\nexplicitly estimate the quality of the current pixel corresponding to sampled\npoints on the epipolar line of the source image and dynamically update hidden\nstates through the uncertainty estimation module. Extensive results on the DTU\ndataset and Tanks&Temple benchmark demonstrate the effectiveness of our method.\nThe code is available at our project page:\nhttps://zju3dv.github.io/GD-PoseMVS/.\n", "date": "2024-11-04", "categories": []}
{"id": "2412.06135", "title": "A CT Image Denoising Method Based on Projection Domain Feature", "abstract": "  In order to improve image quality of projection in industrial applications,\ngenerally, a standard method is to increase the current or exposure time, which\nmight cause overexposure of detector units in areas of thin objects or\nbackgrounds. Increasing the projection sampling is a better method to address\nthe issue, but it also leads to significant noise in the reconstructed image.\nThis paper proposed a projection domain denoising algorithm based on the\nfeatures of the projection domain for this case. This algorithm utilized the\nsimilarity of projections of neighboring veiws to reduce image noise quickly\nand effectively. The availability of the algorithm proposed in this work has\nbeen conducted by numerical simulation and practical data experiments.\n", "date": "2024-12-09", "categories": []}
{"id": "2412.06191", "title": "Event fields: Capturing light fields at high speed, resolution, and\n  dynamic range", "abstract": "  Event cameras, which feature pixels that independently respond to changes in\nbrightness, are becoming increasingly popular in high-speed applications due to\ntheir lower latency, reduced bandwidth requirements, and enhanced dynamic range\ncompared to traditional frame-based cameras. Numerous imaging and vision\ntechniques have leveraged event cameras for high-speed scene understanding by\ncapturing high-framerate, high-dynamic range videos, primarily utilizing the\ntemporal advantages inherent to event cameras. Additionally, imaging and vision\ntechniques have utilized the light field-a complementary dimension to temporal\ninformation-for enhanced scene understanding. In this work, we propose \"Event\nFields\", a new approach that utilizes innovative optical designs for event\ncameras to capture light fields at high speed. We develop the underlying\nmathematical framework for Event Fields and introduce two foundational\nframeworks to capture them practically: spatial multiplexing to capture\ntemporal derivatives and temporal multiplexing to capture angular derivatives.\nTo realize these, we design two complementary optical setups one using a\nkaleidoscope for spatial multiplexing and another using a galvanometer for\ntemporal multiplexing. We evaluate the performance of both designs using a\ncustom-built simulator and real hardware prototypes, showcasing their distinct\nbenefits. Our event fields unlock the full advantages of typical light\nfields-like post-capture refocusing and depth estimation-now supercharged for\nhigh-speed and high-dynamic range scenes. This novel light-sensing paradigm\nopens doors to new applications in photography, robotics, and AR/VR, and\npresents fresh challenges in rendering and machine learning.\n", "date": "2024-12-09", "categories": []}
{"id": "2412.03617", "title": "End-to-end Triple-domain PET Enhancement: A Hybrid\n  Denoising-and-reconstruction Framework for Reconstructing Standard-dose PET\n  Images from Low-dose PET Sinograms", "abstract": "  As a sensitive functional imaging technique, positron emission tomography\n(PET) plays a critical role in early disease diagnosis. However, obtaining a\nhigh-quality PET image requires injecting a sufficient dose (standard dose) of\nradionuclides into the body, which inevitably poses radiation hazards to\npatients. To mitigate radiation hazards, the reconstruction of standard-dose\nPET (SPET) from low-dose PET (LPET) is desired. According to imaging theory,\nPET reconstruction process involves multiple domains (e.g., projection domain\nand image domain), and a significant portion of the difference between SPET and\nLPET arises from variations in the noise levels introduced during the sampling\nof raw data as sinograms. In light of these two facts, we propose an end-to-end\nTriPle-domain LPET EnhancemenT (TriPLET) framework, by leveraging the\nadvantages of a hybrid denoising-and-reconstruction process and a triple-domain\nrepresentation (i.e., sinograms, frequency spectrum maps, and images) to\nreconstruct SPET images from LPET sinograms. Specifically, TriPLET consists of\nthree sequentially coupled components including 1) a Transformer-assisted\ndenoising network that denoises the inputted LPET sinograms in the projection\ndomain, 2) a discrete-wavelet-transform-based reconstruction network that\nfurther reconstructs SPET from LPET in the wavelet domain, and 3) a pair-based\nadversarial network that evaluates the reconstructed SPET images in the image\ndomain. Extensive experiments on the real PET dataset demonstrate that our\nproposed TriPLET can reconstruct SPET images with the highest similarity and\nsignal-to-noise ratio to real data, compared with state-of-the-art methods.\n", "date": "2024-12-04", "categories": []}
{"id": "2412.03745", "title": "Deep Variational Bayesian Modeling of Haze Degradation Process", "abstract": "  Relying on the representation power of neural networks, most recent works\nhave often neglected several factors involved in haze degradation, such as\ntransmission (the amount of light reaching an observer from a scene over\ndistance) and atmospheric light. These factors are generally unknown, making\ndehazing problems ill-posed and creating inherent uncertainties. To account for\nsuch uncertainties and factors involved in haze degradation, we introduce a\nvariational Bayesian framework for single image dehazing. We propose to take\nnot only a clean image and but also transmission map as latent variables, the\nposterior distributions of which are parameterized by corresponding neural\nnetworks: dehazing and transmission networks, respectively. Based on a physical\nmodel for haze degradation, our variational Bayesian framework leads to a new\nobjective function that encourages the cooperation between them, facilitating\nthe joint training of and thereby boosting the performance of each other. In\nour framework, a dehazing network can estimate a clean image independently of a\ntransmission map estimation during inference, introducing no overhead.\nFurthermore, our model-agnostic framework can be seamlessly incorporated with\nother existing dehazing networks, greatly enhancing the performance\nconsistently across datasets and models.\n", "date": "2024-12-04", "categories": []}
{"id": "2412.03898", "title": "4D SlingBAG: spatial-temporal coupled Gaussian ball for large-scale\n  dynamic 3D photoacoustic iterative reconstruction", "abstract": "  Large-scale dynamic three-dimensional (3D) photoacoustic imaging (PAI) is\nsignificantly important in clinical applications. In practical implementations,\nlarge-scale 3D real-time PAI systems typically utilize sparse two-dimensional\n(2D) sensor arrays with certain angular deficiencies, necessitating advanced\niterative reconstruction (IR) algorithms to achieve quantitative PAI and reduce\nreconstruction artifacts. However, for existing IR algorithms, multi-frame 3D\nreconstruction leads to extremely high memory consumption and prolonged\ncomputation time, with limited consideration of the spatial-temporal continuity\nbetween data frames. Here, we propose a novel method, named the 4D sliding\nGaussian ball adaptive growth (4D SlingBAG) algorithm, based on the current\npoint cloud-based IR algorithm sliding Gaussian ball adaptive growth\n(SlingBAG), which has minimal memory consumption among IR methods. Our 4D\nSlingBAG method applies spatial-temporal coupled deformation functions to each\nGaussian sphere in point cloud, thus explicitly learning the deformations\nfeatures of the dynamic 3D PA scene. This allows for the efficient\nrepresentation of various physiological processes (such as pulsation) or\nexternal pressures (e.g., blood perfusion experiments) contributing to changes\nin vessel morphology and blood flow during dynamic 3D PAI, enabling highly\nefficient IR for dynamic 3D PAI. Simulation experiments demonstrate that 4D\nSlingBAG achieves high-quality dynamic 3D PA reconstruction. Compared to\nperforming reconstructions by using SlingBAG algorithm individually for each\nframe, our method significantly reduces computational time and keeps a\nextremely low memory consumption. The project for 4D SlingBAG can be found in\nthe following GitHub repository:\n\\href{https://github.com/JaegerCQ/4D-SlingBAG}{https://github.com/JaegerCQ/4D-SlingBAG}.\n", "date": "2024-12-05", "categories": []}
{"id": "2412.03941", "title": "Enhancing and Accelerating Diffusion-Based Inverse Problem Solving\n  through Measurements Optimization", "abstract": "  Diffusion models have recently demonstrated notable success in solving\ninverse problems. However, current diffusion model-based solutions typically\nrequire a large number of function evaluations (NFEs) to generate high-quality\nimages conditioned on measurements, as they incorporate only limited\ninformation at each step. To accelerate the diffusion-based inverse\nproblem-solving process, we introduce \\textbf{M}easurements\n\\textbf{O}ptimization (MO), a more efficient plug-and-play module for\nintegrating measurement information at each step of the inverse problem-solving\nprocess. This method is comprehensively evaluated across eight diverse linear\nand nonlinear tasks on the FFHQ and ImageNet datasets. By using MO, we\nestablish state-of-the-art (SOTA) performance across multiple tasks, with key\nadvantages: (1) it operates with no more than 100 NFEs, with phase retrieval on\nImageNet being the sole exception; (2) it achieves SOTA or near-SOTA results\neven at low NFE counts; and (3) it can be seamlessly integrated into existing\ndiffusion model-based solutions for inverse problems, such as DPS\n\\cite{chung2022diffusion} and Red-diff \\cite{mardani2023variational}. For\nexample, DPS-MO attains a peak signal-to-noise ratio (PSNR) of 28.71 dB on the\nFFHQ 256 dataset for high dynamic range imaging, setting a new SOTA benchmark\nwith only 100 NFEs, whereas current methods require between 1000 and 4000 NFEs\nfor comparable performance.\n", "date": "2024-12-05", "categories": []}
{"id": "2408.12671", "title": "Joint Image De-noising and Enhancement for Satellite-Based SAR", "abstract": "  The reconstructed images from the Synthetic Aperture Radar (SAR) data suffer\nfrom multiplicative noise as well as low contrast level. These two factors\nimpact the quality of the SAR images significantly and prevent any attempt to\nextract valuable information from the processed data. The necessity for\nmitigating these effects in the field of SAR imaging is of high importance.\nTherefore, in this paper, we address the aforementioned issues and propose a\ntechnique to handle these shortcomings simultaneously. In fact, we combine the\nde-noising and contrast enhancement processes into a unified algorithm. The\nimage enhancement is performed based on the Contrast Limited Adaptive Histogram\nEqualization (CLAHE) technique.\n  The verification of the proposed algorithm is performed by experimental\nresults based on the data that has been collected from the European Space\nAgency's ERS-2 satellite which operates in strip-map mode.\n", "date": "2024-08-06", "categories": []}
{"id": "2410.08688", "title": "Chain-of-Restoration: Multi-Task Image Restoration Models are Zero-Shot\n  Step-by-Step Universal Image Restorers", "abstract": "  Despite previous image restoration (IR) methods have often concentrated on\nisolated degradations, recent research has increasingly focused on addressing\ncomposite degradations involving a complex combination of multiple isolated\ndegradations. However, current IR methods for composite degradations require\nbuilding training data that contain an exponential number of possible\ndegradation combinations, which brings in a significant burden. To alleviate\nthis issue, this paper proposes a new task setting, i.e. Universal Image\nRestoration (UIR). Specifically, UIR doesn't require training on all the\ndegradation combinations but only on a set of degradation bases and then\nremoving any degradation that these bases can potentially compose in a\nzero-shot manner. Inspired by the Chain-of-Thought that prompts large language\nmodels (LLMs) to address problems step-by-step, we propose Chain-of-Restoration\n(CoR) mechanism, which instructs models to remove unknown composite\ndegradations step-by-step. By integrating a simple Degradation Discriminator\ninto pre-trained multi-task models, CoR facilitates the process where models\nremove one degradation basis per step, continuing this process until the image\nis fully restored from the unknown composite degradation. Extensive experiments\nshow that CoR can significantly improve model performance in removing composite\ndegradations, achieving comparable or better results than those\nstate-of-the-art (SoTA) methods trained on all degradations.\n", "date": "2024-10-11", "categories": []}
{"id": "2410.21556", "title": "Super-resolution in disordered media using neural networks", "abstract": "  We propose a methodology that exploits large and diverse data sets to\naccurately estimate the ambient medium's Green's functions in strongly\nscattering media. Given these estimates, obtained with and without the use of\nneural networks, excellent imaging results are achieved, with a resolution that\nis better than that of a homogeneous medium. This phenomenon, also known as\nsuper-resolution, occurs because the ambient scattering medium effectively\nenhances the physical imaging aperture. This work has been submitted to the\nIEEE for possible publication. Copyright may be transferred without notice,\nafter which this version may no longer be accessible.\n", "date": "2024-10-28", "categories": []}
{"id": "2412.03036", "title": "Fan-Beam CT Reconstruction for Unaligned Sparse-View X-ray Baggage\n  Dataset", "abstract": "  Computed Tomography (CT) is a technology that reconstructs cross-sectional\nimages using X-ray images taken from multiple directions. In CT, hundreds of\nX-ray images acquired as the X-ray source and detector rotate around a central\naxis, are used for precise reconstruction. In security baggage inspection,\nX-ray imaging is also widely used; however, unlike the rotating systems in\nmedical CT, stationary X-ray systems are more common, and publicly available\nreconstructed data are limited. This makes it challenging to obtain large-scale\n3D labeled data and voxel representations essential for training. To address\nthese limitations, our study presents a calibration and reconstruction method\nusing an unaligned sparse multi-view X-ray baggage dataset, which has extensive\n2D labeling. Our approach integrates multi-spectral neural attenuation field\nreconstruction with Linear pushbroom (LPB) camera model pose optimization,\nenhancing rendering consistency for novel views through color coding network.\nOur method aims to improve generalization within the security baggage\ninspection domain, where generalization is particularly challenging.\n", "date": "2024-12-04", "categories": []}
{"id": "2412.01971", "title": "Learning a Filtered Backprojection Reconstruction Method for\n  Photoacoustic Computed Tomography with Hemispherical Measurement Geometries", "abstract": "  In certain three-dimensional (3D) applications of photoacoustic computed\ntomography (PACT), including \\textit{in vivo} breast imaging, hemispherical\nmeasurement apertures that enclose the object within their convex hull are\nemployed for data acquisition. Data acquired with such measurement geometries\nare referred to as \\textit{half-scan} data, as only half of a complete\nspherical measurement aperture is employed. Although previous studies have\ndemonstrated that half-scan data can uniquely and stably reconstruct the\nsought-after object, no closed-form reconstruction formula for use with\nhalf-scan data has been reported. To address this, a semi-analytic\nreconstruction method in the form of filtered backprojection (FBP), referred to\nas the half-scan FBP method, is developed in this work. Because the explicit\nform of the filtering operation in the half-scan FBP method is not currently\nknown, a learning-based method is proposed to approximate it. The proposed\nmethod is systematically investigated by use of virtual imaging studies of 3D\nbreast PACT that employ ensembles of numerical breast phantoms and a\nphysics-based model of the data acquisition process. The method is subsequently\napplied to experimental data acquired in an \\textit{in vivo} breast PACT study.\nThe results confirm that the half-scan FBP method can accurately reconstruct 3D\nimages from half-scan data. Importantly, because the sought-after inverse\nmapping is well-posed, the reconstruction method remains accurate even when\napplied to data that differ considerably from those employed to learn the\nfiltering operation.\n", "date": "2024-12-02", "categories": []}
{"id": "2412.02548", "title": "Plug-and-Play Half-Quadratic Splitting for Ptychography", "abstract": "  Ptychography is a coherent diffraction imaging method that uses phase\nretrieval techniques to reconstruct complex-valued images. It achieves this by\nsequentially illuminating overlapping regions of a sample with a coherent beam\nand recording the diffraction pattern. Although this addresses traditional\nimaging system challenges, it is computationally intensive and highly sensitive\nto noise, especially with reduced illumination overlap. Data-driven\nregularisation techniques have been applied in phase retrieval to improve\nreconstruction quality. In particular, plug-and-play (PnP) offers flexibility\nby integrating data-driven denoisers as implicit priors. In this work, we\npropose a half-quadratic splitting framework for using PnP and other\ndata-driven priors for ptychography. We evaluate our method both on natural\nimages and real test objects to validate its effectiveness for ptychographic\nimage reconstruction.\n", "date": "2024-12-03", "categories": []}
{"id": "2410.23178", "title": "Uncertainty quantification for fast reconstruction methods using\n  augmented equivariant bootstrap: Application to radio interferometry", "abstract": "  The advent of next-generation radio interferometers like the Square Kilometer\nArray promises to revolutionise our radio astronomy observational capabilities.\nThe unprecedented volume of data these devices generate requires fast and\naccurate image reconstruction algorithms to solve the ill-posed radio\ninterferometric imaging problem. Most state-of-the-art reconstruction methods\nlack trustworthy and scalable uncertainty quantification, which is critical for\nthe rigorous scientific interpretation of radio observations. We propose an\nunsupervised technique based on a conformalized version of a radio-augmented\nequivariant bootstrapping method, which allows us to quantify uncertainties for\nfast reconstruction methods. Noticeably, we rely on reconstructions from\nultra-fast unrolled algorithms. The proposed method brings more reliable\nuncertainty estimations to our problem than existing alternatives.\n", "date": "2024-10-30", "categories": []}
{"id": "2412.00511", "title": "Energy-Based Prior Latent Space Diffusion model for Reconstruction of\n  Lumbar Vertebrae from Thick Slice MRI", "abstract": "  Lumbar spine problems are ubiquitous, motivating research into targeted\nimaging for treatment planning and guided interventions. While high resolution\nand high contrast CT has been the modality of choice, MRI can capture both bone\nand soft tissue without the ionizing radiation of CT albeit longer acquisition\ntime. The critical trade-off between contrast quality and acquisition time has\nmotivated 'thick slice MRI', which prioritises faster imaging with high\nin-plane resolution but variable contrast and low through-plane resolution. We\ninvestigate a recently developed post-acquisition pipeline which segments\nvertebrae from thick-slice acquisitions and uses a variational autoencoder to\nenhance quality after an initial 3D reconstruction. We instead propose a latent\nspace diffusion energy-based prior to leverage diffusion models, which exhibit\nhigh-quality image generation. Crucially, we mitigate their high computational\ncost and low sample efficiency by learning an energy-based latent\nrepresentation to perform the diffusion processes. Our resulting method\noutperforms existing approaches across metrics including Dice and VS scores,\nand more faithfully captures 3D features.\n", "date": "2024-11-30", "categories": []}
{"id": "2406.08439", "title": "Coherent Optical Modems for Full-Wavefield Lidar", "abstract": "  The advent of the digital age has driven the development of coherent optical\nmodems--devices that modulate the amplitude and phase of light in multiple\npolarization states. These modems transmit data through fiber optic cables that\nare thousands of kilometers in length at data rates exceeding one terabit per\nsecond. This remarkable technology is made possible through near-THz-rate\nprogrammable control and sensing of the full optical wavefield. While coherent\noptical modems form the backbone of telecommunications networks around the\nworld, their extraordinary capabilities also provide unique opportunities for\nimaging. Here, we repurpose off-the-shelf coherent optical modems to introduce\nfull-wavefield lidar: a type of random modulation continuous wave lidar that\nsimultaneously measures depth, axial velocity, and polarization. We demonstrate\nthis modality by combining a 74 GHz-bandwidth coherent optical modem with\nfree-space coupling optics and scanning mirrors. We develop a time-resolved\nimage formation model for this system and formulate a maximum-likelihood\nreconstruction algorithm to recover depth, velocity, and polarization\ninformation at each scene point from the modem's raw transmitted and received\nsymbols. Compared to existing lidars, full-wavefield lidar promises improved\nmm-scale ranging accuracy from brief, microsecond exposure times, reliable\nvelocimetry, and robustness to interference from ambient light or other lidar\nsignals.\n", "date": "2024-06-12", "categories": []}
{"id": "2410.02572", "title": "Combining Pre- and Post-Demosaicking Noise Removal for RAW Video", "abstract": "  Denoising is one of the fundamental steps of the processing pipeline that\nconverts data captured by a camera sensor into a display-ready image or video.\nIt is generally performed early in the pipeline, usually before demosaicking,\nalthough studies swapping their order or even conducting them jointly have been\nproposed. With the advent of deep learning, the quality of denoising algorithms\nhas steadily increased. Even so, modern neural networks still have a hard time\nadapting to new noise levels and scenes, which is indispensable for real-world\napplications. With those in mind, we propose a self-similarity-based denoising\nscheme that weights both a pre- and a post-demosaicking denoiser for\nBayer-patterned CFA video data. We show that a balance between the two leads to\nbetter image quality, and we empirically find that higher noise levels benefit\nfrom a higher influence pre-demosaicking. We also integrate temporal trajectory\nprefiltering steps before each denoiser, which further improve texture\nreconstruction. The proposed method only requires an estimation of the noise\nmodel at the sensor, accurately adapts to any noise level, and is competitive\nwith the state of the art, making it suitable for real-world videography.\n", "date": "2024-10-03", "categories": []}
{"id": "2411.19593", "title": "Self-Supervised Denoiser Framework", "abstract": "  Reconstructing images using Computed Tomography (CT) in an industrial context\nleads to specific challenges that differ from those encountered in other areas,\nsuch as clinical CT. Indeed, non-destructive testing with industrial CT will\noften involve scanning multiple similar objects while maintaining high\nthroughput, requiring short scanning times, which is not a relevant concern in\nclinical CT. Under-sampling the tomographic data (sinograms) is a natural way\nto reduce the scanning time at the cost of image quality since the latter\ndepends on the number of measurements. In such a scenario, post-processing\ntechniques are required to compensate for the image artifacts induced by the\nsinogram sparsity. We introduce the Self-supervised Denoiser Framework (SDF), a\nself-supervised training method that leverages pre-training on highly sampled\nsinogram data to enhance the quality of images reconstructed from undersampled\nsinogram data. The main contribution of SDF is that it proposes to train an\nimage denoiser in the sinogram space by setting the learning task as the\nprediction of one sinogram subset from another. As such, it does not require\nground-truth image data, leverages the abundant data modality in CT, the\nsinogram, and can drastically enhance the quality of images reconstructed from\na fraction of the measurements. We demonstrate that SDF produces better image\nquality, in terms of peak signal-to-noise ratio, than other analytical and\nself-supervised frameworks in both 2D fan-beam or 3D cone-beam CT settings.\nMoreover, we show that the enhancement provided by SDF carries over when\nfine-tuning the image denoiser on a few examples, making it a suitable\npre-training technique in a context where there is little high-quality image\ndata. Our results are established on experimental datasets, making SDF a strong\ncandidate for being the building block of foundational image-enhancement models\nin CT.\n", "date": "2024-11-29", "categories": []}
{"id": "2411.18189", "title": "Towards Lensless Image Deblurring with Prior-Embedded Implicit Neural\n  Representations in the Low-Data Regime", "abstract": "  The field of computational imaging has witnessed a promising paradigm shift\nwith the emergence of untrained neural networks, offering novel solutions to\ninverse computational imaging problems. While existing techniques have\ndemonstrated impressive results, they often operate either in the high-data\nregime, leveraging Generative Adversarial Networks (GANs) as image priors, or\nthrough untrained iterative reconstruction in a data-agnostic manner. This\npaper delves into lensless image reconstruction, a subset of computational\nimaging that replaces traditional lenses with computation, enabling the\ndevelopment of ultra-thin and lightweight imaging systems. To the best of our\nknowledge, we are the first to leverage implicit neural representations for\nlensless image deblurring, achieving reconstructions without the requirement of\nprior training. We perform prior-embedded untrained iterative optimization to\nenhance reconstruction performance and speed up convergence, effectively\nbridging the gap between the no-data and high-data regimes. Through a thorough\ncomparative analysis encompassing various untrained and low-shot methods,\nincluding under-parameterized non-convolutional methods and domain-restricted\nlow-shot methods, we showcase the superior performance of our approach by a\nsignificant margin.\n", "date": "2024-11-27", "categories": []}
{"id": "2411.18588", "title": "Hierarchical Information Flow for Generalized Efficient Image\n  Restoration", "abstract": "  While vision transformers show promise in numerous image restoration (IR)\ntasks, the challenge remains in efficiently generalizing and scaling up a model\nfor multiple IR tasks. To strike a balance between efficiency and model\ncapacity for a generalized transformer-based IR method, we propose a\nhierarchical information flow mechanism for image restoration, dubbed Hi-IR,\nwhich progressively propagates information among pixels in a bottom-up manner.\nHi-IR constructs a hierarchical information tree representing the degraded\nimage across three levels. Each level encapsulates different types of\ninformation, with higher levels encompassing broader objects and concepts and\nlower levels focusing on local details. Moreover, the hierarchical tree\narchitecture removes long-range self-attention, improves the computational\nefficiency and memory utilization, thus preparing it for effective model\nscaling. Based on that, we explore model scaling to improve our method's\ncapabilities, which is expected to positively impact IR in large-scale training\nsettings. Extensive experimental results show that Hi-IR achieves\nstate-of-the-art performance in seven common image restoration tasks, affirming\nits effectiveness and generalizability.\n", "date": "2024-11-27", "categories": []}
{"id": "2408.11758", "title": "MambaCSR: Dual-Interleaved Scanning for Compressed Image\n  Super-Resolution With SSMs", "abstract": "  We present MambaCSR, a simple but effective framework based on Mamba for the\nchallenging compressed image super-resolution (CSR) task. Particularly, the\nscanning strategies of Mamba are crucial for effective contextual knowledge\nmodeling in the restoration process despite it relying on selective state space\nmodeling for all tokens. In this work, we propose an efficient dual-interleaved\nscanning paradigm (DIS) for CSR, which is composed of two scanning strategies:\n(i) hierarchical interleaved scanning is designed to comprehensively capture\nand utilize the most potential contextual information within an image by\nsimultaneously taking advantage of the local window-based and sequential\nscanning methods; (ii) horizontal-to-vertical interleaved scanning is proposed\nto reduce the computational cost by leaving the redundancy between the scanning\nof different directions. To overcome the non-uniform compression artifacts, we\nalso propose position-aligned cross-scale scanning to model multi-scale\ncontextual information. Experimental results on multiple benchmarks have shown\nthe great performance of our MambaCSR in the compressed image super-resolution\ntask. The code will be soon available\nin~\\textcolor{magenta}{\\url{https://github.com/renyulin-f/MambaCSR}}.\n", "date": "2024-08-21", "categories": []}
{"id": "2409.14722", "title": "Neural refractive index field: Unlocking the Potential of\n  Background-oriented Schlieren Tomography in Volumetric Flow Visualization", "abstract": "  Background-oriented Schlieren tomography (BOST) is a prevalent method for\nvisualizing intricate turbulent flows, valued for its ease of implementation\nand capacity to capture three-dimensional distributions of a multitude of flow\nparameters. However, the voxel-based meshing scheme leads to significant\nchallenges, such as inadequate spatial resolution, substantial discretization\nerrors, poor noise immunity, and excessive computational costs. This work\npresents an innovative reconstruction approach termed neural refractive index\nfield (NeRIF) which implicitly represents the flow field with a neural network,\nwhich is trained with tailored strategies. Both numerical simulations and\nexperimental demonstrations on turbulent Bunsen flames suggest that our\napproach can significantly improve the reconstruction accuracy and spatial\nresolution while concurrently reducing computational expenses. Although\nshowcased in the context of background-oriented schlieren tomography here, the\nkey idea embedded in the NeRIF can be readily adapted to various other\ntomographic modalities including tomographic absorption spectroscopy and\ntomographic particle imaging velocimetry, broadening its potential impact\nacross different domains of flow visualization and analysis.\n", "date": "2024-09-23", "categories": []}
{"id": "2411.17139", "title": "Neural-Network-Enhanced Metalens Camera for High-Definition, Dynamic\n  Imaging in the Long-Wave Infrared Spectrum", "abstract": "  To provide a lightweight and cost-effective solution for the long-wave\ninfrared imaging using a singlet, we develop a camera by integrating a\nHigh-Frequency-Enhancing Cycle-GAN neural network into a metalens imaging\nsystem. The High-Frequency-Enhancing Cycle-GAN improves the quality of the\noriginal metalens images by addressing inherent frequency loss introduced by\nthe metalens. In addition to the bidirectional cyclic generative adversarial\nnetwork, it incorporates a high-frequency adversarial learning module. This\nmodule utilizes wavelet transform to extract high-frequency components, and\nthen establishes a high-frequency feedback loop. It enables the generator to\nenhance the camera outputs by integrating adversarial feedback from the\nhigh-frequency discriminator. This ensures that the generator adheres to the\nconstraints imposed by the high-frequency adversarial loss, thereby effectively\nrecovering the camera's frequency loss. This recovery guarantees high-fidelity\nimage output from the camera, facilitating smooth video production. Our camera\nis capable of achieving dynamic imaging at 125 frames per second with an End\nPoint Error value of 12.58. We also achieve 0.42 for Fr\\'echet Inception\nDistance, 30.62 for Peak Signal to Noise Ratio, and 0.69 for Structural\nSimilarity in the recorded videos.\n", "date": "2024-11-26", "categories": []}
{"id": "2405.17818", "title": "Hyperspectral and multispectral image fusion with arbitrary resolution\n  through self-supervised representations", "abstract": "  The fusion of a low-resolution hyperspectral image (LR-HSI) with a\nhigh-resolution multispectral image (HR-MSI) has emerged as an effective\ntechnique for achieving HSI super-resolution (SR). Previous studies have mainly\nconcentrated on estimating the posterior distribution of the latent\nhigh-resolution hyperspectral image (HR-HSI), leveraging an appropriate image\nprior and likelihood computed from the discrepancy between the latent HSI and\nobserved images. Low rankness stands out for preserving latent HSI\ncharacteristics through matrix factorization among the various priors. However,\nthe primary limitation in previous studies lies in the generalization of a\nfusion model with fixed resolution scales, which necessitates retraining\nwhenever output resolutions are changed. To overcome this limitation, we\npropose a novel continuous low-rank factorization (CLoRF) by integrating two\nneural representations into the matrix factorization, capturing spatial and\nspectral information, respectively. This approach enables us to harness both\nthe low rankness from the matrix factorization and the continuity from neural\nrepresentation in a self-supervised manner.Theoretically, we prove the low-rank\nproperty and Lipschitz continuity in the proposed continuous low-rank\nfactorization. Experimentally, our method significantly surpasses existing\ntechniques and achieves user-desired resolutions without the need for neural\nnetwork retraining. Code is available at\nhttps://github.com/wangting1907/CLoRF-Fusion.\n", "date": "2024-05-28", "categories": []}
{"id": "2411.12471", "title": "SCIGS: 3D Gaussians Splatting from a Snapshot Compressive Image", "abstract": "  Snapshot Compressive Imaging (SCI) offers a possibility for capturing\ninformation in high-speed dynamic scenes, requiring efficient reconstruction\nmethod to recover scene information. Despite promising results, current deep\nlearning-based and NeRF-based reconstruction methods face challenges: 1) deep\nlearning-based reconstruction methods struggle to maintain 3D structural\nconsistency within scenes, and 2) NeRF-based reconstruction methods still face\nlimitations in handling dynamic scenes. To address these challenges, we propose\nSCIGS, a variant of 3DGS, and develop a primitive-level transformation network\nthat utilizes camera pose stamps and Gaussian primitive coordinates as\nembedding vectors. This approach resolves the necessity of camera pose in\nvanilla 3DGS and enhances multi-view 3D structural consistency in dynamic\nscenes by utilizing transformed primitives. Additionally, a high-frequency\nfilter is introduced to eliminate the artifacts generated during the\ntransformation. The proposed SCIGS is the first to reconstruct a 3D explicit\nscene from a single compressed image, extending its application to dynamic 3D\nscenes. Experiments on both static and dynamic scenes demonstrate that SCIGS\nnot only enhances SCI decoding but also outperforms current state-of-the-art\nmethods in reconstructing dynamic 3D scenes from a single compressed image. The\ncode will be made available upon publication.\n", "date": "2024-11-19", "categories": []}
{"id": "2411.16217", "title": "Mixed Degradation Image Restoration via Local Dynamic Optimization and\n  Conditional Embedding", "abstract": "  Multiple-in-one image restoration (IR) has made significant progress, aiming\nto handle all types of single degraded image restoration with a single model.\nHowever, in real-world scenarios, images often suffer from combinations of\nmultiple degradation factors. Existing multiple-in-one IR models encounter\nchallenges related to degradation diversity and prompt singularity when\naddressing this issue. In this paper, we propose a novel multiple-in-one IR\nmodel that can effectively restore images with both single and mixed\ndegradations. To address degradation diversity, we design a Local Dynamic\nOptimization (LDO) module which dynamically processes degraded areas of varying\ntypes and granularities. To tackle the prompt singularity issue, we develop an\nefficient Conditional Feature Embedding (CFE) module that guides the decoder in\nleveraging degradation-type-related features, significantly improving the\nmodel's performance in mixed degradation restoration scenarios. To validate the\neffectiveness of our model, we introduce a new dataset containing both single\nand mixed degradation elements. Experimental results demonstrate that our\nproposed model achieves state-of-the-art (SOTA) performance not only on mixed\ndegradation tasks but also on classic single-task restoration benchmarks.\n", "date": "2024-11-25", "categories": []}
{"id": "2406.01294", "title": "CE-VAE: Capsule Enhanced Variational AutoEncoder for Underwater Image\n  Enhancement", "abstract": "  Unmanned underwater image analysis for marine monitoring faces two key\nchallenges: (i) degraded image quality due to light attenuation and (ii)\nhardware storage constraints limiting high-resolution image collection.\nExisting methods primarily address image enhancement with approaches that hinge\non storing the full-size input. In contrast, we introduce the Capsule Enhanced\nVariational AutoEncoder (CE-VAE), a novel architecture designed to efficiently\ncompress and enhance degraded underwater images. Our attention-aware image\nencoder can project the input image onto a latent space representation while\nbeing able to run online on a remote device. The only information that needs to\nbe stored on the device or sent to a beacon is a compressed representation.\nThere is a dual-decoder module that performs offline, full-size enhanced image\ngeneration. One branch reconstructs spatial details from the compressed latent\nspace, while the second branch utilizes a capsule-clustering layer to capture\nentity-level structures and complex spatial relationships. This parallel\ndecoding strategy enables the model to balance fine-detail preservation with\ncontext-aware enhancements. CE-VAE achieves state-of-the-art performance in\nunderwater image enhancement on six benchmark datasets, providing up to 3x\nhigher compression efficiency than existing approaches. Code available at\n\\url{https://github.com/iN1k1/ce-vae-underwater-image-enhancement}.\n", "date": "2024-06-03", "categories": []}
{"id": "2406.05153", "title": "Integrating Physics of the Problem into Data-Driven Methods to Enhance\n  Elastic Full-Waveform Inversion with Uncertainty Quantification", "abstract": "  Full-Waveform Inversion (FWI) is a nonlinear iterative seismic imaging\ntechnique that, by reducing the misfit between recorded and predicted seismic\nwaveforms, can produce detailed estimates of subsurface geophysical properties.\nNevertheless, the strong nonlinearity of FWI can trap the optimization in local\nminima. This issue arises due to factors such as improper initial values, the\nabsence of low frequencies in the measurements, noise, and other related\nconsiderations. To address this challenge and with the advent of advanced\nmachine-learning techniques, data-driven methods, such as deep learning, have\nattracted significantly increasing attention in the geophysical community.\nFurthermore, the elastic wave equation should be included in FWI to represent\nelastic effects accurately. The intersection of data-driven techniques and\nelastic scattering theories presents opportunities and challenges. In this\npaper, by using the knowledge of elastic scattering (physics of the problem)\nand integrating it with machine learning techniques, we propose methods for the\nsolution of time-harmonic FWI to enhance accuracy compared to pure data-driven\nand physics-based approaches. Moreover, to address uncertainty quantification,\nby modifying the structure of the Variational Autoencoder, we introduce a\nprobabilistic deep learning method based on the physics of the problem that\nenables us to explore the uncertainties of the solution. According to the\nlimited availability of datasets in this field and to assess the performance\nand accuracy of the proposed methods, we create a comprehensive dataset close\nto reality and conduct a comparative analysis of the presented approaches to\nit.\n", "date": "2024-06-04", "categories": []}
{"id": "2411.14141", "title": "Differentiable SVD based on Moore-Penrose Pseudoinverse for Inverse\n  Imaging Problems", "abstract": "  Low-rank regularization-based deep unrolling networks have achieved\nremarkable success in various inverse imaging problems (IIPs). However, the\nsingular value decomposition (SVD) is non-differentiable when duplicated\nsingular values occur, leading to severe numerical instability during training.\nIn this paper, we propose a differentiable SVD based on the Moore-Penrose\npseudoinverse to address this issue. To the best of our knowledge, this is the\nfirst work to provide a comprehensive analysis of the differentiability of the\ntrivial SVD. Specifically, we show that the non-differentiability of SVD is\nessentially due to an underdetermined system of linear equations arising in the\nderivation process. We utilize the Moore-Penrose pseudoinverse to solve the\nsystem, thereby proposing a differentiable SVD. A numerical stability analysis\nin the context of IIPs is provided. Experimental results in color image\ncompressed sensing and dynamic MRI reconstruction show that our proposed\ndifferentiable SVD can effectively address the numerical instability issue\nwhile ensuring computational precision. Code is available at\nhttps://github.com/yhao-z/SVD-inv.\n", "date": "2024-11-21", "categories": []}
{"id": "2403.07244", "title": "Time-Efficient Light-Field Acquisition Using Coded Aperture and Events", "abstract": "  We propose a computational imaging method for time-efficient light-field\nacquisition that combines a coded aperture with an event-based camera.\nDifferent from the conventional coded-aperture imaging method, our method\napplies a sequence of coding patterns during a single exposure for an image\nframe. The parallax information, which is related to the differences in coding\npatterns, is recorded as events. The image frame and events, all of which are\nmeasured in a single exposure, are jointly used to computationally reconstruct\na light field. We also designed an algorithm pipeline for our method that is\nend-to-end trainable on the basis of deep optics and compatible with real\ncamera hardware. We experimentally showed that our method can achieve more\naccurate reconstruction than several other imaging methods with a single\nexposure. We also developed a hardware prototype with the potential to complete\nthe measurement on the camera within 22 msec and demonstrated that light fields\nfrom real 3-D scenes can be obtained with convincing visual quality. Our\nsoftware and supplementary video are available from our project website.\n", "date": "2024-03-12", "categories": []}
{"id": "2405.19112", "title": "Reconstructing Interpretable Features in Computational Super-Resolution\n  microscopy via Regularized Latent Search", "abstract": "  Supervised deep learning approaches can artificially increase the resolution\nof microscopy images by learning a mapping between two image resolutions or\nmodalities. However, such methods often require a large set of hard-to-get\nlow-res/high-res image pairs and produce synthetic images with a moderate\nincrease in resolution. Conversely, recent methods based on GAN latent search\noffered a drastic increase in resolution without the need of paired images.\nHowever, they offer limited reconstruction of the high-resolution image\ninterpretable features. Here, we propose a robust super-resolution method based\non regularized latent search~(RLS) that offers an actionable balance between\nfidelity to the ground-truth and realism of the recovered image given a\ndistribution prior. The latter allows to split the analysis of a low-resolution\nimage into a computational super-resolution task performed by deep learning\nfollowed by a quantification task performed by a handcrafted algorithm and\nbased on interpretable biological features. This two-step process holds\npotential for various applications such as diagnostics on mobile devices, where\nthe main aim is not to recover the high-resolution details of a specific sample\nbut rather to obtain high-resolution images that preserve explainable and\nquantifiable differences between conditions.\n", "date": "2024-05-29", "categories": []}
{"id": "2411.12450", "title": "Frequency-Aware Guidance for Blind Image Restoration via Diffusion\n  Models", "abstract": "  Blind image restoration remains a significant challenge in low-level vision\ntasks. Recently, denoising diffusion models have shown remarkable performance\nin image synthesis. Guided diffusion models, leveraging the potent generative\npriors of pre-trained models along with a differential guidance loss, have\nachieved promising results in blind image restoration. However, these models\ntypically consider data consistency solely in the spatial domain, often\nresulting in distorted image content. In this paper, we propose a novel\nfrequency-aware guidance loss that can be integrated into various diffusion\nmodels in a plug-and-play manner. Our proposed guidance loss, based on 2D\ndiscrete wavelet transform, simultaneously enforces content consistency in both\nthe spatial and frequency domains. Experimental results demonstrate the\neffectiveness of our method in three blind restoration tasks: blind image\ndeblurring, imaging through turbulence, and blind restoration for multiple\ndegradations. Notably, our method achieves a significant improvement in PSNR\nscore, with a remarkable enhancement of 3.72\\,dB in image deblurring. Moreover,\nour method exhibits superior capability in generating images with rich details\nand reduced distortion, leading to the best visual quality.\n", "date": "2024-11-19", "categories": []}
{"id": "2406.08048", "title": "3D CBCT Challenge 2024: Improved Cone Beam CT Reconstruction using\n  SwinIR-Based Sinogram and Image Enhancement", "abstract": "  In this paper, we present our approach to the 3D CBCT Challenge 2024, a part\nof ICASSP SP Grand Challenges 2024. Improvement in Cone Beam Computed\nTomography (CBCT) reconstruction has been achieved by integrating Swin Image\nRestoration (SwinIR) based sinogram and image enhancement modules. The proposed\nmethodology uses Nesterov Accelerated Gradient Descent (NAG) to solve the least\nsquares (NAG-LS) problem in CT image reconstruction. The integration of\nsinogram and image enhancement modules aims to enhance image clarity and\npreserve fine details, offering a promising solution for both low dose and\nclinical dose CBCT reconstruction. The averaged mean squared error (MSE) over\nthe validation dataset has decreased significantly, in the case of low dose by\none-fifth and clinical dose by one-tenth. Our solution is one of the top 5\napproaches in this challenge.\n", "date": "2024-06-12", "categories": []}
{"id": "2410.15385", "title": "LoRA-IR: Taming Low-Rank Experts for Efficient All-in-One Image\n  Restoration", "abstract": "  Prompt-based all-in-one image restoration (IR) frameworks have achieved\nremarkable performance by incorporating degradation-specific information into\nprompt modules. Nevertheless, handling the complex and diverse degradations\nencountered in real-world scenarios remains a significant challenge. To tackle\nthis, we propose LoRA-IR, a flexible framework that dynamically leverages\ncompact low-rank experts to facilitate efficient all-in-one image restoration.\nSpecifically, LoRA-IR consists of two training stages: degradation-guided\npre-training and parameter-efficient fine-tuning. In the pre-training stage, we\nenhance the pre-trained CLIP model by introducing a simple mechanism that\nscales it to higher resolutions, allowing us to extract robust degradation\nrepresentations that adaptively guide the IR network. In the fine-tuning stage,\nwe refine the pre-trained IR network through low-rank adaptation (LoRA). Built\nupon a Mixture-of-Experts (MoE) architecture, LoRA-IR dynamically integrates\nmultiple low-rank restoration experts through a degradation-guided router. This\ndynamic integration mechanism significantly enhances our model's adaptability\nto diverse and unknown degradations in complex real-world scenarios. Extensive\nexperiments demonstrate that LoRA-IR achieves SOTA performance across 14 IR\ntasks and 29 benchmarks, while maintaining computational efficiency. Code and\npre-trained models will be available at:\nhttps://github.com/shallowdream204/LoRA-IR.\n", "date": "2024-10-20", "categories": []}
{"id": "2411.10708", "title": "AllRestorer: All-in-One Transformer for Image Restoration under\n  Composite Degradations", "abstract": "  Image restoration models often face the simultaneous interaction of multiple\ndegradations in real-world scenarios. Existing approaches typically handle\nsingle or composite degradations based on scene descriptors derived from text\nor image embeddings. However, due to the varying proportions of different\ndegradations within an image, these scene descriptors may not accurately\ndifferentiate between degradations, leading to suboptimal restoration in\npractical applications. To address this issue, we propose a novel\nTransformer-based restoration framework, AllRestorer. In AllRestorer, we enable\nthe model to adaptively consider all image impairments, thereby avoiding errors\nfrom scene descriptor misdirection. Specifically, we introduce an All-in-One\nTransformer Block (AiOTB), which adaptively removes all degradations present in\na given image by modeling the relationships between all degradations and the\nimage embedding in latent space. To accurately address different variations\npotentially present within the same type of degradation and minimize ambiguity,\nAiOTB utilizes a composite scene descriptor consisting of both image and text\nembeddings to define the degradation. Furthermore, AiOTB includes an adaptive\nweight for each degradation, allowing for precise control of the restoration\nintensity. By leveraging AiOTB, AllRestorer avoids misdirection caused by\ninaccurate scene descriptors, achieving a 5.00 dB increase in PSNR compared to\nthe baseline on the CDD-11 dataset.\n", "date": "2024-11-16", "categories": []}
{"id": "2411.10773", "title": "An End-to-End Real-World Camera Imaging Pipeline", "abstract": "  Recent advances in neural camera imaging pipelines have demonstrated notable\nprogress. Nevertheless, the real-world imaging pipeline still faces challenges\nincluding the lack of joint optimization in system components, computational\nredundancies, and optical distortions such as lens shading.In light of this, we\npropose an end-to-end camera imaging pipeline (RealCamNet) to enhance\nreal-world camera imaging performance. Our methodology diverges from\nconventional, fragmented multi-stage image signal processing towards end-to-end\narchitecture. This architecture facilitates joint optimization across the full\npipeline and the restoration of coordinate-biased distortions. RealCamNet is\ndesigned for high-quality conversion from RAW to RGB and compact image\ncompression. Specifically, we deeply analyze coordinate-dependent optical\ndistortions, e.g., vignetting and dark shading, and design a novel\nCoordinate-Aware Distortion Restoration (CADR) module to restore\ncoordinate-biased distortions. Furthermore, we propose a Coordinate-Independent\nMapping Compression (CIMC) module to implement tone mapping and redundant\ninformation compression. Existing datasets suffer from misalignment and overly\nidealized conditions, making them inadequate for training real-world imaging\npipelines. Therefore, we collected a real-world imaging dataset. Experiment\nresults show that RealCamNet achieves the best rate-distortion performance with\nlower inference latency.\n", "date": "2024-11-16", "categories": []}
{"id": "2411.11717", "title": "RAWMamba: Unified sRGB-to-RAW De-rendering With State Space Model", "abstract": "  Recent advancements in sRGB-to-RAW de-rendering have increasingly emphasized\nmetadata-driven approaches to reconstruct RAW data from sRGB images,\nsupplemented by partial RAW information. In image-based de-rendering, metadata\nis commonly obtained through sampling, whereas in video tasks, it is typically\nderived from the initial frame. The distinct metadata requirements necessitate\nspecialized network architectures, leading to architectural incompatibilities\nthat increase deployment complexity. In this paper, we propose RAWMamba, a\nMamba-based unified framework developed for sRGB-to-RAW de-rendering across\nboth image and video domains. The core of RAWMamba is the Unified Metadata\nEmbedding (UME) module, which harmonizes diverse metadata types into a unified\nrepresentation. In detail, a multi-perspective affinity modeling method is\nproposed to promote the extraction of reference information. In addition, we\nintroduce the Local Tone-Aware Mamba (LTA-Mamba) module, which captures\nlong-range dependencies to enable effective global propagation of metadata.\nExperimental results demonstrate that the proposed RAWMamba achieves\nstate-of-the-art performance, yielding high-quality RAW data reconstruction.\n", "date": "2024-11-18", "categories": []}
{"id": "2410.14994", "title": "Quanta Video Restoration", "abstract": "  The proliferation of single-photon image sensors has opened the door to a\nplethora of high-speed and low-light imaging applications. However, data\ncollected by these sensors are often 1-bit or few-bit, and corrupted by noise\nand strong motion. Conventional video restoration methods are not designed to\nhandle this situation, while specialized quanta burst algorithms have limited\nperformance when the number of input frames is low. In this paper, we introduce\nQuanta Video Restoration (QUIVER), an end-to-end trainable network built on the\ncore ideas of classical quanta restoration methods, i.e., pre-filtering, flow\nestimation, fusion, and refinement. We also collect and publish I2-2000FPS, a\nhigh-speed video dataset with the highest temporal resolution of 2000\nframes-per-second, for training and testing. On simulated and real data, QUIVER\noutperforms existing quanta restoration methods by a significant margin. Code\nand dataset available at\nhttps://github.com/chennuriprateek/Quanta_Video_Restoration-QUIVER-\n", "date": "2024-10-19", "categories": []}
{"id": "2411.10189", "title": "NeISF++: Neural Incident Stokes Field for Polarized Inverse Rendering of\n  Conductors and Dielectrics", "abstract": "  Recent inverse rendering methods have greatly improved shape, material, and\nillumination reconstruction by utilizing polarization cues. However, existing\nmethods only support dielectrics, ignoring conductors that are found everywhere\nin life. Since conductors and dielectrics have different reflection properties,\nusing previous conductor methods will lead to obvious errors. In addition,\nconductors are glossy, which may cause strong specular reflection and is hard\nto reconstruct. To solve the above issues, we propose NeISF++, an inverse\nrendering pipeline that supports conductors and dielectrics. The key ingredient\nfor our proposal is a general pBRDF that describes both conductors and\ndielectrics. As for the strong specular reflection problem, we propose a novel\ngeometry initialization method using DoLP images. This physical cue is\ninvariant to intensities and thus robust to strong specular reflections.\nExperimental results on our synthetic and real datasets show that our method\nsurpasses the existing polarized inverse rendering methods for geometry and\nmaterial decomposition as well as downstream tasks like relighting.\n", "date": "2024-11-15", "categories": []}
{"id": "2411.10200", "title": "Block based Adaptive Compressive Sensing with Sampling Rate Control", "abstract": "  Compressive sensing (CS), acquiring and reconstructing signals below the\nNyquist rate, has great potential in image and video acquisition to exploit\ndata redundancy and greatly reduce the amount of sampled data. To further\nreduce the sampled data while keeping the video quality, this paper explores\nthe temporal redundancy in video CS and proposes a block based adaptive\ncompressive sensing framework with a sampling rate (SR) control strategy. To\navoid redundant compression of non-moving regions, we first incorporate moving\nblock detection between consecutive frames, and only transmit the measurements\nof moving blocks. The non-moving regions are reconstructed from the previous\nframe. In addition, we propose a block storage system and a dynamic threshold\nto achieve adaptive SR allocation to each frame based on the area of moving\nregions and target SR for controlling the average SR within the target SR.\nFinally, to reduce blocking artifacts and improve reconstruction quality, we\nadopt a cooperative reconstruction of the moving and non-moving blocks by\nreferring to the measurements of the non-moving blocks from the previous frame.\nExtensive experiments have demonstrated that this work is able to control SR\nand obtain better performance than existing works.\n", "date": "2024-11-15", "categories": []}
{"id": "2411.08995", "title": "Computed tomography using meta-optics", "abstract": "  Computer vision tasks require processing large amounts of data to perform\nimage classification, segmentation, and feature extraction. Optical\npreprocessors can potentially reduce the number of floating point operations\nrequired by computer vision tasks, enabling low-power and low-latency\noperation. However, existing optical preprocessors are mostly learned and hence\nstrongly depend on the training data, and thus lack universal applicability. In\nthis paper, we present a metaoptic imager, which implements the Radon transform\nobviating the need for training the optics. High quality image reconstruction\nwith a large compression ratio of 0.6% is presented through the use of the\nSimultaneous Algebraic Reconstruction Technique. Image classification with 90%\naccuracy is presented on an experimentally measured Radon dataset through\nneural network trained on digitally transformed images.\n", "date": "2024-11-13", "categories": []}
{"id": "2411.09133", "title": "Computational metaoptics for imaging", "abstract": "  Metasurfaces -- ultrathin structures composed of subwavelength optical\nelements -- have revolutionized light manipulation by enabling precise control\nover electromagnetic waves' amplitude, phase, polarization, and spectral\nproperties. Concurrently, computational imaging leverages algorithms to\nreconstruct images from optically processed signals, overcoming limitations of\ntraditional imaging systems. This review explores the synergistic integration\nof metaoptics and computational imaging, \"computational metaoptics,\" which\ncombines the physical wavefront shaping ability of metasurfaces with advanced\ncomputational algorithms to enhance imaging performance beyond conventional\nlimits. We discuss how computational metaoptics addresses the inherent\nlimitations of single-layer metasurfaces in achieving multifunctionality\nwithout compromising efficiency. By treating metasurfaces as physical\npreconditioners and co-designing them with reconstruction algorithms through\nend-to-end (inverse) design, it is possible to jointly optimize the optical\nhardware and computational software. This holistic approach allows for the\nautomatic discovery of optimal metasurface designs and reconstruction methods\nthat significantly improve imaging capabilities. Advanced applications enabled\nby computational metaoptics are highlighted, including phase imaging and\nquantum state measurement, which benefit from the metasurfaces' ability to\nmanipulate complex light fields and the computational algorithms' capacity to\nreconstruct high-dimensional information. We also examine performance\nevaluation challenges, emphasizing the need for new metrics that account for\nthe combined optical and computational nature of these systems. Finally, we\nidentify new frontiers in computational metaoptics which point toward a future\nwhere computational metaoptics may play a central role in advancing imaging\nscience and technology.\n", "date": "2024-11-14", "categories": []}
{"id": "2411.08291", "title": "Restoration algorithms and system performance evaluation for active\n  imagers", "abstract": "  This paper deals with two fields related to active imaging system. First, we\nbegin to explore image processing algorithms to restore the artefacts like\nspeckle, scintillation and image dancing caused by atmospheric turbulence.\nNext, we examine how to evaluate the performance of this kind of systems. To do\nthis task, we propose a modified version of the german TRM3 metric which\npermits to get MTF-like measures. We use the database acquired during NATO-TG40\nfield trials to make our tests.\n", "date": "2024-11-13", "categories": []}
{"id": "2403.06443", "title": "Temporal-Mapping Photography for Event Cameras", "abstract": "  Event cameras, or Dynamic Vision Sensors (DVS) are novel neuromorphic sensors\nthat capture brightness changes as a continuous stream of \"events\" rather than\ntraditional intensity frames. Converting sparse events to dense intensity\nframes faithfully has long been an ill-posed problem. Previous methods have\nprimarily focused on converting events to video in dynamic scenes or with a\nmoving camera. In this paper, for the first time, we realize events to dense\nintensity image conversion using a stationary event camera in static scenes\nwith a transmittance adjustment device for brightness modulation. Different\nfrom traditional methods that mainly rely on event integration, the proposed\nEvent-Based Temporal Mapping Photography (EvTemMap) measures the time of event\nemitting for each pixel. Then, the resulting Temporal Matrix is converted to an\nintensity frame with a temporal mapping neural network. At the hardware level,\nthe proposed EvTemMap is implemented by combining a transmittance adjustment\ndevice with a DVS, named Adjustable Transmittance Dynamic Vision Sensor\n(AT-DVS). Additionally, we collected TemMat dataset under various conditions\nincluding low-light and high dynamic range scenes. The experimental results\nshowcase the high dynamic range, fine-grained details, and high-grayscale\nresolution of the proposed EvTemMap. The code and dataset are available in\nhttps://github.com/YuHanBaozju/EvTemMap\n", "date": "2024-03-11", "categories": []}
{"id": "2411.07578", "title": "Atmospheric turbulence restoration by diffeomorphic image registration\n  and blind deconvolution", "abstract": "  A novel approach is presented in this paper to improve images which are\naltered by atmospheric turbulence. Two new algorithms are presented based on\ntwo combinations of a blind deconvolution block, an elastic registration block\nand a temporal filter block. The algorithms are tested on real images acquired\nin the desert in New Mexico by the NATO RTG40 group.\n", "date": "2024-11-12", "categories": []}
{"id": "2311.13682", "title": "Single-Shot Plug-and-Play Methods for Inverse Problems", "abstract": "  The utilisation of Plug-and-Play (PnP) priors in inverse problems has become\nincreasingly prominent in recent years. This preference is based on the\nmathematical equivalence between the general proximal operator and the\nregularised denoiser, facilitating the adaptation of various off-the-shelf\ndenoiser priors to a wide range of inverse problems. However, existing PnP\nmodels predominantly rely on pre-trained denoisers using large datasets. In\nthis work, we introduce Single-Shot PnP methods (SS-PnP), shifting the focus to\nsolving inverse problems with minimal data. First, we integrate Single-Shot\nproximal denoisers into iterative methods, enabling training with single\ninstances. Second, we propose implicit neural priors based on a novel function\nthat preserves relevant frequencies to capture fine details while avoiding the\nissue of vanishing gradients. We demonstrate, through extensive numerical and\nvisual experiments, that our method leads to better approximations.\n", "date": "2023-11-22", "categories": []}
{"id": "2406.02785", "title": "Event-horizon-scale Imaging of M87* under Different Assumptions via Deep\n  Generative Image Priors", "abstract": "  Reconstructing images from the Event Horizon Telescope (EHT) observations of\nM87*, the supermassive black hole at the center of the galaxy M87, depends on a\nprior to impose desired image statistics. However, given the impossibility of\ndirectly observing black holes, there is no clear choice for a prior. We\npresent a framework for flexibly designing a range of priors, each bringing\ndifferent biases to the image reconstruction. These priors can be weak (e.g.,\nimpose only basic natural-image statistics) or strong (e.g., impose assumptions\nof black-hole structure). Our framework uses Bayesian inference with\nscore-based priors, which are data-driven priors arising from a deep generative\nmodel that can learn complicated image distributions. Using our Bayesian\nimaging approach with sophisticated data-driven priors, we can assess how\nvisual features and uncertainty of reconstructed images change depending on the\nprior. In addition to simulated data, we image the real EHT M87* data and\ndiscuss how recovered features are influenced by the choice of prior.\n", "date": "2024-06-04", "categories": []}
{"id": "2411.05883", "title": "Benchmarking 3D multi-coil NC-PDNet MRI reconstruction", "abstract": "  Deep learning has shown great promise for MRI reconstruction from\nundersampled data, yet there is a lack of research on validating its\nperformance in 3D parallel imaging acquisitions with non-Cartesian\nundersampling. In addition, the artifacts and the resulting image quality\ndepend on the under-sampling pattern. To address this uncharted territory, we\nextend the Non-Cartesian Primal-Dual Network (NC-PDNet), a state-of-the-art\nunrolled neural network, to a 3D multi-coil setting. We evaluated the impact of\nchannel-specific versus channel-agnostic training configurations and examined\nthe effect of coil compression. Finally, we benchmark four distinct\nnon-Cartesian undersampling patterns, with an acceleration factor of six, using\nthe publicly available Calgary-Campinas dataset. Our results show that NC-PDNet\ntrained on compressed data with varying input channel numbers achieves an\naverage PSNR of 42.98 dB for 1 mm isotropic 32 channel whole-brain 3D\nreconstruction. With an inference time of 4.95sec and a GPU memory usage of\n5.49 GB, our approach demonstrates significant potential for clinical research\napplication.\n", "date": "2024-11-08", "categories": []}
{"id": "2411.06181", "title": "Epi-NAF: Enhancing Neural Attenuation Fields for Limited-Angle CT with\n  Epipolar Consistency Conditions", "abstract": "  Neural field methods, initially successful in the inverse rendering domain,\nhave recently been extended to CT reconstruction, marking a paradigm shift from\ntraditional techniques. While these approaches deliver state-of-the-art results\nin sparse-view CT reconstruction, they struggle in limited-angle settings,\nwhere input projections are captured over a restricted angle range. We present\na novel loss term based on consistency conditions between corresponding\nepipolar lines in X-ray projection images, aimed at regularizing neural\nattenuation field optimization. By enforcing these consistency conditions, our\napproach, Epi-NAF, propagates supervision from input views within the\nlimited-angle range to predicted projections over the full cone-beam CT range.\nThis loss results in both qualitative and quantitative improvements in\nreconstruction compared to baseline methods.\n", "date": "2024-11-09", "categories": []}
{"id": "2403.10012", "title": "Representing Domain-Mixing Optical Degradation for Real-World\n  Computational Aberration Correction via Vector Quantization", "abstract": "  Relying on paired synthetic data, existing learning-based Computational\nAberration Correction (CAC) methods are confronted with the intricate and\nmultifaceted synthetic-to-real domain gap, which leads to suboptimal\nperformance in real-world applications. In this paper, in contrast to improving\nthe simulation pipeline, we deliver a novel insight into real-world CAC from\nthe perspective of Unsupervised Domain Adaptation (UDA). By incorporating\nreadily accessible unpaired real-world data into training, we formalize the\nDomain Adaptive CAC (DACAC) task, and then introduce a comprehensive Real-world\naberrated images (Realab) dataset to benchmark it. The setup task presents a\nformidable challenge due to the intricacy of understanding the target optical\ndegradation domain. To this intent, we propose a novel Quantized Domain-Mixing\nRepresentation (QDMR) framework as a potent solution to the issue. Centering\naround representing and quantizing the optical degradation which is consistent\nacross different images, QDMR adapts the CAC model to the target domain from\nthree key aspects: (1) reconstructing aberrated images of both domains by a\nVQGAN to learn a Domain-Mixing Codebook (DMC) characterizing the optical\ndegradation; (2) modulating the deep features in CAC model with DMC to transfer\nthe target domain knowledge; and (3) leveraging the trained VQGAN to generate\npseudo target aberrated images from the source ones for convincing target\ndomain supervision. Extensive experiments on both synthetic and real-world\nbenchmarks reveal that the models with QDMR consistently surpass the\ncompetitive methods in mitigating the synthetic-to-real gap, which produces\nvisually pleasant real-world CAC results with fewer artifacts. Codes and\ndatasets are made publicly available at https://github.com/zju-jiangqi/QDMR.\n", "date": "2024-03-15", "categories": []}
{"id": "2405.18782", "title": "Principled Probabilistic Imaging using Diffusion Models as Plug-and-Play\n  Priors", "abstract": "  Diffusion models (DMs) have recently shown outstanding capabilities in\nmodeling complex image distributions, making them expressive image priors for\nsolving Bayesian inverse problems. However, most existing DM-based methods rely\non approximations in the generative process to be generic to different inverse\nproblems, leading to inaccurate sample distributions that deviate from the\ntarget posterior defined within the Bayesian framework. To harness the\ngenerative power of DMs while avoiding such approximations, we propose a Markov\nchain Monte Carlo algorithm that performs posterior sampling for general\ninverse problems by reducing it to sampling the posterior of a Gaussian\ndenoising problem. Crucially, we leverage a general DM formulation as a unified\ninterface that allows for rigorously solving the denoising problem with a range\nof state-of-the-art DMs. We demonstrate the effectiveness of the proposed\nmethod on six inverse problems (three linear and three nonlinear), including a\nreal-world black hole imaging problem. Experimental results indicate that our\nproposed method offers more accurate reconstructions and posterior estimation\ncompared to existing DM-based imaging inverse methods.\n", "date": "2024-05-29", "categories": []}
{"id": "2407.06633", "title": "Variational Zero-shot Multispectral Pansharpening", "abstract": "  Pansharpening aims to generate a high spatial resolution multispectral image\n(HRMS) by fusing a low spatial resolution multispectral image (LRMS) and a\npanchromatic image (PAN). The most challenging issue for this task is that only\nthe to-be-fused LRMS and PAN are available, and the existing deep\nlearning-based methods are unsuitable since they rely on many training pairs.\nTraditional variational optimization (VO) based methods are well-suited for\naddressing such a problem. They focus on carefully designing explicit fusion\nrules as well as regularizations for an optimization problem, which are based\non the researcher's discovery of the image relationships and image structures.\nUnlike previous VO-based methods, in this work, we explore such complex\nrelationships by a parameterized term rather than a manually designed one.\nSpecifically, we propose a zero-shot pansharpening method by introducing a\nneural network into the optimization objective. This network estimates a\nrepresentation component of HRMS, which mainly describes the relationship\nbetween HRMS and PAN. In this way, the network achieves a similar goal to the\nso-called deep image prior because it implicitly regulates the relationship\nbetween the HRMS and PAN images through its inherent structure. We directly\nminimize this optimization objective via network parameters and the expected\nHRMS image through iterative updating. Extensive experiments on various\nbenchmark datasets demonstrate that our proposed method can achieve better\nperformance compared with other state-of-the-art methods. The codes are\navailable at https://github.com/xyrui/PSDip.\n", "date": "2024-07-09", "categories": []}
{"id": "2410.23247", "title": "bit2bit: 1-bit quanta video reconstruction via self-supervised photon\n  prediction", "abstract": "  Quanta image sensors, such as SPAD arrays, are an emerging sensor technology,\nproducing 1-bit arrays representing photon detection events over exposures as\nshort as a few nanoseconds. In practice, raw data are post-processed using\nheavy spatiotemporal binning to create more useful and interpretable images at\nthe cost of degrading spatiotemporal resolution. In this work, we propose\nbit2bit, a new method for reconstructing high-quality image stacks at the\noriginal spatiotemporal resolution from sparse binary quanta image data.\nInspired by recent work on Poisson denoising, we developed an algorithm that\ncreates a dense image sequence from sparse binary photon data by predicting the\nphoton arrival location probability distribution. However, due to the binary\nnature of the data, we show that the assumption of a Poisson distribution is\ninadequate. Instead, we model the process with a Bernoulli lattice process from\nthe truncated Poisson. This leads to the proposal of a novel self-supervised\nsolution based on a masked loss function. We evaluate our method using both\nsimulated and real data. On simulated data from a conventional video, we\nachieve 34.35 mean PSNR with extremely photon-sparse binary input (<0.06\nphotons per pixel per frame). We also present a novel dataset containing a wide\nrange of real SPAD high-speed videos under various challenging imaging\nconditions. The scenes cover strong/weak ambient light, strong motion,\nultra-fast events, etc., which will be made available to the community, on\nwhich we demonstrate the promise of our approach. Both reconstruction quality\nand throughput substantially surpass the state-of-the-art methods (e.g., Quanta\nBurst Photography (QBP)). Our approach significantly enhances the visualization\nand usability of the data, enabling the application of existing analysis\ntechniques.\n", "date": "2024-10-30", "categories": []}
{"id": "2411.04457", "title": "Efficient single image non-uniformity correction algorithm", "abstract": "  This paper introduces a new way to correct the non-uniformity (NU) in\nuncooled infrared-type images. The main defect of these uncooled images is the\nlack of a column (resp. line) time-dependent cross-calibration, resulting in a\nstrong column (resp. line) and time dependent noise. This problem can be\nconsidered as a 1D flicker of the columns inside each frame. Thus, classic\nmovie deflickering algorithms can be adapted, to equalize the columns (resp.\nthe lines). The proposed method therefore applies to the series formed by the\ncolumns of an infrared image a movie deflickering algorithm. The obtained\nsingle image method works on static images, and therefore requires no\nregistration, no camera motion compensation, and no closed aperture sensor\nequalization. Thus, the method has only one camera dependent parameter, and is\nlandscape independent. This simple method will be compared to a state of the\nart total variation single image correction on raw real and simulated images.\nThe method is real time, requiring only two operations per pixel. It involves\nno test-pattern calibration and produces no \"ghost artifacts\".\n", "date": "2024-11-07", "categories": []}
{"id": "2411.04810", "title": "GANESH: Generalizable NeRF for Lensless Imaging", "abstract": "  Lensless imaging offers a significant opportunity to develop ultra-compact\ncameras by removing the conventional bulky lens system. However, without a\nfocusing element, the sensor's output is no longer a direct image but a complex\nmultiplexed scene representation. Traditional methods have attempted to address\nthis challenge by employing learnable inversions and refinement models, but\nthese methods are primarily designed for 2D reconstruction and do not\ngeneralize well to 3D reconstruction. We introduce GANESH, a novel framework\ndesigned to enable simultaneous refinement and novel view synthesis from\nmulti-view lensless images. Unlike existing methods that require scene-specific\ntraining, our approach supports on-the-fly inference without retraining on each\nscene. Moreover, our framework allows us to tune our model to specific scenes,\nenhancing the rendering and refinement quality. To facilitate research in this\narea, we also present the first multi-view lensless dataset, LenslessScenes.\nExtensive experiments demonstrate that our method outperforms current\napproaches in reconstruction accuracy and refinement quality. Code and video\nresults are available at https://rakesh-123-cryp.github.io/Rakesh.github.io/\n", "date": "2024-11-07", "categories": []}
{"id": "2404.03202", "title": "OmniGS: Fast Radiance Field Reconstruction using Omnidirectional\n  Gaussian Splatting", "abstract": "  Photorealistic reconstruction relying on 3D Gaussian Splatting has shown\npromising potential in various domains. However, the current 3D Gaussian\nSplatting system only supports radiance field reconstruction using undistorted\nperspective images. In this paper, we present OmniGS, a novel omnidirectional\nGaussian splatting system, to take advantage of omnidirectional images for fast\nradiance field reconstruction. Specifically, we conduct a theoretical analysis\nof spherical camera model derivatives in 3D Gaussian Splatting. According to\nthe derivatives, we then implement a new GPU-accelerated omnidirectional\nrasterizer that directly splats 3D Gaussians onto the equirectangular screen\nspace for omnidirectional image rendering. We realize differentiable\noptimization of the omnidirectional radiance field without the requirement of\ncube-map rectification or tangent-plane approximation. Extensive experiments\nconducted in egocentric and roaming scenarios demonstrate that our method\nachieves state-of-the-art reconstruction quality and high rendering speed using\nomnidirectional images. The code will be publicly available.\n", "date": "2024-04-04", "categories": []}
{"id": "2405.16749", "title": "DMPlug: A Plug-in Method for Solving Inverse Problems with Diffusion\n  Models", "abstract": "  Pretrained diffusion models (DMs) have recently been popularly used in\nsolving inverse problems (IPs). The existing methods mostly interleave\niterative steps in the reverse diffusion process and iterative steps to bring\nthe iterates closer to satisfying the measurement constraint. However, such\ninterleaving methods struggle to produce final results that look like natural\nobjects of interest (i.e., manifold feasibility) and fit the measurement (i.e.,\nmeasurement feasibility), especially for nonlinear IPs. Moreover, their\ncapabilities to deal with noisy IPs with unknown types and levels of\nmeasurement noise are unknown. In this paper, we advocate viewing the reverse\nprocess in DMs as a function and propose a novel plug-in method for solving IPs\nusing pretrained DMs, dubbed DMPlug. DMPlug addresses the issues of manifold\nfeasibility and measurement feasibility in a principled manner, and also shows\ngreat potential for being robust to unknown types and levels of noise. Through\nextensive experiments across various IP tasks, including two linear and three\nnonlinear IPs, we demonstrate that DMPlug consistently outperforms\nstate-of-the-art methods, often by large margins especially for nonlinear IPs.\nThe code is available at https://github.com/sun-umn/DMPlug.\n", "date": "2024-05-27", "categories": []}
{"id": "2411.03615", "title": "ADMIRE: a locally adaptive single-image, non-uniformity correction and\n  denoising algorithm: application to uncooled IR camera", "abstract": "  We propose a new way to correct for the non-uniformity (NU) and the noise in\nuncooled infrared-type images. This method works on static images, needs no\nregistration, no camera motion and no model for the non uniformity. The\nproposed method uses an hybrid scheme including an automatic locally-adaptive\ncontrast adjustment and a state-of-the-art image denoising method. It permits\nto correct for a fully non-linear NU and the noise efficiently using only one\nimage. We compared it with total variation on real raw and simulated NU\ninfrared images. The strength of this approach lies in its simplicity, low\ncomputational cost. It needs no test-pattern or calibration and produces no\n\"ghost-artefact\".\n", "date": "2024-11-06", "categories": []}
{"id": "2407.11781", "title": "Sliding Gaussian ball adaptive growth (SlingBAG): point cloud-based\n  iterative algorithm for large-scale 3D photoacoustic imaging", "abstract": "  Large-scale 3D photoacoustic (PA) imaging has become increasingly important\nfor both clinical and pre-clinical applications. Limited by cost and system\ncomplexity, only systems with sparsely-distributed sensors can be widely\nimplemented, which desires advanced reconstruction algorithms to reduce\nartifacts. However, high computing memory and time consumption of traditional\niterative reconstruction (IR) algorithms is practically unacceptable for\nlarge-scale 3D PA imaging. Here, we propose a point cloud-based IR algorithm\nthat reduces memory consumption by several orders, wherein the 3D PA scene is\nmodeled as a series of Gaussian-distributed spherical sources stored in form of\npoint cloud. During the IR process, not only are properties of each Gaussian\nsource, including its peak intensity (initial pressure value), standard\ndeviation (size) and mean (position) continuously optimized, but also each\nGaussian source itself adaptively undergoes destroying, splitting, and\nduplication along the gradient direction. This method, named the sliding\nGaussian ball adaptive growth (SlingBAG) algorithm, enables high-quality\nlarge-scale 3D PA reconstruction with fast iteration and extremely low memory\nusage. We validated SlingBAG algorithm in both simulation study and in vivo\nanimal experiments. The source code and data for SlingBAG, along with\nsupplementary materials and demonstration videos, are now available in the\nfollowing GitHub repository: https://github.com/JaegerCQ/SlingBAG.\n", "date": "2024-07-16", "categories": []}
{"id": "2411.02889", "title": "Turbulence stabilization", "abstract": "  We recently developed a new approach to get a stabilized image from a\nsequence of frames acquired through atmospheric turbulence. The goal of this\nalgorihtm is to remove the geometric distortions due by the atmosphere\nmovements. This method is based on a variational formulation and is efficiently\nsolved by the use of Bregman iterations and the operator splitting method. In\nthis paper we propose to study the influence of the choice of the regularizing\nterm in the model. Then we proposed to experiment some of the most used\nregularization constraints available in the litterature.\n", "date": "2024-11-05", "categories": []}
{"id": "2411.02890", "title": "Fried deconvolution", "abstract": "  In this paper we present a new approach to deblur the effect of atmospheric\nturbulence in the case of long range imaging. Our method is based on an\nanalytical formulation, the Fried kernel, of the atmosphere modulation transfer\nfunction (MTF) and a framelet based deconvolution algorithm. An important\nparameter is the refractive index structure which requires specific\nmeasurements to be known. Then we propose a method which provides a good\nestimation of this parameter from the input blurred image. The final algorithms\nare very easy to implement and show very good results on both simulated blur\nand real images.\n", "date": "2024-11-05", "categories": []}
{"id": "2411.02985", "title": "Sparse Reconstruction of Wavefronts using an Over-Complete Phase\n  Dictionary", "abstract": "  Wavefront reconstruction is a critical component in various optical systems,\nincluding adaptive optics, interferometry, and phase contrast imaging.\nTraditional reconstruction methods often employ either the Cartesian (pixel)\nbasis or the Zernike polynomial basis. While the Cartesian basis is adept at\ncapturing high-frequency features, it is susceptible to overfitting and\ninefficiencies due to the high number of degrees of freedom. The Zernike basis\nefficiently represents common optical aberrations but struggles with complex or\nnon-standard wavefronts such as optical vortices, Bessel beams, or wavefronts\nwith sharp discontinuities. This paper introduces a novel approach to wavefront\nreconstruction using an over-complete phase dictionary combined with sparse\nrepresentation techniques. By constructing a dictionary that includes a diverse\nset of basis functions - ranging from Zernike polynomials to specialized\nfunctions representing optical vortices and other complex modes - we enable a\nmore flexible and efficient representation of complex wavefronts. Furthermore,\na trainable affine transform is implemented to account for misalignment.\nUtilizing principles from compressed sensing and sparse coding, we enforce\nsparsity in the coefficient space to avoid overfitting and enhance robustness\nto noise.\n", "date": "2024-11-05", "categories": []}
{"id": "2312.10568", "title": "IntraSeismic: a coordinate-based learning approach to seismic inversion", "abstract": "  Seismic imaging is the numerical process of creating a volumetric\nrepresentation of the subsurface geological structures from elastic waves\nrecorded at the surface of the Earth. As such, it is widely utilized in the\nenergy and construction sectors for applications ranging from oil and gas\nprospection, to geothermal production and carbon capture and storage\nmonitoring, to geotechnical assessment of infrastructures. Extracting\nquantitative information from seismic recordings, such as an acoustic impedance\nmodel, is however a highly ill-posed inverse problem, due to the band-limited\nand noisy nature of the data. This paper introduces IntraSeismic, a novel\nhybrid seismic inversion method that seamlessly combines coordinate-based\nlearning with the physics of the post-stack modeling operator. Key features of\nIntraSeismic are i) unparalleled performance in 2D and 3D post-stack seismic\ninversion, ii) rapid convergence rates, iii) ability to seamlessly include hard\nconstraints (i.e., well data) and perform uncertainty quantification, and iv)\npotential data compression and fast randomized access to portions of the\ninverted model. Synthetic and field data applications of IntraSeismic are\npresented to validate the effectiveness of the proposed method.\n", "date": "2023-12-17", "categories": []}
{"id": "2312.13220", "title": "SISMIK for brain MRI: Deep-learning-based motion estimation and\n  model-based motion correction in k-space", "abstract": "  MRI, a widespread non-invasive medical imaging modality, is highly sensitive\nto patient motion. Despite many attempts over the years, motion correction\nremains a difficult problem and there is no general method applicable to all\nsituations. We propose a retrospective method for motion estimation and\ncorrection to tackle the problem of in-plane rigid-body motion, apt for\nclassical 2D Spin-Echo scans of the brain, which are regularly used in clinical\npractice. Due to the sequential acquisition of k-space, motion artifacts are\nwell localized. The method leverages the power of deep neural networks to\nestimate motion parameters in k-space and uses a model-based approach to\nrestore degraded images to avoid ''hallucinations''. Notable advantages are its\nability to estimate motion occurring in high spatial frequencies without the\nneed of a motion-free reference. The proposed method operates on the whole\nk-space dynamic range and is moderately affected by the lower SNR of higher\nharmonics. As a proof of concept, we provide models trained using supervised\nlearning on 600k motion simulations based on motion-free scans of 43 different\nsubjects. Generalization performance was tested with simulations as well as\nin-vivo. Qualitative and quantitative evaluations are presented for motion\nparameter estimations and image reconstruction. Experimental results show that\nour approach is able to obtain good generalization performance on simulated\ndata and in-vivo acquisitions. We provide a Python implementation at\nhttps://gitlab.unige.ch/Oscar.Dabrowski/sismik_mri/.\n", "date": "2023-12-20", "categories": []}
{"id": "2411.00893", "title": "Blind Time-of-Flight Imaging: Sparse Deconvolution on the Continuum with\n  Unknown Kernels", "abstract": "  In recent years, computational Time-of-Flight (ToF) imaging has emerged as an\nexciting and a novel imaging modality that offers new and powerful\ninterpretations of natural scenes, with applications extending to 3D,\nlight-in-flight, and non-line-of-sight imaging. Mathematically, ToF imaging\nrelies on algorithmic super-resolution, as the back-scattered sparse light\nechoes lie on a finer time resolution than what digital devices can capture.\nTraditional methods necessitate knowledge of the emitted light pulses or\nkernels and employ sparse deconvolution to recover scenes. Unlike previous\napproaches, this paper introduces a novel, blind ToF imaging technique that\ndoes not require kernel calibration and recovers sparse spikes on a continuum,\nrather than a discrete grid. By studying the shared characteristics of various\nToF modalities, we capitalize on the fact that most physical pulses\napproximately satisfy the Strang-Fix conditions from approximation theory. This\nleads to a new mathematical formulation for sparse super-resolution. Our\nrecovery approach uses an optimization method that is pivoted on an alternating\nminimization strategy. We benchmark our blind ToF method against traditional\nkernel calibration methods, which serve as the baseline. Extensive hardware\nexperiments across different ToF modalities demonstrate the algorithmic\nadvantages, flexibility and empirical robustness of our approach. We show that\nour work facilitates super-resolution in scenarios where distinguishing between\nclosely spaced objects is challenging, while maintaining performance comparable\nto known kernel situations. Examples of light-in-flight imaging and light-sweep\nvideos highlight the practical benefits of our blind super-resolution method in\nenhancing the understanding of natural scenes.\n", "date": "2024-10-31", "categories": []}
{"id": "2411.01788", "title": "Non rigid geometric distortions correction -- Application to atmospheric\n  turbulence stabilization", "abstract": "  A novel approach is presented to recover an image degraded by atmospheric\nturbulence. Given a sequence of frames affected by turbulence, we construct a\nvariational model to characterize the static image. The optimization problem is\nsolved by Bregman Iteration and the operator splitting method. Our algorithm is\nsimple, efficient, and can be easily generalized for different scenarios.\n", "date": "2024-11-04", "categories": []}
{"id": "2411.00605", "title": "pcaGAN: Improving Posterior-Sampling cGANs via Principal Component\n  Regularization", "abstract": "  In ill-posed imaging inverse problems, there can exist many hypotheses that\nfit both the observed measurements and prior knowledge of the true image.\nRather than returning just one hypothesis of that image, posterior samplers aim\nto explore the full solution space by generating many probable hypotheses,\nwhich can later be used to quantify uncertainty or construct recoveries that\nappropriately navigate the perception/distortion trade-off. In this work, we\npropose a fast and accurate posterior-sampling conditional generative\nadversarial network (cGAN) that, through a novel form of regularization, aims\nfor correctness in the posterior mean as well as the trace and K principal\ncomponents of the posterior covariance matrix. Numerical experiments\ndemonstrate that our method outperforms contemporary cGANs and diffusion models\nin imaging inverse problems like denoising, large-scale inpainting, and\naccelerated MRI recovery. The code for our model can be found here:\nhttps://github.com/matt-bendel/pcaGAN.\n", "date": "2024-11-01", "categories": []}
{"id": "2411.00688", "title": "Why do we regularise in every iteration for imaging inverse problems?", "abstract": "  Regularisation is commonly used in iterative methods for solving imaging\ninverse problems. Many algorithms involve the evaluation of the proximal\noperator of the regularisation term in every iteration, leading to a\nsignificant computational overhead since such evaluation can be costly. In this\ncontext, the ProxSkip algorithm, recently proposed for federated learning\npurposes, emerges as an solution. It randomly skips regularisation steps,\nreducing the computational time of an iterative algorithm without affecting its\nconvergence. Here we explore for the first time the efficacy of ProxSkip to a\nvariety of imaging inverse problems and we also propose a novel PDHGSkip\nversion. Extensive numerical results highlight the potential of these methods\nto accelerate computations while maintaining high-quality reconstructions.\n", "date": "2024-11-01", "categories": []}
{"id": "2410.22802", "title": "Wavelet Burst Accumulation for turbulence mitigation", "abstract": "  In this paper, we investigate the extension of the recently proposed weighted\nFourier burst accumulation (FBA) method into the wavelet domain. The purpose of\nFBA is to reconstruct a clean and sharp image from a sequence of blurred\nframes. This concept lies in the construction of weights to amplify dominant\nfrequencies in the Fourier spectrum of each frame. The reconstructed image is\nthen obtained by taking the inverse Fourier transform of the average of all\nprocessed spectra. In this paper, we first suggest to replace the rigid\nregistration step used in the original algorithm by a non-rigid registration in\norder to be able to process sequences acquired through atmospheric turbulence.\nSecond, we propose to work in a wavelet domain instead of the Fourier one. This\nleads us to the construction of two types of algorithms. Finally, we propose an\nalternative approach to replace the weighting idea by an approach promoting the\nsparsity in the used space. Several experiments are provided to illustrate the\nefficiency of the proposed methods.\n", "date": "2024-10-30", "categories": []}
{"id": "2403.09486", "title": "SpikeReveal: Unlocking Temporal Sequences from Real Blurry Inputs with\n  Spike Streams", "abstract": "  Reconstructing a sequence of sharp images from the blurry input is crucial\nfor enhancing our insights into the captured scene and poses a significant\nchallenge due to the limited temporal features embedded in the image. Spike\ncameras, sampling at rates up to 40,000 Hz, have proven effective in capturing\nmotion features and beneficial for solving this ill-posed problem. Nonetheless,\nexisting methods fall into the supervised learning paradigm, which suffers from\nnotable performance degradation when applied to real-world scenarios that\ndiverge from the synthetic training data domain. Moreover, the quality of\nreconstructed images is capped by the generated images based on motion analysis\ninterpolation, which inherently differs from the actual scene, affecting the\ngeneralization ability of these methods in real high-speed scenarios. To\naddress these challenges, we propose the first self-supervised framework for\nthe task of spike-guided motion deblurring. Our approach begins with the\nformulation of a spike-guided deblurring model that explores the theoretical\nrelationships among spike streams, blurry images, and their corresponding sharp\nsequences. We subsequently develop a self-supervised cascaded framework to\nalleviate the issues of spike noise and spatial-resolution mismatching\nencountered in the deblurring model. With knowledge distillation and\nre-blurring loss, we further design a lightweight deblur network to generate\nhigh-quality sequences with brightness and texture consistency with the\noriginal input. Quantitative and qualitative experiments conducted on our\nreal-world and synthetic datasets with spikes validate the superior\ngeneralization of the proposed framework. Our code, data and trained models\nwill be available at \\url{https://github.com/chenkang455/S-SDM}.\n", "date": "2024-03-14", "categories": []}
{"id": "2410.18666", "title": "DreamClear: High-Capacity Real-World Image Restoration with Privacy-Safe\n  Dataset Curation", "abstract": "  Image restoration (IR) in real-world scenarios presents significant\nchallenges due to the lack of high-capacity models and comprehensive datasets.\nTo tackle these issues, we present a dual strategy: GenIR, an innovative data\ncuration pipeline, and DreamClear, a cutting-edge Diffusion Transformer\n(DiT)-based image restoration model. GenIR, our pioneering contribution, is a\ndual-prompt learning pipeline that overcomes the limitations of existing\ndatasets, which typically comprise only a few thousand images and thus offer\nlimited generalizability for larger models. GenIR streamlines the process into\nthree stages: image-text pair construction, dual-prompt based fine-tuning, and\ndata generation & filtering. This approach circumvents the laborious data\ncrawling process, ensuring copyright compliance and providing a cost-effective,\nprivacy-safe solution for IR dataset construction. The result is a large-scale\ndataset of one million high-quality images. Our second contribution,\nDreamClear, is a DiT-based image restoration model. It utilizes the generative\npriors of text-to-image (T2I) diffusion models and the robust perceptual\ncapabilities of multi-modal large language models (MLLMs) to achieve\nphotorealistic restoration. To boost the model's adaptability to diverse\nreal-world degradations, we introduce the Mixture of Adaptive Modulator (MoAM).\nIt employs token-wise degradation priors to dynamically integrate various\nrestoration experts, thereby expanding the range of degradations the model can\naddress. Our exhaustive experiments confirm DreamClear's superior performance,\nunderlining the efficacy of our dual strategy for real-world image restoration.\nCode and pre-trained models are available at:\nhttps://github.com/shallowdream204/DreamClear.\n", "date": "2024-10-24", "categories": []}
{"id": "2410.19415", "title": "Integration of Communication and Computational Imaging", "abstract": "  Communication enables the expansion of human visual perception beyond the\nlimitations of time and distance, while computational imaging overcomes the\nconstraints of depth and breadth. Although impressive achievements have been\nwitnessed with the two types of technologies, the occlusive information flow\nbetween the two domains is a bottleneck hindering their ulterior progression.\nHerein, we propose a novel framework that integrates communication and\ncomputational imaging (ICCI) to break through the inherent isolation between\ncommunication and computational imaging for remote perception. By jointly\nconsidering the sensing and transmitting of remote visual information, the ICCI\nframework performs a full-link information transfer optimization, aiming to\nminimize information loss from the generation of the information source to the\nexecution of the final vision tasks. We conduct numerical analysis and\nexperiments to demonstrate the ICCI framework by integrating communication\nsystems and snapshot compressive imaging systems. Compared with straightforward\ncombination schemes, which sequentially execute sensing and transmitting, the\nICCI scheme shows greater robustness against channel noise and impairments\nwhile achieving higher data compression. Moreover, an 80 km 27-band\nhyperspectral video perception with a rate of 30 fps is experimentally\nachieved. This new ICCI remote perception paradigm offers a highefficiency\nsolution for various real-time computer vision tasks.\n", "date": "2024-10-25", "categories": []}
{"id": "2410.21535", "title": "ECMamba: Consolidating Selective State Space Model with Retinex Guidance\n  for Efficient Multiple Exposure Correction", "abstract": "  Exposure Correction (EC) aims to recover proper exposure conditions for\nimages captured under over-exposure or under-exposure scenarios. While existing\ndeep learning models have shown promising results, few have fully embedded\nRetinex theory into their architecture, highlighting a gap in current\nmethodologies. Additionally, the balance between high performance and\nefficiency remains an under-explored problem for exposure correction task.\nInspired by Mamba which demonstrates powerful and highly efficient sequence\nmodeling, we introduce a novel framework based on Mamba for Exposure Correction\n(ECMamba) with dual pathways, each dedicated to the restoration of reflectance\nand illumination map, respectively. Specifically, we firstly derive the Retinex\ntheory and we train a Retinex estimator capable of mapping inputs into two\nintermediary spaces, each approximating the target reflectance and illumination\nmap, respectively. This setup facilitates the refined restoration process of\nthe subsequent Exposure Correction Mamba Module (ECMM). Moreover, we develop a\nnovel 2D Selective State-space layer guided by Retinex information\n(Retinex-SS2D) as the core operator of ECMM. This architecture incorporates an\ninnovative 2D scanning strategy based on deformable feature aggregation,\nthereby enhancing both efficiency and effectiveness. Extensive experiment\nresults and comprehensive ablation studies demonstrate the outstanding\nperformance and the importance of each component of our proposed ECMamba. Code\nis available at https://github.com/LowlevelAI/ECMamba.\n", "date": "2024-10-28", "categories": []}
{"id": "2211.16005", "title": "Convex Relaxations for Isometric and Equiareal NRSfM", "abstract": "  Extensible objects form a challenging case for NRSfM, owing to the lack of a\nsufficiently constrained extensible model of the point-cloud. We tackle the\nchallenge by proposing 1) convex relaxations of the isometric model up to\nquasi-isometry, and 2) convex relaxations involving the equiareal deformation\nmodel, which preserves local area and has not been used in NRSfM. The equiareal\nmodel is appealing because it is physically plausible and widely applicable.\nHowever, it has two main difficulties: first, when used on its own, it is\nambiguous, and second, it involves quartic, hence highly nonconvex,\nconstraints. Our approach handles the first difficulty by mixing the equiareal\nwith the isometric model and the second difficulty by new convex relaxations.\nWe validate our methods on multiple real and synthetic data, including\nwell-known benchmarks.\n", "date": "2022-11-29", "categories": []}
{"id": "2309.17389", "title": "Prompt-based test-time real image dehazing: a novel pipeline", "abstract": "  Existing methods attempt to improve models' generalization ability on\nreal-world hazy images by exploring well-designed training schemes (\\eg,\nCycleGAN, prior loss). However, most of them need very complicated training\nprocedures to achieve satisfactory results. For the first time, we present a\nnovel pipeline called Prompt-based Test-Time Dehazing (PTTD) to help generate\nvisually pleasing results of real-captured hazy images during the inference\nphase. We experimentally observe that given a dehazing model trained on\nsynthetic data, fine-tuning the statistics (\\ie, mean and standard deviation)\nof encoding features is able to narrow the domain gap, boosting the performance\nof real image dehazing. Accordingly, we first apply a prompt generation module\n(PGM) to generate a visual prompt, which is the reference of appropriate\nstatistical perturbations for mean and standard deviation. Then, we employ a\nfeature adaptation module (FAM) into the existing dehazing models for adjusting\nthe original statistics with the guidance of the generated prompt. PTTD is\nmodel-agnostic and can be equipped with various state-of-the-art dehazing\nmodels trained on synthetic hazy-clean pairs to tackle the real image dehazing\ntask. Extensive experimental results demonstrate that our PTTD is effective,\nachieving superior performance against state-of-the-art dehazing methods in\nreal-world scenarios. The code is available at\n\\url{https://github.com/cecret3350/PTTD-Dehazing}.\n", "date": "2023-09-29", "categories": []}
{"id": "2404.06693", "title": "Binomial Self-compensation for Motion Error in Dynamic 3D Scanning", "abstract": "  Phase shifting profilometry (PSP) is favored in high-precision 3D scanning\ndue to its high accuracy, robustness, and pixel-wise property. However, a\nfundamental assumption of PSP that the object should remain static is violated\nin dynamic measurement, making PSP susceptible to object moving, resulting in\nripple-like errors in the point clouds. We propose a pixel-wise and frame-wise\nloopable binomial self-compensation (BSC) algorithm to effectively and flexibly\neliminate motion error in the four-step PSP. Our mathematical model\ndemonstrates that by summing successive motion-affected phase frames weighted\nby binomial coefficients, motion error exponentially diminishes as the binomial\norder increases, accomplishing automatic error compensation through the\nmotion-affected phase sequence, without the assistance of any intermediate\nvariable. Extensive experiments show that our BSC outperforms the existing\nmethods in reducing motion error, while achieving a depth map frame rate equal\nto the camera's acquisition rate (90 fps), enabling high-accuracy 3D\nreconstruction with a quasi-single-shot frame rate.\n", "date": "2024-04-10", "categories": []}
{"id": "2405.20693", "title": "R$^2$-Gaussian: Rectifying Radiative Gaussian Splatting for Tomographic\n  Reconstruction", "abstract": "  3D Gaussian splatting (3DGS) has shown promising results in image rendering\nand surface reconstruction. However, its potential in volumetric reconstruction\ntasks, such as X-ray computed tomography, remains under-explored. This paper\nintroduces R$^2$-Gaussian, the first 3DGS-based framework for sparse-view\ntomographic reconstruction. By carefully deriving X-ray rasterization\nfunctions, we discover a previously unknown integration bias in the standard\n3DGS formulation, which hampers accurate volume retrieval. To address this\nissue, we propose a novel rectification technique via refactoring the\nprojection from 3D to 2D Gaussians. Our new method presents three key\ninnovations: (1) introducing tailored Gaussian kernels, (2) extending\nrasterization to X-ray imaging, and (3) developing a CUDA-based differentiable\nvoxelizer. Experiments on synthetic and real-world datasets demonstrate that\nour method outperforms state-of-the-art approaches in accuracy and efficiency.\nCrucially, it delivers high-quality results in 4 minutes, which is 12$\\times$\nfaster than NeRF-based methods and on par with traditional algorithms. Code and\nmodels are available on the project page\nhttps://github.com/Ruyi-Zha/r2_gaussian.\n", "date": "2024-05-31", "categories": []}
{"id": "2408.05449", "title": "Unidirectional imaging with partially coherent light", "abstract": "  Unidirectional imagers form images of input objects only in one direction,\ne.g., from field-of-view (FOV) A to FOV B, while blocking the image formation\nin the reverse direction, from FOV B to FOV A. Here, we report unidirectional\nimaging under spatially partially coherent light and demonstrate high-quality\nimaging only in the forward direction (A->B) with high power efficiency while\ndistorting the image formation in the backward direction (B->A) along with low\npower efficiency. Our reciprocal design features a set of spatially engineered\nlinear diffractive layers that are statistically optimized for partially\ncoherent illumination with a given phase correlation length. Our analyses\nreveal that when illuminated by a partially coherent beam with a correlation\nlength of ~1.5 w or larger, where w is the wavelength of light, diffractive\nunidirectional imagers achieve robust performance, exhibiting asymmetric\nimaging performance between the forward and backward directions - as desired. A\npartially coherent unidirectional imager designed with a smaller correlation\nlength of less than 1.5 w still supports unidirectional image transmission, but\nwith a reduced figure of merit. These partially coherent diffractive\nunidirectional imagers are compact (axially spanning less than 75 w),\npolarization-independent, and compatible with various types of illumination\nsources, making them well-suited for applications in asymmetric visual\ninformation processing and communication.\n", "date": "2024-08-10", "categories": []}
{"id": "2410.20558", "title": "Neural rendering enables dynamic tomography", "abstract": "  Interrupted X-ray computed tomography (X-CT) has been the common way to\nobserve the deformation of materials during an experiment. While this approach\nis effective for quasi-static experiments, it has never been possible to\nreconstruct a full 3d tomography during a dynamic experiment which cannot be\ninterrupted. In this work, we propose that neural rendering tools can be used\nto drive the paradigm shift to enable 3d reconstruction during dynamic events.\nFirst, we derive theoretical results to support the selection of projections\nangles. Via a combination of synthetic and experimental data, we demonstrate\nthat neural radiance fields can reconstruct data modalities of interest more\nefficiently than conventional reconstruction methods. Finally, we develop a\nspatio-temporal model with spline-based deformation field and demonstrate that\nsuch model can reconstruct the spatio-temporal deformation of lattice samples\nin real-world experiments.\n", "date": "2024-10-27", "categories": []}
{"id": "2410.20716", "title": "Physics-Free Spectrally Multiplexed Photometric Stereo under Unknown\n  Spectral Composition", "abstract": "  In this paper, we present a groundbreaking spectrally multiplexed photometric\nstereo approach for recovering surface normals of dynamic surfaces without the\nneed for calibrated lighting or sensors, a notable advancement in the field\ntraditionally hindered by stringent prerequisites and spectral ambiguity. By\nembracing spectral ambiguity as an advantage, our technique enables the\ngeneration of training data without specialized multispectral rendering\nframeworks. We introduce a unique, physics-free network architecture,\nSpectraM-PS, that effectively processes multiplexed images to determine surface\nnormals across a wide range of conditions and material types, without relying\non specific physically-based knowledge. Additionally, we establish the first\nbenchmark dataset, SpectraM14, for spectrally multiplexed photometric stereo,\nfacilitating comprehensive evaluations against existing calibrated methods. Our\ncontributions significantly enhance the capabilities for dynamic surface\nrecovery, particularly in uncalibrated setups, marking a pivotal step forward\nin the application of photometric stereo across various domains.\n", "date": "2024-10-28", "categories": []}
{"id": "2206.03361", "title": "Hierarchical Similarity Learning for Aliasing Suppression Image\n  Super-Resolution", "abstract": "  As a highly ill-posed issue, single image super-resolution (SISR) has been\nwidely investigated in recent years. The main task of SISR is to recover the\ninformation loss caused by the degradation procedure. According to the Nyquist\nsampling theory, the degradation leads to aliasing effect and makes it hard to\nrestore the correct textures from low-resolution (LR) images. In practice,\nthere are correlations and self-similarities among the adjacent patches in the\nnatural images. This paper considers the self-similarity and proposes a\nhierarchical image super-resolution network (HSRNet) to suppress the influence\nof aliasing. We consider the SISR issue in the optimization perspective, and\npropose an iterative solution pattern based on the half-quadratic splitting\n(HQS) method. To explore the texture with local image prior, we design a\nhierarchical exploration block (HEB) and progressive increase the receptive\nfield. Furthermore, multi-level spatial attention (MSA) is devised to obtain\nthe relations of adjacent feature and enhance the high-frequency information,\nwhich acts as a crucial role for visual experience. Experimental result shows\nHSRNet achieves better quantitative and visual performance than other works,\nand remits the aliasing more effectively.\n", "date": "2022-06-07", "categories": []}
{"id": "2206.05647", "title": "A Fast Alternating Minimization Algorithm for Coded Aperture Snapshot\n  Spectral Imaging Based on Sparsity and Deep Image Priors", "abstract": "  Coded aperture snapshot spectral imaging (CASSI) is a technique used to\nreconstruct three-dimensional hyperspectral images (HSIs) from one or several\ntwo-dimensional projection measurements. However, fewer projection measurements\nor more spectral channels leads to a severly ill-posed problem, in which case\nregularization methods have to be applied. In order to significantly improve\nthe accuracy of reconstruction, this paper proposes a fast alternating\nminimization algorithm based on the sparsity and deep image priors (Fama-SDIP)\nof natural images. By integrating deep image prior (DIP) into the principle of\ncompressive sensing (CS) reconstruction, the proposed algorithm can achieve\nstate-of-the-art results without any training dataset. Extensive experiments\nshow that Fama-SDIP method significantly outperforms prevailing leading methods\non simulation and real HSI datasets.\n", "date": "2022-06-12", "categories": []}
{"id": "2212.00069", "title": "MrSARP: A Hierarchical Deep Generative Prior for SAR Image\n  Super-resolution", "abstract": "  Generative models learned from training using deep learning methods can be\nused as priors in inverse under-determined inverse problems, including imaging\nfrom sparse set of measurements. In this paper, we present a novel hierarchical\ndeep-generative model MrSARP for SAR imagery that can synthesize SAR images of\na target at different resolutions jointly. MrSARP is trained in conjunction\nwith a critic that scores multi resolution images jointly to decide if they are\nrealistic images of a target at different resolutions. We show how this deep\ngenerative model can be used to retrieve the high spatial resolution image from\nlow resolution images of the same target. The cost function of the generator is\nmodified to improve its capability to retrieve the input parameters for a given\nset of resolution images. We evaluate the model's performance using the three\nstandard error metrics used for evaluating super-resolution performance on\nsimulated data and compare it to upsampling and sparsity based image sharpening\napproaches.\n", "date": "2022-11-30", "categories": []}
{"id": "2311.10523", "title": "Removing Adverse Volumetric Effects From Trained Neural Radiance Fields", "abstract": "  While the use of neural radiance fields (NeRFs) in different challenging\nsettings has been explored, only very recently have there been any\ncontributions that focus on the use of NeRF in foggy environments. We argue\nthat the traditional NeRF models are able to replicate scenes filled with fog\nand propose a method to remove the fog when synthesizing novel views. By\ncalculating the global contrast of a scene, we can estimate a density threshold\nthat, when applied, removes all visible fog. This makes it possible to use NeRF\nas a way of rendering clear views of objects of interest located in fog-filled\nenvironments. Additionally, to benchmark performance on such scenes, we\nintroduce a new dataset that expands some of the original synthetic NeRF scenes\nthrough the addition of fog and natural environments. The code, dataset, and\nvideo results can be found on our project page: https://vegardskui.com/fognerf/\n", "date": "2023-11-17", "categories": []}
{"id": "2401.16039", "title": "Data-Driven Filter Design in FBP: Transforming CT Reconstruction with\n  Trainable Fourier Series", "abstract": "  In this study, we introduce a Fourier series-based trainable filter for\ncomputed tomography (CT) reconstruction within the filtered backprojection\n(FBP) framework. This method overcomes the limitation in noise reduction by\noptimizing Fourier series coefficients to construct the filter, maintaining\ncomputational efficiency with minimal increment for the trainable parameters\ncompared to other deep learning frameworks. Additionally, we propose Gaussian\nedge-enhanced (GEE) loss function that prioritizes the $L_1$ norm of\nhigh-frequency magnitudes, effectively countering the blurring problems\nprevalent in mean squared error (MSE) approaches. The model's foundation in the\nFBP algorithm ensures excellent interpretability, as it relies on a data-driven\nfilter with all other parameters derived through rigorous mathematical\nprocedures. Designed as a plug-and-play solution, our Fourier series-based\nfilter can be easily integrated into existing CT reconstruction models, making\nit an adaptable tool for a wide range of practical applications. Code and data\nare available at https://github.com/sypsyp97/Trainable-Fourier-Series.\n", "date": "2024-01-29", "categories": []}
{"id": "2406.08177", "title": "One-Step Effective Diffusion Network for Real-World Image\n  Super-Resolution", "abstract": "  The pre-trained text-to-image diffusion models have been increasingly\nemployed to tackle the real-world image super-resolution (Real-ISR) problem due\nto their powerful generative image priors. Most of the existing methods start\nfrom random noise to reconstruct the high-quality (HQ) image under the guidance\nof the given low-quality (LQ) image. While promising results have been\nachieved, such Real-ISR methods require multiple diffusion steps to reproduce\nthe HQ image, increasing the computational cost. Meanwhile, the random noise\nintroduces uncertainty in the output, which is unfriendly to image restoration\ntasks. To address these issues, we propose a one-step effective diffusion\nnetwork, namely OSEDiff, for the Real-ISR problem. We argue that the LQ image\ncontains rich information to restore its HQ counterpart, and hence the given LQ\nimage can be directly taken as the starting point for diffusion, eliminating\nthe uncertainty introduced by random noise sampling. We finetune the\npre-trained diffusion network with trainable layers to adapt it to complex\nimage degradations. To ensure that the one-step diffusion model could yield HQ\nReal-ISR output, we apply variational score distillation in the latent space to\nconduct KL-divergence regularization. As a result, our OSEDiff model can\nefficiently and effectively generate HQ images in just one diffusion step. Our\nexperiments demonstrate that OSEDiff achieves comparable or even better\nReal-ISR results, in terms of both objective metrics and subjective\nevaluations, than previous diffusion model-based Real-ISR methods that require\ndozens or hundreds of steps. The source codes are released at\nhttps://github.com/cswry/OSEDiff.\n", "date": "2024-06-12", "categories": []}
{"id": "2410.18789", "title": "Single-Shot Phase Diversity Wavefront Sensing in Deep Turbulence via\n  Metasurface Optics", "abstract": "  Free-space optical communication (FSOC) systems offer high-bandwidth and\nsecure communication with minimal capital costs. Adaptive optics (AO) are\ntypically added to these systems to decrease atmospheric channel losses;\nhowever, the performance of traditional AO wavefront sensors degrades in\nlong-range, deep turbulence conditions. Alternative wavefront sensors using\nphase diversity can successfully reconstruct wavefronts in deep turbulence, but\ncurrent implementations require bulky setups with high latency. In this work,\nwe employ a nanostructured birefringent metasurface optic that enables\nlow-latency phase diversity wavefront sensing in a compact form factor. We\nprove the effectiveness of this approach in mid-to-high turbulence (Rytov\nnumbers from 0.2 to 0.6) through simulation and experimental demonstration. In\nboth cases an average 16-fold increase in signal from the corrected beam is\nobtained. Our approach opens a pathway for compact, robust wavefront sensing\nthat enhances range and accuracy of FSOC systems.\n", "date": "2024-10-24", "categories": []}
{"id": "2410.17377", "title": "PtychoFormer: A Transformer-based Model for Ptychographic Phase\n  Retrieval", "abstract": "  Ptychography is a computational method of microscopy that recovers\nhigh-resolution transmission images of samples from a series of diffraction\npatterns. While conventional phase retrieval algorithms can iteratively recover\nthe images, they require oversampled diffraction patterns, incur significant\ncomputational costs, and struggle to recover the absolute phase of the sample's\ntransmission function. Deep learning algorithms for ptychography are a\npromising approach to resolving the limitations of iterative algorithms. We\npresent PtychoFormer, a hierarchical transformer-based model for data-driven\nsingle-shot ptychographic phase retrieval. PtychoFormer processes subsets of\ndiffraction patterns, generating local inferences that are seamlessly stitched\ntogether to produce a high-quality reconstruction. Our model exhibits tolerance\nto sparsely scanned diffraction patterns and achieves up to 3600 times faster\nimaging speed than the extended ptychographic iterative engine (ePIE). We also\npropose the extended-PtychoFormer (ePF), a hybrid approach that combines the\nbenefits of PtychoFormer with the ePIE. ePF minimizes global phase shifts and\nsignificantly enhances reconstruction quality, achieving state-of-the-art phase\nretrieval in ptychography.\n", "date": "2024-10-22", "categories": []}
{"id": "2310.04432", "title": "Training-free Linear Image Inverses via Flows", "abstract": "  Solving inverse problems without any training involves using a pretrained\ngenerative model and making appropriate modifications to the generation process\nto avoid finetuning of the generative model. While recent methods have explored\nthe use of diffusion models, they still require the manual tuning of many\nhyperparameters for different inverse problems. In this work, we propose a\ntraining-free method for solving linear inverse problems by using pretrained\nflow models, leveraging the simplicity and efficiency of Flow Matching models,\nusing theoretically-justified weighting schemes, and thereby significantly\nreducing the amount of manual tuning. In particular, we draw inspiration from\ntwo main sources: adopting prior gradient correction methods to the flow\nregime, and a solver scheme based on conditional Optimal Transport paths. As\npretrained diffusion models are widely accessible, we also show how to\npractically adapt diffusion models for our method. Empirically, our approach\nrequires no problem-specific tuning across an extensive suite of noisy linear\ninverse problems on high-dimensional datasets, ImageNet-64/128 and AFHQ-256,\nand we observe that our flow-based method for solving inverse problems improves\nupon closely-related diffusion-based methods in most settings.\n", "date": "2023-09-25", "categories": []}
{"id": "2310.11838", "title": "Equivariant Bootstrapping for Uncertainty Quantification in Imaging\n  Inverse Problems", "abstract": "  Scientific imaging problems are often severely ill-posed, and hence have\nsignificant intrinsic uncertainty. Accurately quantifying the uncertainty in\nthe solutions to such problems is therefore critical for the rigorous\ninterpretation of experimental results as well as for reliably using the\nreconstructed images as scientific evidence. Unfortunately, existing imaging\nmethods are unable to quantify the uncertainty in the reconstructed images in a\nmanner that is robust to experiment replications. This paper presents a new\nuncertainty quantification methodology based on an equivariant formulation of\nthe parametric bootstrap algorithm that leverages symmetries and invariance\nproperties commonly encountered in imaging problems. Additionally, the proposed\nmethodology is general and can be easily applied with any image reconstruction\ntechnique, including unsupervised training strategies that can be trained from\nobserved data alone, thus enabling uncertainty quantification in situations\nwhere there is no ground truth data available. We demonstrate the proposed\napproach with a series of numerical experiments and through comparisons with\nalternative uncertainty quantification strategies from the state-of-the-art,\nsuch as Bayesian strategies involving score-based diffusion models and Langevin\nsamplers. In all our experiments, the proposed method delivers remarkably\naccurate high-dimensional confidence regions and outperforms the competing\napproaches in terms of estimation accuracy, uncertainty quantification\naccuracy, and computing time.\n", "date": "2023-10-18", "categories": []}
{"id": "2312.08881", "title": "Parameter Efficient Adaptation for Image Restoration with Heterogeneous\n  Mixture-of-Experts", "abstract": "  Designing single-task image restoration models for specific degradation has\nseen great success in recent years. To achieve generalized image restoration,\nall-in-one methods have recently been proposed and shown potential for multiple\nrestoration tasks using one single model. Despite the promising results, the\nexisting all-in-one paradigm still suffers from high computational costs as\nwell as limited generalization on unseen degradations. In this work, we\nintroduce an alternative solution to improve the generalization of image\nrestoration models. Drawing inspiration from recent advancements in Parameter\nEfficient Transfer Learning (PETL), we aim to tune only a small number of\nparameters to adapt pre-trained restoration models to various tasks. However,\ncurrent PETL methods fail to generalize across varied restoration tasks due to\ntheir homogeneous representation nature. To this end, we propose AdaptIR, a\nMixture-of-Experts (MoE) with orthogonal multi-branch design to capture local\nspatial, global spatial, and channel representation bases, followed by adaptive\nbase combination to obtain heterogeneous representation for different\ndegradations. Extensive experiments demonstrate that our AdaptIR achieves\nstable performance on single-degradation tasks, and excels in\nhybrid-degradation tasks, with fine-tuning only 0.6% parameters for 8 hours.\n", "date": "2023-12-12", "categories": []}
{"id": "2410.14214", "title": "MambaSCI: Efficient Mamba-UNet for Quad-Bayer Patterned Video Snapshot\n  Compressive Imaging", "abstract": "  Color video snapshot compressive imaging (SCI) employs computational imaging\ntechniques to capture multiple sequential video frames in a single\nBayer-patterned measurement. With the increasing popularity of quad-Bayer\npattern in mainstream smartphone cameras for capturing high-resolution videos,\nmobile photography has become more accessible to a wider audience. However,\nexisting color video SCI reconstruction algorithms are designed based on the\ntraditional Bayer pattern. When applied to videos captured by quad-Bayer\ncameras, these algorithms often result in color distortion and ineffective\ndemosaicing, rendering them impractical for primary equipment. To address this\nchallenge, we propose the MambaSCI method, which leverages the Mamba and UNet\narchitectures for efficient reconstruction of quad-Bayer patterned color video\nSCI. To the best of our knowledge, our work presents the first algorithm for\nquad-Bayer patterned SCI reconstruction, and also the initial application of\nthe Mamba model to this task. Specifically, we customize Residual-Mamba-Blocks,\nwhich residually connect the Spatial-Temporal Mamba (STMamba),\nEdge-Detail-Reconstruction (EDR) module, and Channel Attention (CA) module.\nRespectively, STMamba is used to model long-range spatial-temporal dependencies\nwith linear complexity, EDR is for better edge-detail reconstruction, and CA is\nused to compensate for the missing channel information interaction in Mamba\nmodel. Experiments demonstrate that MambaSCI surpasses state-of-the-art methods\nwith lower computational and memory costs. PyTorch style pseudo-code for the\ncore modules is provided in the supplementary materials.\n", "date": "2024-10-18", "categories": []}
{"id": "2410.13570", "title": "RGB to Hyperspectral: Spectral Reconstruction for Enhanced Surgical\n  Imaging", "abstract": "  This study investigates the reconstruction of hyperspectral signatures from\nRGB data to enhance surgical imaging, utilizing the publicly available\nHeiPorSPECTRAL dataset from porcine surgery and an in-house neurosurgery\ndataset. Various architectures based on convolutional neural networks (CNNs)\nand transformer models are evaluated using comprehensive metrics. Transformer\nmodels exhibit superior performance in terms of RMSE, SAM, PSNR and SSIM by\neffectively integrating spatial information to predict accurate spectral\nprofiles, encompassing both visible and extended spectral ranges. Qualitative\nassessments demonstrate the capability to predict spectral profiles critical\nfor informed surgical decision-making during procedures. Challenges associated\nwith capturing both the visible and extended hyperspectral ranges are\nhighlighted using the MAE, emphasizing the complexities involved. The findings\nopen up the new research direction of hyperspectral reconstruction for surgical\napplications and clinical use cases in real-time surgical environments.\n", "date": "2024-10-17", "categories": []}
{"id": "2406.13607", "title": "Ultra-High-Definition Image Restoration: New Benchmarks and A Dual\n  Interaction Prior-Driven Solution", "abstract": "  Ultra-High-Definition (UHD) image restoration has acquired remarkable\nattention due to its practical demand. In this paper, we construct UHD snow and\nrain benchmarks, named UHD-Snow and UHD-Rain, to remedy the deficiency in this\nfield. The UHD-Snow/UHD-Rain is established by simulating the physics process\nof rain/snow into consideration and each benchmark contains 3200 degraded/clear\nimage pairs of 4K resolution. Furthermore, we propose an effective UHD image\nrestoration solution by considering gradient and normal priors in model design\nthanks to these priors' spatial and detail contributions. Specifically, our\nmethod contains two branches: (a) feature fusion and reconstruction branch in\nhigh-resolution space and (b) prior feature interaction branch in\nlow-resolution space. The former learns high-resolution features and fuses\nprior-guided low-resolution features to reconstruct clear images, while the\nlatter utilizes normal and gradient priors to mine useful spatial features and\ndetail features to guide high-resolution recovery better. To better utilize\nthese priors, we introduce single prior feature interaction and dual prior\nfeature interaction, where the former respectively fuses normal and gradient\npriors with high-resolution features to enhance prior ones, while the latter\ncalculates the similarity between enhanced prior ones and further exploits dual\nguided filtering to boost the feature interaction of dual priors. We conduct\nexperiments on both new and existing public datasets and demonstrate the\nstate-of-the-art performance of our method on UHD image low-light enhancement,\ndehazing, deblurring, desonwing, and deraining. The source codes and benchmarks\nare available at \\url{https://github.com/wlydlut/UHDDIP}.\n", "date": "2024-06-19", "categories": []}
{"id": "2402.15648", "title": "MambaIR: A Simple Baseline for Image Restoration with State-Space Model", "abstract": "  Recent years have seen significant advancements in image restoration, largely\nattributed to the development of modern deep neural networks, such as CNNs and\nTransformers. However, existing restoration backbones often face the dilemma\nbetween global receptive fields and efficient computation, hindering their\napplication in practice. Recently, the Selective Structured State Space Model,\nespecially the improved version Mamba, has shown great potential for long-range\ndependency modeling with linear complexity, which offers a way to resolve the\nabove dilemma. However, the standard Mamba still faces certain challenges in\nlow-level vision such as local pixel forgetting and channel redundancy. In this\nwork, we introduce a simple but effective baseline, named MambaIR, which\nintroduces both local enhancement and channel attention to improve the vanilla\nMamba. In this way, our MambaIR takes advantage of the local pixel similarity\nand reduces the channel redundancy. Extensive experiments demonstrate the\nsuperiority of our method, for example, MambaIR outperforms SwinIR by up to\n0.45dB on image SR, using similar computational cost but with a global\nreceptive field. Code is available at \\url{https://github.com/csguoh/MambaIR}.\n", "date": "2024-02-23", "categories": []}
{"id": "2410.11730", "title": "Patch-Based Diffusion Models Beat Whole-Image Models for Mismatched\n  Distribution Inverse Problems", "abstract": "  Diffusion models have achieved excellent success in solving inverse problems\ndue to their ability to learn strong image priors, but existing approaches\nrequire a large training dataset of images that should come from the same\ndistribution as the test dataset. When the training and test distributions are\nmismatched, artifacts and hallucinations can occur in reconstructed images due\nto the incorrect priors. In this work, we systematically study out of\ndistribution (OOD) problems where a known training distribution is first\nprovided. We first study the setting where only a single measurement obtained\nfrom the unknown test distribution is available. Next we study the setting\nwhere a very small sample of data belonging to the test distribution is\navailable, and our goal is still to reconstruct an image from a measurement\nthat came from the test distribution. In both settings, we use a patch-based\ndiffusion prior that learns the image distribution solely from patches.\nFurthermore, in the first setting, we include a self-supervised loss that helps\nthe network output maintain consistency with the measurement. Extensive\nexperiments show that in both settings, the patch-based method can obtain high\nquality image reconstructions that can outperform whole-image models and can\ncompete with methods that have access to large in-distribution training\ndatasets. Furthermore, we show how whole-image models are prone to memorization\nand overfitting, leading to artifacts in the reconstructions, while a\npatch-based model can resolve these issues.\n", "date": "2024-10-15", "categories": []}
{"id": "2405.13805", "title": "Perceptual Fairness in Image Restoration", "abstract": "  Fairness in image restoration tasks is the desire to treat different\nsub-groups of images equally well. Existing definitions of fairness in image\nrestoration are highly restrictive. They consider a reconstruction to be a\ncorrect outcome for a group (e.g., women) only if it falls within the group's\nset of ground truth images (e.g., natural images of women); otherwise, it is\nconsidered entirely incorrect. Consequently, such definitions are prone to\ncontroversy, as errors in image restoration can manifest in various ways. In\nthis work we offer an alternative approach towards fairness in image\nrestoration, by considering the Group Perceptual Index (GPI), which we define\nas the statistical distance between the distribution of the group's ground\ntruth images and the distribution of their reconstructions. We assess the\nfairness of an algorithm by comparing the GPI of different groups, and say that\nit achieves perfect Perceptual Fairness (PF) if the GPIs of all groups are\nidentical. We motivate and theoretically study our new notion of fairness, draw\nits connection to previous ones, and demonstrate its utility on\nstate-of-the-art face image restoration algorithms.\n", "date": "2024-05-22", "categories": []}
{"id": "2408.15143", "title": "A Preliminary Exploration Towards General Image Restoration", "abstract": "  Despite the tremendous success of deep models in various individual image\nrestoration tasks, there are at least two major technical challenges preventing\nthese works from being applied to real-world usages: (1) the lack of\ngeneralization ability and (2) the complex and unknown degradations in\nreal-world scenarios. Existing deep models, tailored for specific individual\nimage restoration tasks, often fall short in effectively addressing these\nchallenges. In this paper, we present a new problem called general image\nrestoration (GIR) which aims to address these challenges within a unified\nmodel. GIR covers most individual image restoration tasks (\\eg, image\ndenoising, deblurring, deraining and super-resolution) and their combinations\nfor general purposes. This paper proceeds to delineate the essential aspects of\nGIR, including problem definition and the overarching significance of\ngeneralization performance. Moreover, the establishment of new datasets and a\nthorough evaluation framework for GIR models is discussed. We conduct a\ncomprehensive evaluation of existing approaches for tackling the GIR challenge,\nilluminating their strengths and pragmatic challenges. By analyzing these\napproaches, we not only underscore the effectiveness of GIR but also highlight\nthe difficulties in its practical implementation. At last, we also try to\nunderstand and interpret these models' behaviors to inspire the future\ndirection. Our work can open up new valuable research directions and contribute\nto the research of general vision.\n", "date": "2024-08-27", "categories": []}
{"id": "2409.16907", "title": "An Adaptive Screen-Space Meshing Approach for Normal Integration", "abstract": "  Reconstructing surfaces from normals is a key component of photometric\nstereo. This work introduces an adaptive surface triangulation in the image\ndomain and afterwards performs the normal integration on a triangle mesh. Our\nkey insight is that surface curvature can be computed from normals. Based on\nthe curvature, we identify flat areas and aggregate pixels into triangles. The\napproximation quality is controlled by a single user parameter facilitating a\nseamless generation of low- to high-resolution meshes. Compared to pixel grids,\nour triangle meshes adapt locally to surface details and allow for a sparser\nrepresentation. Our new mesh-based formulation of the normal integration\nproblem is strictly derived from discrete differential geometry and leads to\nwell-conditioned linear systems. Results on real and synthetic data show that\n10 to 100 times less vertices are required than pixels. Experiments suggest\nthat this sparsity translates into a sublinear runtime in the number of pixels.\nFor 64 MP normal maps, our meshing-first approach generates and integrates\nmeshes in minutes while pixel-based approaches require hours just for the\nintegration.\n", "date": "2024-09-25", "categories": []}
{"id": "2410.10085", "title": "NeRF-enabled Analysis-Through-Synthesis for ISAR Imaging of Small\n  Everyday Objects with Sparse and Noisy UWB Radar Data", "abstract": "  Inverse Synthetic Aperture Radar (ISAR) imaging presents a formidable\nchallenge when it comes to small everyday objects due to their limited Radar\nCross-Section (RCS) and the inherent resolution constraints of radar systems.\nExisting ISAR reconstruction methods including backprojection (BP) often\nrequire complex setups and controlled environments, rendering them impractical\nfor many real-world noisy scenarios. In this paper, we propose a novel\nAnalysis-through-Synthesis (ATS) framework enabled by Neural Radiance Fields\n(NeRF) for high-resolution coherent ISAR imaging of small objects using sparse\nand noisy Ultra-Wideband (UWB) radar data with an inexpensive and portable\nsetup. Our end-to-end framework integrates ultra-wideband radar wave\npropagation, reflection characteristics, and scene priors, enabling efficient\n2D scene reconstruction without the need for costly anechoic chambers or\ncomplex measurement test beds. With qualitative and quantitative comparisons,\nwe demonstrate that the proposed method outperforms traditional techniques and\ngenerates ISAR images of complex scenes with multiple targets and complex\nstructures in Non-Line-of-Sight (NLOS) and noisy scenarios, particularly with\nlimited number of views and sparse UWB radar scans. This work represents a\nsignificant step towards practical, cost-effective ISAR imaging of small\neveryday objects, with broad implications for robotics and mobile sensing\napplications.\n", "date": "2024-10-14", "categories": []}
{"id": "2410.08498", "title": "On a Hidden Property in Computational Imaging", "abstract": "  Computational imaging plays a vital role in various scientific and medical\napplications, such as Full Waveform Inversion (FWI), Computed Tomography (CT),\nand Electromagnetic (EM) inversion. These methods address inverse problems by\nreconstructing physical properties (e.g., the acoustic velocity map in FWI)\nfrom measurement data (e.g., seismic waveform data in FWI), where both\nmodalities are governed by complex mathematical equations. In this paper, we\nempirically demonstrate that despite their differing governing equations, three\ninverse problems (FWI, CT, and EM inversion) share a hidden property within\ntheir latent spaces. Specifically, using FWI as an example, we show that both\nmodalities (the velocity map and seismic waveform data) follow the same set of\none-way wave equations in the latent space, yet have distinct initial\nconditions that are linearly correlated. This suggests that after projection\ninto the latent embedding space, the two modalities correspond to different\nsolutions of the same equation, connected through their initial conditions. Our\nexperiments confirm that this hidden property is consistent across all three\nimaging problems, providing a novel perspective for understanding these\ncomputational imaging tasks.\n", "date": "2024-10-11", "categories": []}
{"id": "2407.04241", "title": "AnySR: Realizing Image Super-Resolution as Any-Scale, Any-Resource", "abstract": "  In an effort to improve the efficiency and scalability of single-image\nsuper-resolution (SISR) applications, we introduce AnySR, to rebuild existing\narbitrary-scale SR methods into any-scale, any-resource implementation. As a\ncontrast to off-the-shelf methods that solve SR tasks across various scales\nwith the same computing costs, our AnySR innovates in: 1) building\narbitrary-scale tasks as any-resource implementation, reducing resource\nrequirements for smaller scales without additional parameters; 2) enhancing\nany-scale performance in a feature-interweaving fashion, inserting scale pairs\ninto features at regular intervals and ensuring correct feature/scale\nprocessing. The efficacy of our AnySR is fully demonstrated by rebuilding most\nexisting arbitrary-scale SISR methods and validating on five popular SISR test\ndatasets. The results show that our AnySR implements SISR tasks in a\ncomputing-more-efficient fashion, and performs on par with existing\narbitrary-scale SISR methods. For the first time, we realize SISR tasks as not\nonly any-scale in literature, but also as any-resource. Code is available at\nhttps://github.com/CrispyFeSo4/AnySR.\n", "date": "2024-07-05", "categories": []}
{"id": "2410.07545", "title": "Calibration of 3D Single-pixel Imaging Systems with a Calibration Field", "abstract": "  3D single-pixel imaging (SPI) is a promising imaging technique that can be\nffexibly applied to various wavebands. The main challenge in 3D SPI is that the\ncalibration usually requires a large number of standard points as references,\nwhich are tricky to capture using single-pixel detectors. Conventional\nsolutions involve sophisticated device deployment and cumbersome operations,\nresulting in hundreds of images needed for calibration. In our work, we\nconstruct a Calibration Field (CaliF) to efffciently generate the standard\npoints from one single image. A high accuracy of the CaliF is guaranteed by the\ntechnique of deep learning and digital twin. We perform experiments with our\nnew method to verify its validity and accuracy. We believe our work holds great\npotential in 3D SPI systems or even general imaging systems.\n", "date": "2024-10-10", "categories": []}
{"id": "2405.07777", "title": "GMSR:Gradient-Guided Mamba for Spectral Reconstruction from RGB Images", "abstract": "  Mainstream approaches to spectral reconstruction (SR) primarily focus on\ndesigning Convolution- and Transformer-based architectures. However, CNN\nmethods often face challenges in handling long-range dependencies, whereas\nTransformers are constrained by computational efficiency limitations. Recent\nbreakthroughs in state-space model (e.g., Mamba) has attracted significant\nattention due to its near-linear computational efficiency and superior\nperformance, prompting our investigation into its potential for SR problem. To\nthis end, we propose the Gradient-guided Mamba for Spectral Reconstruction from\nRGB Images, dubbed GMSR-Net. GMSR-Net is a lightweight model characterized by a\nglobal receptive field and linear computational complexity. Its core comprises\nmultiple stacked Gradient Mamba (GM) blocks, each featuring a tri-branch\nstructure. In addition to benefiting from efficient global feature\nrepresentation by Mamba block, we further innovatively introduce spatial\ngradient attention and spectral gradient attention to guide the reconstruction\nof spatial and spectral cues. GMSR-Net demonstrates a significant\naccuracy-efficiency trade-off, achieving state-of-the-art performance while\nmarkedly reducing the number of parameters and computational burdens. Compared\nto existing approaches, GMSR-Net slashes parameters and FLOPS by substantial\nmargins of 10 times and 20 times, respectively. Code is available at\nhttps://github.com/wxy11-27/GMSR.\n", "date": "2024-05-13", "categories": []}
{"id": "2410.06478", "title": "MaskBlur: Spatial and Angular Data Augmentation for Light Field Image\n  Super-Resolution", "abstract": "  Data augmentation (DA) is an effective approach for enhancing model\nperformance with limited data, such as light field (LF) image super-resolution\n(SR). LF images inherently possess rich spatial and angular information.\nNonetheless, there is a scarcity of DA methodologies explicitly tailored for LF\nimages, and existing works tend to concentrate solely on either the spatial or\nangular domain. This paper proposes a novel spatial and angular DA strategy\nnamed MaskBlur for LF image SR by concurrently addressing spatial and angular\naspects. MaskBlur consists of spatial blur and angular dropout two components.\nSpatial blur is governed by a spatial mask, which controls where pixels are\nblurred, i.e., pasting pixels between the low-resolution and high-resolution\ndomains. The angular mask is responsible for angular dropout, i.e., selecting\nwhich views to perform the spatial blur operation. By doing so, MaskBlur\nenables the model to treat pixels differently in the spatial and angular\ndomains when super-resolving LF images rather than blindly treating all pixels\nequally. Extensive experiments demonstrate the efficacy of MaskBlur in\nsignificantly enhancing the performance of existing SR methods. We further\nextend MaskBlur to other LF image tasks such as denoising, deblurring,\nlow-light enhancement, and real-world SR. Code is publicly available at\n\\url{https://github.com/chaowentao/MaskBlur}.\n", "date": "2024-10-09", "categories": []}
{"id": "2410.07043", "title": "Z-upscaling: Optical Flow Guided Frame Interpolation for Isotropic\n  Reconstruction of 3D EM Volumes", "abstract": "  We propose a novel optical flow based approach to enhance the axial\nresolution of anisotropic 3D EM volumes to achieve isotropic 3D reconstruction.\nAssuming spatial continuity of 3D biological structures in well aligned EM\nvolumes, we reasoned that optical flow estimation techniques, often applied for\ntemporal resolution enhancement in videos, can be utilized. Pixel level motion\nis estimated between neighboring 2D slices along z, using spatial gradient flow\nestimates to interpolate and generate new 2D slices resulting in isotropic\nvoxels. We leverage recent state-of-the-art learning methods for video frame\ninterpolation and transfer learning techniques, and demonstrate the success of\nour approach on publicly available ultrastructure EM volumes.\n", "date": "2024-10-09", "categories": []}
{"id": "2303.14739", "title": "Geometry-Aware Attenuation Learning for Sparse-View CBCT Reconstruction", "abstract": "  Cone Beam Computed Tomography (CBCT) plays a vital role in clinical imaging.\nTraditional methods typically require hundreds of 2D X-ray projections to\nreconstruct a high-quality 3D CBCT image, leading to considerable radiation\nexposure. This has led to a growing interest in sparse-view CBCT reconstruction\nto reduce radiation doses. While recent advances, including deep learning and\nneural rendering algorithms, have made strides in this area, these methods\neither produce unsatisfactory results or suffer from time inefficiency of\nindividual optimization. In this paper, we introduce a novel geometry-aware\nencoder-decoder framework to solve this problem. Our framework starts by\nencoding multi-view 2D features from various 2D X-ray projections with a 2D CNN\nencoder. Leveraging the geometry of CBCT scanning, it then back-projects the\nmulti-view 2D features into the 3D space to formulate a comprehensive\nvolumetric feature map, followed by a 3D CNN decoder to recover 3D CBCT image.\nImportantly, our approach respects the geometric relationship between 3D CBCT\nimage and its 2D X-ray projections during feature back projection stage, and\nenjoys the prior knowledge learned from the data population. This ensures its\nadaptability in dealing with extremly sparse view inputs without individual\ntraining, such as scenarios with only 5 or 10 X-ray projections. Extensive\nevaluations on two simulated datasets and one real-world dataset demonstrate\nexceptional reconstruction quality and time efficiency of our method.\n", "date": "2023-03-26", "categories": []}
{"id": "2409.17996", "title": "PhoCoLens: Photorealistic and Consistent Reconstruction in Lensless\n  Imaging", "abstract": "  Lensless cameras offer significant advantages in size, weight, and cost\ncompared to traditional lens-based systems. Without a focusing lens, lensless\ncameras rely on computational algorithms to recover the scenes from multiplexed\nmeasurements. However, current algorithms struggle with inaccurate forward\nimaging models and insufficient priors to reconstruct high-quality images. To\novercome these limitations, we introduce a novel two-stage approach for\nconsistent and photorealistic lensless image reconstruction. The first stage of\nour approach ensures data consistency by focusing on accurately reconstructing\nthe low-frequency content with a spatially varying deconvolution method that\nadjusts to changes in the Point Spread Function (PSF) across the camera's field\nof view. The second stage enhances photorealism by incorporating a generative\nprior from pre-trained diffusion models. By conditioning on the low-frequency\ncontent retrieved in the first stage, the diffusion model effectively\nreconstructs the high-frequency details that are typically lost in the lensless\nimaging process, while also maintaining image fidelity. Our method achieves a\nsuperior balance between data fidelity and visual quality compared to existing\nmethods, as demonstrated with two popular lensless systems, PhlatCam and\nDiffuserCam. Project website: https://phocolens.github.io/.\n", "date": "2024-09-26", "categories": []}
{"id": "2406.10455", "title": "CryoSPIN: Improving Ab-Initio Cryo-EM Reconstruction with Semi-Amortized\n  Pose Inference", "abstract": "  Cryo-EM is an increasingly popular method for determining the atomic\nresolution 3D structure of macromolecular complexes (eg, proteins) from noisy\n2D images captured by an electron microscope. The computational task is to\nreconstruct the 3D density of the particle, along with 3D pose of the particle\nin each 2D image, for which the posterior pose distribution is highly\nmulti-modal. Recent developments in cryo-EM have focused on deep learning for\nwhich amortized inference has been used to predict pose. Here, we address key\nproblems with this approach, and propose a new semi-amortized method, cryoSPIN,\nin which reconstruction begins with amortized inference and then switches to a\nform of auto-decoding to refine poses locally using stochastic gradient\ndescent. Through evaluation on synthetic datasets, we demonstrate that cryoSPIN\nis able to handle multi-modal pose distributions during the amortized inference\nstage, while the later, more flexible stage of direct pose optimization yields\nfaster and more accurate convergence of poses compared to baselines. On\nexperimental data, we show that cryoSPIN outperforms the state-of-the-art\ncryoAI in speed and reconstruction quality.\n", "date": "2024-06-15", "categories": []}
{"id": "2410.02764", "title": "Flash-Splat: 3D Reflection Removal with Flash Cues and Gaussian Splats", "abstract": "  We introduce a simple yet effective approach for separating transmitted and\nreflected light. Our key insight is that the powerful novel view synthesis\ncapabilities provided by modern inverse rendering methods (e.g.,~3D Gaussian\nsplatting) allow one to perform flash/no-flash reflection separation using\nunpaired measurements -- this relaxation dramatically simplifies image\nacquisition over conventional paired flash/no-flash reflection separation\nmethods. Through extensive real-world experiments, we demonstrate our method,\nFlash-Splat, accurately reconstructs both transmitted and reflected scenes in\n3D. Our method outperforms existing 3D reflection separation methods, which do\nnot leverage illumination control, by a large margin. Our project webpage is at\nhttps://flash-splat.github.io/.\n", "date": "2024-10-03", "categories": []}
{"id": "2410.01365", "title": "Anti-biofouling Lensless Camera System with Deep Learning based Image\n  Reconstruction", "abstract": "  In recent years, there has been an increasing demand for underwater cameras\nthat monitor the condition of offshore structures and check the number of\nindividuals in aqua culture environments with long-period observation. One of\nthe significant issues with this observation is that biofouling sticks to the\naperture and lens densely and prevents cameras from capturing clear images.\nThis study examines an underwater camera that applies material technologies\nwith high inherent resistance to biofouling and computer vision technologies\nbased on image reconstruction by deep learning to lens-less cameras. For this\npurpose, our prototype camera uses a coded aperture with 1k rectangular shape\npinholes in a thin metal plate, such as copper, which hinder the growth of\nbiofouling and keep the surface clean. Although images taken by lens-less\ncameras are usually not well formed due to lack of the traditional glass-based\nlens, a deep learning approach using ViT (Vision Transformer) has recently\ndemonstrated reconstructing original photo images well and our study shows that\nusing gated MLP (Multilayer Perceptron) also yields good results. On the other\nhand, a certain degree of thickness for bio-repellence materials is required to\nexhibit their effect the thickness of aperture is necessary to use apertures\nsufficiently thinner than the size of the pinholes to avoid unintentional\nreflection and absorption on the sidewalls. Therefore, we prepared a\nsufficiently thin plate for image reconstruction and now currently we conduct\ntests of the lens-less camera of the bio-repellence aperture with actual\nseawater environments to determine whether it can sufficiently demonstrate the\nbiofouling effect compared with usual camera with only waterproof.\n", "date": "2024-10-02", "categories": []}
{"id": "2404.12361", "title": "Learning the Domain Specific Inverse NUFFT for Accelerated Spiral MRI\n  using Diffusion Models", "abstract": "  Deep learning methods for accelerated MRI achieve state-of-the-art results\nbut largely ignore additional speedups possible with noncartesian sampling\ntrajectories. To address this gap, we created a generative diffusion\nmodel-based reconstruction algorithm for multi-coil highly undersampled spiral\nMRI. This model uses conditioning during training as well as frequency-based\nguidance to ensure consistency between images and measurements. Evaluated on\nretrospective data, we show high quality (structural similarity > 0.87) in\nreconstructed images with ultrafast scan times (0.02 seconds for a 2D image).\nWe use this algorithm to identify a set of optimal variable-density spiral\ntrajectories and show large improvements in image quality compared to\nconventional reconstruction using the non-uniform fast Fourier transform. By\ncombining efficient spiral sampling trajectories, multicoil imaging, and deep\nlearning reconstruction, these methods could enable the extremely high\nacceleration factors needed for real-time 3D imaging.\n", "date": "2024-04-18", "categories": []}
{"id": "2408.07932", "title": "MobileMEF: Fast and Efficient Method for Multi-Exposure Fusion", "abstract": "  Recent advances in camera design and imaging technology have enabled the\ncapture of high-quality images using smartphones. However, due to the limited\ndynamic range of digital cameras, the quality of photographs captured in\nenvironments with highly imbalanced lighting often results in poor-quality\nimages. To address this issue, most devices capture multi-exposure frames and\nthen use some multi-exposure fusion method to merge those frames into a final\nfused image. Nevertheless, most traditional and current deep learning\napproaches are unsuitable for real-time applications on mobile devices due to\ntheir heavy computational and memory requirements. We propose a new method for\nmulti-exposure fusion based on an encoder-decoder deep learning architecture\nwith efficient building blocks tailored for mobile devices. This efficient\ndesign makes our model capable of processing 4K resolution images in less than\n2 seconds on mid-range smartphones. Our method outperforms state-of-the-art\ntechniques regarding full-reference quality measures and computational\nefficiency (runtime and memory usage), making it ideal for real-time\napplications on hardware-constrained devices. Our code is available at:\nhttps://github.com/LucasKirsten/MobileMEF.\n", "date": "2024-08-15", "categories": []}
{"id": "2312.02235", "title": "CryoGEM: Physics-Informed Generative Cryo-Electron Microscopy", "abstract": "  In the past decade, deep conditional generative models have revolutionized\nthe generation of realistic images, extending their application from\nentertainment to scientific domains. Single-particle cryo-electron microscopy\n(cryo-EM) is crucial in resolving near-atomic resolution 3D structures of\nproteins, such as the SARS- COV-2 spike protein. To achieve high-resolution\nreconstruction, a comprehensive data processing pipeline has been adopted.\nHowever, its performance is still limited as it lacks high-quality annotated\ndatasets for training. To address this, we introduce physics-informed\ngenerative cryo-electron microscopy (CryoGEM), which for the first time\nintegrates physics-based cryo-EM simulation with a generative unpaired noise\ntranslation to generate physically correct synthetic cryo-EM datasets with\nrealistic noises. Initially, CryoGEM simulates the cryo-EM imaging process\nbased on a virtual specimen. To generate realistic noises, we leverage an\nunpaired noise translation via contrastive learning with a novel mask-guided\nsampling scheme. Extensive experiments show that CryoGEM is capable of\ngenerating authentic cryo-EM images. The generated dataset can used as training\ndata for particle picking and pose estimation models, eventually improving the\nreconstruction resolution.\n", "date": "2023-12-04", "categories": []}
{"id": "2409.17988", "title": "Deblur e-NeRF: NeRF from Motion-Blurred Events under High-speed or\n  Low-light Conditions", "abstract": "  The stark contrast in the design philosophy of an event camera makes it\nparticularly ideal for operating under high-speed, high dynamic range and\nlow-light conditions, where standard cameras underperform. Nonetheless, event\ncameras still suffer from some amount of motion blur, especially under these\nchallenging conditions, in contrary to what most think. This is attributed to\nthe limited bandwidth of the event sensor pixel, which is mostly proportional\nto the light intensity. Thus, to ensure that event cameras can truly excel in\nsuch conditions where it has an edge over standard cameras, it is crucial to\naccount for event motion blur in downstream applications, especially\nreconstruction. However, none of the recent works on reconstructing Neural\nRadiance Fields (NeRFs) from events, nor event simulators, have considered the\nfull effects of event motion blur. To this end, we propose, Deblur e-NeRF, a\nnovel method to directly and effectively reconstruct blur-minimal NeRFs from\nmotion-blurred events generated under high-speed motion or low-light\nconditions. The core component of this work is a physically-accurate pixel\nbandwidth model proposed to account for event motion blur under arbitrary speed\nand lighting conditions. We also introduce a novel threshold-normalized total\nvariation loss to improve the regularization of large textureless patches.\nExperiments on real and novel realistically simulated sequences verify our\neffectiveness. Our code, event simulator and synthetic event dataset will be\nopen-sourced.\n", "date": "2024-09-26", "categories": []}
{"id": "2409.19220", "title": "Extending Depth of Field for Varifocal Multiview Images", "abstract": "  Optical imaging systems are generally limited by the depth of field because\nof the nature of the optics. Therefore, extending depth of field (EDoF) is a\nfundamental task for meeting the requirements of emerging visual applications.\nTo solve this task, the common practice is using multi-focus images from a\nsingle viewpoint. This method can obtain acceptable quality of EDoF under the\ncondition of fixed field of view, but it is only applicable to static scenes\nand the field of view is limited and fixed. An emerging data type, varifocal\nmultiview images have the potential to become a new paradigm for solving the\nEDoF, because the data contains more field of view information than multi-focus\nimages. To realize EDoF of varifocal multiview images, we propose an end-to-end\nmethod for the EDoF, including image alignment, image optimization and image\nfusion. Experimental results demonstrate the efficiency of the proposed method.\n", "date": "2024-09-28", "categories": []}
{"id": "2409.19701", "title": "Hyperspectral Unmixing of Agricultural Images taken from UAV Using\n  Adapted U-Net Architecture", "abstract": "  The hyperspectral unmixing method is an algorithm that extracts material\n(usually called endmember) data from hyperspectral data cube pixels along with\ntheir abundances. Due to a lower spatial resolution of hyperspectral sensors\ndata in each of the pixels may contain mixed information from multiple\nendmembers. In this paper we create a hyperspectral unmixing dataset, created\nfrom blueberry field data gathered by a hyperspectral camera mounted on a UAV.\nWe also propose a hyperspectral unmixing algorithm based on U-Net network\narchitecture to achieve more accurate unmixing results on existing and newly\ncreated hyperspectral unmixing datasets.\n", "date": "2024-09-29", "categories": []}
{"id": "2409.20013", "title": "Single-shot reconstruction of three-dimensional morphology of biological\n  cells in digital holographic microscopy using a physics-driven neural network", "abstract": "  Recent advances in deep learning-based image reconstruction techniques have\nled to significant progress in phase retrieval using digital in-line\nholographic microscopy (DIHM). However, existing deep learning-based phase\nretrieval methods have technical limitations in generalization performance and\nthree-dimensional (3D) morphology reconstruction from a single-shot hologram of\nbiological cells. In this study, we propose a novel deep learning model, named\nMorpHoloNet, for single-shot reconstruction of 3D morphology by integrating\nphysics-driven and coordinate-based neural networks. By simulating the optical\ndiffraction of coherent light through a 3D phase shift distribution, the\nproposed MorpHoloNet is optimized by minimizing the loss between the simulated\nand input holograms on the sensor plane. Compared to existing DIHM methods that\nface challenges with twin image and phase retrieval problems, MorpHoloNet\nenables direct reconstruction of 3D complex light field and 3D morphology of a\ntest sample from its single-shot hologram without requiring multiple\nphase-shifted holograms or angle scanning. The performance of the proposed\nMorpHoloNet is validated by reconstructing 3D morphologies and refractive index\ndistributions from synthetic holograms of ellipsoids and experimental holograms\nof biological cells. The proposed deep learning model is utilized to\nreconstruct spatiotemporal variations in 3D translational and rotational\nbehaviors and morphological deformations of biological cells from consecutive\nsingle-shot holograms captured using DIHM. MorpHoloNet would pave the way for\nadvancing label-free, real-time 3D imaging and dynamic analysis of biological\ncells under various cellular microenvironments in biomedical and engineering\nfields.\n", "date": "2024-09-30", "categories": []}
{"id": "2310.04964", "title": "Learning Many-to-Many Mapping for Unpaired Real-World Image\n  Super-resolution and Downscaling", "abstract": "  Learning based single image super-resolution (SISR) for real-world images has\nbeen an active research topic yet a challenging task, due to the lack of paired\nlow-resolution (LR) and high-resolution (HR) training images. Most of the\nexisting unsupervised real-world SISR methods adopt a two-stage training\nstrategy by synthesizing realistic LR images from their HR counterparts first,\nthen training the super-resolution (SR) models in a supervised manner. However,\nthe training of image degradation and SR models in this strategy are separate,\nignoring the inherent mutual dependency between downscaling and its inverse\nupscaling process. Additionally, the ill-posed nature of image degradation is\nnot fully considered. In this paper, we propose an image downscaling and SR\nmodel dubbed as SDFlow, which simultaneously learns a bidirectional\nmany-to-many mapping between real-world LR and HR images unsupervisedly. The\nmain idea of SDFlow is to decouple image content and degradation information in\nthe latent space, where content information distribution of LR and HR images is\nmatched in a common latent space. Degradation information of the LR images and\nthe high-frequency information of the HR images are fitted to an easy-to-sample\nconditional distribution. Experimental results on real-world image SR datasets\nindicate that SDFlow can generate diverse realistic LR and SR images both\nquantitatively and qualitatively.\n", "date": "2023-10-08", "categories": []}
{"id": "2409.18223", "title": "PNR: Physics-informed Neural Representation for high-resolution LFM\n  reconstruction", "abstract": "  Light field microscopy (LFM) has been widely utilized in various fields for\nits capability to efficiently capture high-resolution 3D scenes. Despite the\nrapid advancements in neural representations, there are few methods\nspecifically tailored for microscopic scenes. Existing approaches often do not\nadequately address issues such as the loss of high-frequency information due to\ndefocus and sample aberration, resulting in suboptimal performance. In\naddition, existing methods, including RLD, INR, and supervised U-Net, face\nchallenges such as sensitivity to initial estimates, reliance on extensive\nlabeled data, and low computational efficiency, all of which significantly\ndiminish the practicality in complex biological scenarios. This paper\nintroduces PNR (Physics-informed Neural Representation), a method for\nhigh-resolution LFM reconstruction that significantly enhances performance. Our\nmethod incorporates an unsupervised and explicit feature representation\napproach, resulting in a 6.1 dB improvement in PSNR than RLD. Additionally, our\nmethod employs a frequency-based training loss, enabling better recovery of\nhigh-frequency details, which leads to a reduction in LPIPS by at least half\ncompared to SOTA methods (1.762 V.S. 3.646 of DINER). Moreover, PNR integrates\na physics-informed aberration correction strategy that optimizes Zernike\npolynomial parameters during optimization, thereby reducing the information\nloss caused by aberrations and improving spatial resolution. These advancements\nmake PNR a promising solution for long-term high-resolution biological imaging\napplications. Our code and dataset will be made publicly available.\n", "date": "2024-09-26", "categories": []}
{"id": "2409.18337", "title": "Photon Inhibition for Energy-Efficient Single-Photon Imaging", "abstract": "  Single-photon cameras (SPCs) are emerging as sensors of choice for various\nchallenging imaging applications. One class of SPCs based on the single-photon\navalanche diode (SPAD) detects individual photons using an avalanche process;\nthe raw photon data can then be processed to extract scene information under\nextremely low light, high dynamic range, and rapid motion. Yet, single-photon\nsensitivity in SPADs comes at a cost -- each photon detection consumes more\nenergy than that of a CMOS camera. This avalanche power significantly limits\nsensor resolution and could restrict widespread adoption of SPAD-based SPCs. We\npropose a computational-imaging approach called \\emph{photon inhibition} to\naddress this challenge. Photon inhibition strategically allocates detections in\nspace and time based on downstream inference task goals and resource\nconstraints. We develop lightweight, on-sensor computational inhibition\npolicies that use past photon data to disable SPAD pixels in real-time, to\nselect the most informative future photons. As case studies, we design policies\ntailored for image reconstruction and edge detection, and demonstrate, both via\nsimulations and real SPC captured data, considerable reduction in photon\ndetections (over 90\\% of photons) while maintaining task performance metrics.\nOur work raises the question of ``which photons should be detected?'', and\npaves the way for future energy-efficient single-photon imaging.\n", "date": "2024-09-26", "categories": []}
{"id": "2406.04769", "title": "Diffusion-based Generative Image Outpainting for Recovery of\n  FOV-Truncated CT Images", "abstract": "  Field-of-view (FOV) recovery of truncated chest CT scans is crucial for\naccurate body composition analysis, which involves quantifying skeletal muscle\nand subcutaneous adipose tissue (SAT) on CT slices. This, in turn, enables\ndisease prognostication. Here, we present a method for recovering truncated CT\nslices using generative image outpainting. We train a diffusion model and apply\nit to truncated CT slices generated by simulating a small FOV. Our model\nreliably recovers the truncated anatomy and outperforms the previous\nstate-of-the-art despite being trained on 87% less data.\n", "date": "2024-06-07", "categories": []}
{"id": "2409.17830", "title": "Unsupervised Learning Based Multi-Scale Exposure Fusion", "abstract": "  Unsupervised learning based multi-scale exposure fusion (ULMEF) is efficient\nfor fusing differently exposed low dynamic range (LDR) images into a higher\nquality LDR image for a high dynamic range (HDR) scene. Unlike supervised\nlearning, loss functions play a crucial role in the ULMEF. In this paper, novel\nloss functions are proposed for the ULMEF and they are defined by using all the\nimages to be fused and other differently exposed images from the same HDR\nscene. The proposed loss functions can guide the proposed ULMEF to learn more\nreliable information from the HDR scene than existing loss functions which are\ndefined by only using the set of images to be fused. As such, the quality of\nthe fused image is significantly improved. The proposed ULMEF also adopts a\nmulti-scale strategy that includes a multi-scale attention module to\neffectively preserve the scene depth and local contrast in the fused image.\nMeanwhile, the proposed ULMEF can be adopted to achieve exposure interpolation\nand exposure extrapolation. Extensive experiments show that the proposed ULMEF\nalgorithm outperforms state-of-the-art exposure fusion algorithms.\n", "date": "2024-09-26", "categories": []}
{"id": "2404.11569", "title": "Simple Image Signal Processing using Global Context Guidance", "abstract": "  In modern smartphone cameras, the Image Signal Processor (ISP) is the core\nelement that converts the RAW readings from the sensor into perceptually\npleasant RGB images for the end users. The ISP is typically proprietary and\nhandcrafted and consists of several blocks such as white balance, color\ncorrection, and tone mapping. Deep learning-based ISPs aim to transform RAW\nimages into DSLR-like RGB images using deep neural networks. However, most\nlearned ISPs are trained using patches (small regions) due to computational\nlimitations. Such methods lack global context, which limits their efficacy on\nfull-resolution images and harms their ability to capture global properties\nsuch as color constancy or illumination. First, we propose a novel module that\ncan be integrated into any neural ISP to capture the global context information\nfrom the full RAW images. Second, we propose an efficient and simple neural ISP\nthat utilizes our proposed module. Our model achieves state-of-the-art results\non different benchmarks using diverse and real smartphone images.\n", "date": "2024-04-17", "categories": []}
{"id": "2403.15770", "title": "Graph Image Prior for Unsupervised Dynamic Cardiac Cine MRI\n  Reconstruction", "abstract": "  The inductive bias of the convolutional neural network (CNN) can be a strong\nprior for image restoration, which is known as the Deep Image Prior (DIP).\nRecently, DIP is utilized in unsupervised dynamic MRI reconstruction, which\nadopts a generative model from the latent space to the image space. However,\nexisting methods usually use a pyramid-shaped CNN generator shared by all\nframes, embedding the temporal modeling within the latent space, which may\nhamper the model expression capability. In this work, we propose a novel scheme\nfor dynamic MRI representation, named ``Graph Image Prior'' (GIP). GIP adopts a\ntwo-stage generative network in a new modeling methodology, which first employs\nindependent CNNs to recover the image structure for each frame, and then\nexploits the spatio-temporal correlations within the feature space\nparameterized by a graph model. A graph convolutional network is utilized for\nfeature fusion and dynamic image generation. In addition, we devise an ADMM\nalgorithm to alternately optimize the images and the network parameters to\nimprove the reconstruction performance. Experiments were conducted on cardiac\ncine MRI reconstruction, which demonstrate that GIP outperforms compressed\nsensing methods and other DIP-based unsupervised methods, significantly\nreducing the performance gap with state-of-the-art supervised algorithms.\nMoreover, GIP displays superior generalization ability when transferred to a\ndifferent reconstruction setting, without the need for any additional data.\n", "date": "2024-03-23", "categories": []}
{"id": "2409.15916", "title": "Deep convolutional framelets for dose reconstruction in BNCT with\n  Compton camera detector", "abstract": "  Boron Neutron Capture Therapy (BNCT) is an innovative binary form of\nradiation therapy with high selectivity towards cancer tissue based on the\nneutron capture reaction 10B(n,$\\alpha$)7Li, consisting in the exposition of\npatients to neutron beams after administration of a boron compound with\npreferential accumulation in cancer cells. The high linear energy transfer\nproducts of the ensuing reaction deposit their energy at cell level, sparing\nnormal tissue. Although progress in accelerator-based BNCT has led to renewed\ninterest in this cancer treatment modality, in vivo dose monitoring during\ntreatment still remains not feasible and several approaches are under\ninvestigation. While Compton imaging presents various advantages over other\nimaging methods, it typically requires long reconstruction times, comparable\nwith BNCT treatment duration. This study aims to develop deep neural network\nmodels to estimate the dose distribution by using a simulated dataset of BNCT\nCompton camera images. The models pursue the avoidance of the iteration time\nassociated with the maximum-likelihood expectation-maximization algorithm\n(MLEM), enabling a prompt dose reconstruction during the treatment. The U-Net\narchitecture and two variants based on the deep convolutional framelets\nframework have been used for noise and artifacts reduction in few-iterations\nreconstructed images, leading to promising results in terms of reconstruction\naccuracy and processing time.\n", "date": "2024-09-24", "categories": []}
{"id": "2304.01994", "title": "Waving Goodbye to Low-Res: A Diffusion-Wavelet Approach for Image\n  Super-Resolution", "abstract": "  This paper presents a novel Diffusion-Wavelet (DiWa) approach for\nSingle-Image Super-Resolution (SISR). It leverages the strengths of Denoising\nDiffusion Probabilistic Models (DDPMs) and Discrete Wavelet Transformation\n(DWT). By enabling DDPMs to operate in the DWT domain, our DDPM models\neffectively hallucinate high-frequency information for super-resolved images on\nthe wavelet spectrum, resulting in high-quality and detailed reconstructions in\nimage space. Quantitatively, we outperform state-of-the-art diffusion-based\nSISR methods, namely SR3 and SRDiff, regarding PSNR, SSIM, and LPIPS on both\nface (8x scaling) and general (4x scaling) SR benchmarks. Meanwhile, using DWT\nenabled us to use fewer parameters than the compared models: 92M parameters\ninstead of 550M compared to SR3 and 9.3M instead of 12M compared to SRDiff.\nAdditionally, our method outperforms other state-of-the-art generative methods\non classical general SR datasets while saving inference time. Finally, our work\nhighlights its potential for various applications.\n", "date": "2023-04-04", "categories": []}
{"id": "2409.13803", "title": "Intrinsic Single-Image HDR Reconstruction", "abstract": "  The low dynamic range (LDR) of common cameras fails to capture the rich\ncontrast in natural scenes, resulting in loss of color and details in saturated\npixels. Reconstructing the high dynamic range (HDR) of luminance present in the\nscene from single LDR photographs is an important task with many applications\nin computational photography and realistic display of images. The HDR\nreconstruction task aims to infer the lost details using the context present in\nthe scene, requiring neural networks to understand high-level geometric and\nillumination cues. This makes it challenging for data-driven algorithms to\ngenerate accurate and high-resolution results. In this work, we introduce a\nphysically-inspired remodeling of the HDR reconstruction problem in the\nintrinsic domain. The intrinsic model allows us to train separate networks to\nextend the dynamic range in the shading domain and to recover lost color\ndetails in the albedo domain. We show that dividing the problem into two\nsimpler sub-tasks improves performance in a wide variety of photographs.\n", "date": "2024-09-20", "categories": []}
{"id": "2409.14011", "title": "Generalizable Non-Line-of-Sight Imaging with Learnable Physical Priors", "abstract": "  Non-line-of-sight (NLOS) imaging, recovering the hidden volume from indirect\nreflections, has attracted increasing attention due to its potential\napplications. Despite promising results, existing NLOS reconstruction\napproaches are constrained by the reliance on empirical physical priors, e.g.,\nsingle fixed path compensation. Moreover, these approaches still possess\nlimited generalization ability, particularly when dealing with scenes at a low\nsignal-to-noise ratio (SNR). To overcome the above problems, we introduce a\nnovel learning-based solution, comprising two key designs: Learnable Path\nCompensation (LPC) and Adaptive Phasor Field (APF). The LPC applies tailored\npath compensation coefficients to adapt to different objects in the scene,\neffectively reducing light wave attenuation, especially in distant regions.\nMeanwhile, the APF learns the precise Gaussian window of the illumination\nfunction for the phasor field, dynamically selecting the relevant spectrum band\nof the transient measurement. Experimental validations demonstrate that our\nproposed approach, only trained on synthetic data, exhibits the capability to\nseamlessly generalize across various real-world datasets captured by different\nimaging systems and characterized by low SNRs.\n", "date": "2024-09-21", "categories": []}
{"id": "2409.14035", "title": "Implicit Neural Representations for Speed-of-Sound Estimation in\n  Ultrasound", "abstract": "  Accurate estimation of the speed-of-sound (SoS) is important for ultrasound\n(US) image reconstruction techniques and tissue characterization. Various\napproaches have been proposed to calculate SoS, ranging from\ntomography-inspired algorithms like CUTE to convolutional networks, and more\nrecently, physics-informed optimization frameworks based on differentiable\nbeamforming. In this work, we utilize implicit neural representations (INRs)\nfor SoS estimation in US. INRs are a type of neural network architecture that\nencodes continuous functions, such as images or physical quantities, through\nthe weights of a network. Implicit networks may overcome the current\nlimitations of SoS estimation techniques, which mainly arise from the use of\nnon-adaptable and oversimplified physical models of tissue. Moreover,\nconvolutional networks for SoS estimation, usually trained using simulated\ndata, often fail when applied to real tissues due to out-of-distribution and\ndata-shift issues. In contrast, implicit networks do not require extensive\ntraining datasets since each implicit network is optimized for an individual\ndata case. This adaptability makes them suitable for processing US data\ncollected from varied tissues and across different imaging protocols.\n  We evaluated the proposed SoS estimation method based on INRs using data\ncollected from a tissue-mimicking phantom containing four cylindrical\ninclusions, with SoS values ranging from 1480 m/s to 1600 m/s. The inclusions\nwere immersed in a material with an SoS value of 1540 m/s. In experiments, the\nproposed method achieved strong performance, clearly demonstrating the\nusefulness of implicit networks for quantitative US applications.\n", "date": "2024-09-21", "categories": []}
{"id": "2409.14394", "title": "Frequency-regularized Neural Representation Method for Sparse-view\n  Tomographic Reconstruction", "abstract": "  Sparse-view tomographic reconstruction is a pivotal direction for reducing\nradiation dose and augmenting clinical applicability. While many research works\nhave proposed the reconstruction of tomographic images from sparse 2D\nprojections, existing models tend to excessively focus on high-frequency\ninformation while overlooking low-frequency components within the sparse input\nimages. This bias towards high-frequency information often leads to\noverfitting, particularly intense at edges and boundaries in the reconstructed\nslices. In this paper, we introduce the Frequency Regularized Neural\nAttenuation/Activity Field (Freq-NAF) for self-supervised sparse-view\ntomographic reconstruction. Freq-NAF mitigates overfitting by incorporating\nfrequency regularization, directly controlling the visible frequency bands in\nthe neural network input. This approach effectively balances high-frequency and\nlow-frequency information. We conducted numerical experiments on CBCT and SPECT\ndatasets, and our method demonstrates state-of-the-art accuracy.\n", "date": "2024-09-22", "categories": []}
{"id": "2409.14587", "title": "Deep Learning Techniques for Atmospheric Turbulence Removal: A Review", "abstract": "  The influence of atmospheric turbulence on acquired imagery makes image\ninterpretation and scene analysis extremely difficult and reduces the\neffectiveness of conventional approaches for classifying and tracking objects\nof interest in the scene. Restoring a scene distorted by atmospheric turbulence\nis also a challenging problem. The effect, which is caused by random, spatially\nvarying perturbations, makes conventional model-based approaches difficult and,\nin most cases, impractical due to complexity and memory requirements. Deep\nlearning approaches offer faster operation and are capable of implementation on\nsmall devices. This paper reviews the characteristics of atmospheric turbulence\nand its impact on acquired imagery. It compares the performance of various\nstate-of-the-art deep neural networks, including Transformers, SWIN and Mamba,\nwhen used to mitigate spatio-temporal image distortions.\n", "date": "2024-09-03", "categories": []}
{"id": "2403.19376", "title": "NIGHT -- Non-Line-of-Sight Imaging from Indirect Time of Flight Data", "abstract": "  The acquisition of objects outside the Line-of-Sight of cameras is a very\nintriguing but also extremely challenging research topic. Recent works showed\nthe feasibility of this idea exploiting transient imaging data produced by\ncustom direct Time of Flight sensors. In this paper, for the first time, we\ntackle this problem using only data from an off-the-shelf indirect Time of\nFlight sensor without any further hardware requirement. We introduced a Deep\nLearning model able to re-frame the surfaces where light bounces happen as a\nvirtual mirror. This modeling makes the task easier to handle and also\nfacilitates the construction of annotated training data. From the obtained data\nit is possible to retrieve the depth information of the hidden scene. We also\nprovide a first-in-its-kind synthetic dataset for the task and demonstrate the\nfeasibility of the proposed idea over it.\n", "date": "2024-03-28", "categories": []}
{"id": "2409.13690", "title": "Colorful Diffuse Intrinsic Image Decomposition in the Wild", "abstract": "  Intrinsic image decomposition aims to separate the surface reflectance and\nthe effects from the illumination given a single photograph. Due to the\ncomplexity of the problem, most prior works assume a single-color illumination\nand a Lambertian world, which limits their use in illumination-aware image\nediting applications. In this work, we separate an input image into its diffuse\nalbedo, colorful diffuse shading, and specular residual components. We arrive\nat our result by gradually removing first the single-color illumination and\nthen the Lambertian-world assumptions. We show that by dividing the problem\ninto easier sub-problems, in-the-wild colorful diffuse shading estimation can\nbe achieved despite the limited ground-truth datasets. Our extended intrinsic\nmodel enables illumination-aware analysis of photographs and can be used for\nimage editing applications such as specularity removal and per-pixel white\nbalancing.\n", "date": "2024-09-20", "categories": []}
{"id": "2409.12401", "title": "MambaRecon: MRI Reconstruction with Structured State Space Models", "abstract": "  Magnetic Resonance Imaging (MRI) is one of the most important medical imaging\nmodalities as it provides superior resolution of soft tissues, albeit with a\nnotable limitation in scanning speed. The advent of deep learning has catalyzed\nthe development of cutting-edge methods for the expedited reconstruction of MRI\nscans, utilizing convolutional neural networks and, more recently, vision\ntransformers. Recently proposed structured state space models (e.g., Mamba)\nhave gained some traction due to their efficiency and low computational\nrequirements compared to transformer models. We propose an innovative MRI\nreconstruction framework that employs structured state space models at its\ncore, aimed at amplifying both long-range contextual sensitivity and\nreconstruction efficacy. Comprehensive experiments on public brain MRI datasets\nshow that our model sets new benchmarks beating state-of-the-art reconstruction\nbaselines. Code will be available\n(https://github.com/yilmazkorkmaz1/MambaRecon).\n", "date": "2024-09-19", "categories": []}
{"id": "2409.12661", "title": "Manifold Sampling for Differentiable Uncertainty in Radiance Fields", "abstract": "  Radiance fields are powerful and, hence, popular models for representing the\nappearance of complex scenes. Yet, constructing them based on image\nobservations gives rise to ambiguities and uncertainties. We propose a\nversatile approach for learning Gaussian radiance fields with explicit and\nfine-grained uncertainty estimates that impose only little additional cost\ncompared to uncertainty-agnostic training. Our key observation is that\nuncertainties can be modeled as a low-dimensional manifold in the space of\nradiance field parameters that is highly amenable to Monte Carlo sampling.\nImportantly, our uncertainties are differentiable and, thus, allow for\ngradient-based optimization of subsequent captures that optimally reduce\nambiguities. We demonstrate state-of-the-art performance on next-best-view\nplanning tasks, including high-dimensional illumination planning for optimal\nradiance field relighting quality.\n", "date": "2024-09-19", "categories": []}
{"id": "2210.14416", "title": "Residual Back Projection With Untrained Neural Networks", "abstract": "  Background and Objective: The success of neural networks in a number of image\nprocessing tasks has motivated their application in image reconstruction\nproblems in computed tomography (CT). While progress has been made in this\narea, the lack of stability and theoretical guarantees for accuracy, together\nwith the scarcity of high-quality training data for specific imaging domains\npose challenges for many CT applications. In this paper, we present a framework\nfor iterative reconstruction (IR) in CT that leverages the hierarchical\nstructure of neural networks, without the need for training. Our framework\nincorporates this structural information as a deep image prior (DIP), and uses\na novel residual back projection (RBP) connection that forms the basis for our\niterations.\n  Methods: We propose using an untrained U-net in conjunction with a novel\nresidual back projection to minimize an objective function and achieve\nhigh-accuracy reconstruction. In each iteration, the weights of the untrained\nU-net are optimized, and the output of the U-net in the current iteration is\nused to update the input of the U-net in the next iteration through the\naforementioned RBP connection.\n  Results: Experimental results demonstrate that the RBP-DIP framework offers\nimprovements over other state-of-the-art conventional IR methods, as well as\npre-trained and untrained models with similar network structures under multiple\nconditions. These improvements are particularly significant in the few-view,\nlimited-angle, and low-dose imaging configurations.\n  Conclusions: Applying to both parallel and fan beam X-ray imaging, our\nframework shows significant improvement under multiple conditions. Furthermore,\nthe proposed framework requires no training data and can be adjusted on-demand\nto adapt to different conditions (e.g. noise level, geometry, and imaged\nobject).\n", "date": "2022-10-26", "categories": []}
{"id": "2404.01360", "title": "Deep learning phase recovery: data-driven, physics-driven, or combining\n  both?", "abstract": "  Phase recovery, calculating the phase of a light wave from its intensity\nmeasurements, is essential for various applications, such as coherent\ndiffraction imaging, adaptive optics, and biomedical imaging. It enables the\nreconstruction of an object's refractive index distribution or topography as\nwell as the correction of imaging system aberrations. In recent years, deep\nlearning has been proven to be highly effective in addressing phase recovery\nproblems. Two most direct deep learning phase recovery strategies are\ndata-driven (DD) with supervised learning mode and physics-driven (PD) with\nself-supervised learning mode. DD and PD achieve the same goal in different\nways and lack the necessary study to reveal similarities and differences.\nTherefore, in this paper, we comprehensively compare these two deep learning\nphase recovery strategies in terms of time consumption, accuracy,\ngeneralization ability, ill-posedness adaptability, and prior capacity. What's\nmore, we propose a co-driven (CD) strategy of combining datasets and physics\nfor the balance of high- and low-frequency information. The codes for DD, PD,\nand CD are publicly available at https://github.com/kqwang/DLPR.\n", "date": "2024-04-01", "categories": []}
{"id": "2409.12016", "title": "Computational Imaging for Long-Term Prediction of Solar Irradiance", "abstract": "  The occlusion of the sun by clouds is one of the primary sources of\nuncertainties in solar power generation, and is a factor that affects the\nwide-spread use of solar power as a primary energy source. Real-time\nforecasting of cloud movement and, as a result, solar irradiance is necessary\nto schedule and allocate energy across grid-connected photovoltaic systems.\nPrevious works monitored cloud movement using wide-angle field of view imagery\nof the sky. However, such images have poor resolution for clouds that appear\nnear the horizon, which reduces their effectiveness for long term prediction of\nsolar occlusion. Specifically, to be able to predict occlusion of the sun over\nlong time periods, clouds that are near the horizon need to be detected, and\ntheir velocities estimated precisely. To enable such a system, we design and\ndeploy a catadioptric system that delivers wide-angle imagery with uniform\nspatial resolution of the sky over its field of view. To enable prediction over\na longer time horizon, we design an algorithm that uses carefully selected\nspatio-temporal slices of the imagery using estimated wind direction and\nvelocity as inputs. Using ray-tracing simulations as well as a real testbed\ndeployed outdoors, we show that the system is capable of predicting solar\nocclusion as well as irradiance for tens of minutes in the future, which is an\norder of magnitude improvement over prior work.\n", "date": "2024-09-18", "categories": []}
{"id": "2409.12165", "title": "NSSR-DIL: Null-Shot Image Super-Resolution Using Deep Identity Learning", "abstract": "  The present State-of-the-Art (SotA) Image Super-Resolution (ISR) methods\nemploy Deep Learning (DL) techniques using a large amount of image data. The\nprimary limitation to extending the existing SotA ISR works for real-world\ninstances is their computational and time complexities. In this paper, contrary\nto the existing methods, we present a novel and computationally efficient ISR\nalgorithm that is independent of the image dataset to learn the ISR task. The\nproposed algorithm reformulates the ISR task from generating the Super-Resolved\n(SR) images to computing the inverse of the kernels that span the degradation\nspace. We introduce Deep Identity Learning, exploiting the identity relation\nbetween the degradation and inverse degradation models. The proposed approach\nneither relies on the ISR dataset nor on a single input low-resolution (LR)\nimage (like the self-supervised method i.e. ZSSR) to model the ISR task. Hence\nwe term our model as Null-Shot Super-Resolution Using Deep Identity Learning\n(NSSR-DIL). The proposed NSSR-DIL model requires fewer computational resources,\nat least by an order of 10, and demonstrates a competitive performance on\nbenchmark ISR datasets. Another salient aspect of our proposition is that the\nNSSR-DIL framework detours retraining the model and remains the same for\nvarying scale factors like X2, X3, and X4. This makes our highly efficient ISR\nmodel more suitable for real-world applications.\n", "date": "2024-09-17", "categories": []}
{"id": "2409.10725", "title": "Depth from Coupled Optical Differentiation", "abstract": "  We propose depth from coupled optical differentiation, a low-computation\npassive-lighting 3D sensing mechanism. It is based on our discovery that\nper-pixel object distance can be rigorously determined by a coupled pair of\noptical derivatives of a defocused image using a simple, closed-form\nrelationship. Unlike previous depth-from-defocus (DfD) methods that leverage\nspatial derivatives of the image to estimate scene depths, the proposed\nmechanism's use of only optical derivatives makes it significantly more robust\nto noise. Furthermore, unlike many previous DfD algorithms with requirements on\naperture code, this relationship is proved to be universal to a broad range of\naperture codes.\n  We build the first 3D sensor based on depth from coupled optical\ndifferentiation. Its optical assembly includes a deformable lens and a\nmotorized iris, which enables dynamic adjustments to the optical power and\naperture radius. The sensor captures two pairs of images: one pair with a\ndifferential change of optical power and the other with a differential change\nof aperture scale. From the four images, a depth and confidence map can be\ngenerated with only 36 floating point operations per output pixel (FLOPOP),\nmore than ten times lower than the previous lowest passive-lighting depth\nsensing solution to our knowledge. Additionally, the depth map generated by the\nproposed sensor demonstrates more than twice the working range of previous DfD\nmethods while using significantly lower computation.\n", "date": "2024-09-16", "categories": []}
{"id": "2409.10794", "title": "Multi-frequency Electrical Impedance Tomography Reconstruction with\n  Multi-Branch Attention Image Prior", "abstract": "  Multi-frequency Electrical Impedance Tomography (mfEIT) is a promising\nbiomedical imaging technique that estimates tissue conductivities across\ndifferent frequencies. Current state-of-the-art (SOTA) algorithms, which rely\non supervised learning and Multiple Measurement Vectors (MMV), require\nextensive training data, making them time-consuming, costly, and less practical\nfor widespread applications. Moreover, the dependency on training data in\nsupervised MMV methods can introduce erroneous conductivity contrasts across\nfrequencies, posing significant concerns in biomedical applications. To address\nthese challenges, we propose a novel unsupervised learning approach based on\nMulti-Branch Attention Image Prior (MAIP) for mfEIT reconstruction. Our method\nemploys a carefully designed Multi-Branch Attention Network (MBA-Net) to\nrepresent multiple frequency-dependent conductivity images and simultaneously\nreconstructs mfEIT images by iteratively updating its parameters. By leveraging\nthe implicit regularization capability of the MBA-Net, our algorithm can\ncapture significant inter- and intra-frequency correlations, enabling robust\nmfEIT reconstruction without the need for training data. Through simulation and\nreal-world experiments, our approach demonstrates performance comparable to, or\nbetter than, SOTA algorithms while exhibiting superior generalization\ncapability. These results suggest that the MAIP-based method can be used to\nimprove the reliability and applicability of mfEIT in various settings.\n", "date": "2024-09-17", "categories": []}
{"id": "2212.14181", "title": "Efficient Image Super-Resolution with Feature Interaction Weighted\n  Hybrid Network", "abstract": "  Lightweight image super-resolution aims to reconstruct high-resolution images\nfrom low-resolution images using low computational costs. However, existing\nmethods result in the loss of middle-layer features due to activation\nfunctions. To minimize the impact of intermediate feature loss on\nreconstruction quality, we propose a Feature Interaction Weighted Hybrid\nNetwork (FIWHN), which comprises a series of Wide-residual Distillation\nInteraction Block (WDIB) as the backbone. Every third WDIB forms a Feature\nShuffle Weighted Group (FSWG) by applying mutual information shuffle and\nfusion. Moreover, to mitigate the negative effects of intermediate feature\nloss, we introduce Wide Residual Weighting units within WDIB. These units\neffectively fuse features of varying levels of detail through a Wide-residual\nDistillation Connection (WRDC) and a Self-Calibrating Fusion (SCF). To\ncompensate for global feature deficiencies, we incorporate a Transformer and\nexplore a novel architecture to combine CNN and Transformer. We show that our\nFIWHN achieves a favorable balance between performance and efficiency through\nextensive experiments on low-level and high-level tasks. Codes will be\navailable at \\url{https://github.com/IVIPLab/FIWHN}.\n", "date": "2022-12-29", "categories": []}
{"id": "2409.08551", "title": "Think Twice Before You Act: Improving Inverse Problem Solving With MCMC", "abstract": "  Recent studies demonstrate that diffusion models can serve as a strong prior\nfor solving inverse problems. A prominent example is Diffusion Posterior\nSampling (DPS), which approximates the posterior distribution of data given the\nmeasure using Tweedie's formula. Despite the merits of being versatile in\nsolving various inverse problems without re-training, the performance of DPS is\nhindered by the fact that this posterior approximation can be inaccurate\nespecially for high noise levels. Therefore, we propose \\textbf{D}iffusion\n\\textbf{P}osterior \\textbf{MC}MC (\\textbf{DPMC}), a novel inference algorithm\nbased on Annealed MCMC to solve inverse problems with pretrained diffusion\nmodels. We define a series of intermediate distributions inspired by the\napproximated conditional distributions used by DPS. Through annealed MCMC\nsampling, we encourage the samples to follow each intermediate distribution\nmore closely before moving to the next distribution at a lower noise level, and\ntherefore reduce the accumulated error along the path. We test our algorithm in\nvarious inverse problems, including super resolution, Gaussian deblurring,\nmotion deblurring, inpainting, and phase retrieval. Our algorithm outperforms\nDPS with less number of evaluations across nearly all tasks, and is competitive\namong existing approaches.\n", "date": "2024-09-13", "categories": []}
{"id": "2212.09129", "title": "SUCRe: Leveraging Scene Structure for Underwater Color Restoration", "abstract": "  Underwater images are altered by the physical characteristics of the medium\nthrough which light rays pass before reaching the optical sensor. Scattering\nand wavelength-dependent absorption significantly modify the captured colors\ndepending on the distance of observed elements to the image plane. In this\npaper, we aim to recover an image of the scene as if the water had no effect on\nlight propagation. We introduce SUCRe, a novel method that exploits the scene's\n3D structure for underwater color restoration. By following points in multiple\nimages and tracking their intensities at different distances to the sensor, we\nconstrain the optimization of the parameters in an underwater image formation\nmodel and retrieve unattenuated pixel intensities. We conduct extensive\nquantitative and qualitative analyses of our approach in a variety of scenarios\nranging from natural light to deep-sea environments using three underwater\ndatasets acquired from real-world scenarios and one synthetic dataset. We also\ncompare the performance of the proposed approach with that of a wide range of\nexisting state-of-the-art methods. The results demonstrate a consistent benefit\nof exploiting multiple views across a spectrum of objective metrics. Our code\nis publicly available at https://github.com/clementinboittiaux/sucre.\n", "date": "2022-12-18", "categories": []}
{"id": "2409.07457", "title": "LSST: Learned Single-Shot Trajectory and Reconstruction Network for MR\n  Imaging", "abstract": "  Single-shot magnetic resonance (MR) imaging acquires the entire k-space data\nin a single shot and it has various applications in whole-body imaging.\nHowever, the long acquisition time for the entire k-space in single-shot fast\nspin echo (SSFSE) MR imaging poses a challenge, as it introduces T2-blur in the\nacquired images. This study aims to enhance the reconstruction quality of SSFSE\nMR images by (a) optimizing the trajectory for measuring the k-space, (b)\nacquiring fewer samples to speed up the acquisition process, and (c) reducing\nthe impact of T2-blur. The proposed method adheres to physics constraints due\nto maximum gradient strength and slew-rate available while optimizing the\ntrajectory within an end-to-end learning framework. Experiments were conducted\non publicly available fastMRI multichannel dataset with 8-fold and 16-fold\nacceleration factors. An experienced radiologist's evaluation on a five-point\nLikert scale indicates improvements in the reconstruction quality as the ACL\nfibers are sharper than comparative methods.\n", "date": "2024-08-08", "categories": []}
{"id": "2409.07171", "title": "AC-IND: Sparse CT reconstruction based on attenuation coefficient\n  estimation and implicit neural distribution", "abstract": "  Computed tomography (CT) reconstruction plays a crucial role in industrial\nnondestructive testing and medical diagnosis. Sparse view CT reconstruction\naims to reconstruct high-quality CT images while only using a small number of\nprojections, which helps to improve the detection speed of industrial assembly\nlines and is also meaningful for reducing radiation in medical scenarios.\nSparse CT reconstruction methods based on implicit neural representations\n(INRs) have recently shown promising performance, but still produce artifacts\nbecause of the difficulty of obtaining useful prior information. In this work,\nwe incorporate a powerful prior: the total number of material categories of\nobjects. To utilize the prior, we design AC-IND, a self-supervised method based\non Attenuation Coefficient Estimation and Implicit Neural Distribution.\nSpecifically, our method first transforms the traditional INR from scalar\nmapping to probability distribution mapping. Then we design a compact\nattenuation coefficient estimator initialized with values from a rough\nreconstruction and fast segmentation. Finally, our algorithm finishes the CT\nreconstruction by jointly optimizing the estimator and the generated\ndistribution. Through experiments, we find that our method not only outperforms\nthe comparative methods in sparse CT reconstruction but also can automatically\ngenerate semantic segmentation maps.\n", "date": "2024-09-11", "categories": []}
{"id": "2307.12297", "title": "Simultaneous temperature estimation and nonuniformity correction from\n  multiple frames", "abstract": "  IR cameras are widely used for temperature measurements in various\napplications, including agriculture, medicine, and security. Low-cost IR\ncameras have the immense potential to replace expensive radiometric cameras in\nthese applications; however, low-cost microbolometer-based IR cameras are prone\nto spatially variant nonuniformity and to drift in temperature measurements,\nwhich limit their usability in practical scenarios.\n  To address these limitations, we propose a novel approach for simultaneous\ntemperature estimation and nonuniformity correction (NUC) from multiple frames\ncaptured by low-cost microbolometer-based IR cameras. We leverage the camera's\nphysical image-acquisition model and incorporate it into a deep-learning\narchitecture termed kernel prediction network (KPN), which enables us to\ncombine multiple frames despite imperfect registration between them. We also\npropose a novel offset block that incorporates the ambient temperature into the\nmodel and enables us to estimate the offset of the camera, which is a key\nfactor in temperature estimation.\n  Our findings demonstrate that the number of frames has a significant impact\non the accuracy of the temperature estimation and NUC. Moreover, introduction\nof the offset block results in significantly improved performance compared to\nvanilla KPN. The method was tested on real data collected by a low-cost IR\ncamera mounted on an unmanned aerial vehicle, showing only a small average\nerror of $0.27-0.54^\\circ C$ relative to costly scientific-grade radiometric\ncameras.\n  Our method provides an accurate and efficient solution for simultaneous\ntemperature estimation and NUC, which has important implications for a wide\nrange of practical applications.\n", "date": "2023-07-23", "categories": []}
{"id": "2409.06334", "title": "Multi-Weather Image Restoration via Histogram-Based Transformer Feature\n  Enhancement", "abstract": "  Currently, the mainstream restoration tasks under adverse weather conditions\nhave predominantly focused on single-weather scenarios. However, in reality,\nmultiple weather conditions always coexist and their degree of mixing is\nusually unknown. Under such complex and diverse weather conditions,\nsingle-weather restoration models struggle to meet practical demands. This is\nparticularly critical in fields such as autonomous driving, where there is an\nurgent need for a model capable of effectively handling mixed weather\nconditions and enhancing image quality in an automated manner. In this paper,\nwe propose a Task Sequence Generator module that, in conjunction with the Task\nIntra-patch Block, effectively extracts task-specific features embedded in\ndegraded images. The Task Intra-patch Block introduces an external learnable\nsequence that aids the network in capturing task-specific information.\nAdditionally, we employ a histogram-based transformer module as the backbone of\nour network, enabling the capture of both global and local dynamic range\nfeatures. Our proposed model achieves state-of-the-art performance on public\ndatasets.\n", "date": "2024-09-10", "categories": []}
{"id": "2307.12130", "title": "Estimating temperatures with low-cost infrared cameras using deep neural\n  networks", "abstract": "  Low-cost thermal cameras are inaccurate (usually $\\pm 3^\\circ C$) and have\nspace-variant nonuniformity across their detector. Both inaccuracy and\nnonuniformity are dependent on the ambient temperature of the camera. The goal\nof this work was to estimate temperatures with low-cost infrared cameras, and\nrectify the nonuniformity.\n  A nonuniformity simulator that accounts for the ambient temperature was\ndeveloped. An end-to-end neural network that incorporates both the physical\nmodel of the camera and the ambient camera temperature was introduced. The\nneural network was trained with the simulated nonuniformity data to estimate\nthe object's temperature and correct the nonuniformity, using only a single\nimage and the ambient temperature measured by the camera itself. Results of the\nproposed method significantly improved the mean temperature error compared to\nprevious works by up to $0.5^\\circ C$. In addition, constraining the physical\nmodel of the camera with the network lowered the error by an additional\n$0.1^\\circ C$.\n  The mean temperature error over an extensive validation dataset was\n$0.37^\\circ C$. The method was verified on real data in the field and produced\nequivalent results.\n", "date": "2023-07-22", "categories": []}
{"id": "2407.06064", "title": "Pan-denoising: Guided Hyperspectral Image Denoising via Weighted\n  Represent Coefficient Total Variation", "abstract": "  This paper introduces a novel paradigm for hyperspectral image (HSI)\ndenoising, which is termed \\textit{pan-denoising}. In a given scene,\npanchromatic (PAN) images capture similar structures and textures to HSIs but\nwith less noise. This enables the utilization of PAN images to guide the HSI\ndenoising process. Consequently, pan-denoising, which incorporates an\nadditional prior, has the potential to uncover underlying structures and\ndetails beyond the internal information modeling of traditional HSI denoising\nmethods. However, the proper modeling of this additional prior poses a\nsignificant challenge. To alleviate this issue, the paper proposes a novel\nregularization term, Panchromatic Weighted Representation Coefficient Total\nVariation (PWRCTV). It employs the gradient maps of PAN images to automatically\nassign different weights of TV regularization for each pixel, resulting in\nlarger weights for smooth areas and smaller weights for edges. This\nregularization forms the basis of a pan-denoising model, which is solved using\nthe Alternating Direction Method of Multipliers. Extensive experiments on\nsynthetic and real-world datasets demonstrate that PWRCTV outperforms several\nstate-of-the-art methods in terms of metrics and visual quality. Furthermore,\nan HSI classification experiment confirms that PWRCTV, as a preprocessing\nmethod, can enhance the performance of downstream classification tasks. The\ncode and data are available at https://github.com/shuangxu96/PWRCTV.\n", "date": "2024-07-08", "categories": []}
{"id": "2409.04733", "title": "A Sample Efficient Alternating Minimization-based Algorithm For Robust\n  Phase Retrieval", "abstract": "  In this work, we study the robust phase retrieval problem where the task is\nto recover an unknown signal $\\theta^* \\in \\mathbb{R}^d$ in the presence of\npotentially arbitrarily corrupted magnitude-only linear measurements. We\npropose an alternating minimization approach that incorporates an oracle solver\nfor a non-convex optimization problem as a subroutine. Our algorithm guarantees\nconvergence to $\\theta^*$ and provides an explicit polynomial dependence of the\nconvergence rate on the fraction of corrupted measurements. We then provide an\nefficient construction of the aforementioned oracle under a sparse arbitrary\noutliers model and offer valuable insights into the geometric properties of the\nloss landscape in phase retrieval with corrupted measurements. Our proposed\noracle avoids the need for computationally intensive spectral initialization,\nusing a simple gradient descent algorithm with a constant step size and random\ninitialization instead. Additionally, our overall algorithm achieves nearly\nlinear sample complexity, $\\mathcal{O}(d \\, \\mathrm{polylog}(d))$.\n", "date": "2024-09-07", "categories": []}
{"id": "2301.06084", "title": "Scattering-induced entropy boost for highly-compressed optical sensing\n  and encryption", "abstract": "  Image sensing often relies on a high-quality machine vision system with a\nlarge field of view and high resolution. It requires fine imaging optics, has\nhigh computational costs, and requires a large communication bandwidth between\nimage sensors and computing units. In this paper, we propose a novel image-free\nsensing framework for resource-efficient image classification, where the\nrequired number of measurements can be reduced by up to two orders of\nmagnitude. In the proposed framework for single-pixel detection, the optical\nfield for a target is first scattered by an optical diffuser and then\ntwo-dimensionally modulated by a spatial light modulator. The optical diffuser\nsimultaneously serves as a compressor and an encryptor for the target\ninformation, effectively narrowing the field of view and improving the system's\nsecurity. The one-dimensional sequence of intensity values, which is measured\nwith time-varying patterns on the spatial light modulator, is then used to\nextract semantic information based on end-to-end deep learning. The proposed\nsensing framework is shown to obtain over a 95\\% accuracy at sampling rates of\n1% and 5% for classification on the MNIST dataset and the recognition of\nChinese license plates, respectively, and the framework is up to 24% more\nefficient than the approach without an optical diffuser. The proposed framework\nrepresents a significant breakthrough in high-throughput machine intelligence\nfor scene analysis with low bandwidth, low costs, and strong encryption.\n", "date": "2022-12-16", "categories": []}
{"id": "2312.02344", "title": "STEREOFOG -- Computational DeFogging via Image-to-Image Translation on a\n  real-world Dataset", "abstract": "  Image-to-Image translation (I2I) is a subtype of Machine Learning (ML) that\nhas tremendous potential in applications where two domains of images and the\nneed for translation between the two exist, such as the removal of fog. For\nexample, this could be useful for autonomous vehicles, which currently struggle\nwith adverse weather conditions like fog. However, datasets for I2I tasks are\nnot abundant and typically hard to acquire. Here, we introduce STEREOFOG, a\ndataset comprised of $10,067$ paired fogged and clear images, captured using a\ncustom-built device, with the purpose of exploring I2I's potential in this\ndomain. It is the only real-world dataset of this kind to the best of our\nknowledge. Furthermore, we apply and optimize the pix2pix I2I ML framework to\nthis dataset. With the final model achieving an average Complex\nWavelet-Structural Similarity (CW-SSIM) score of $0.76$, we prove the\ntechnique's suitability for the problem.\n", "date": "2023-12-04", "categories": []}
{"id": "2407.13520", "title": "EaDeblur-GS: Event assisted 3D Deblur Reconstruction with Gaussian\n  Splatting", "abstract": "  3D deblurring reconstruction techniques have recently seen significant\nadvancements with the development of Neural Radiance Fields (NeRF) and 3D\nGaussian Splatting (3DGS). Although these techniques can recover relatively\nclear 3D reconstructions from blurry image inputs, they still face limitations\nin handling severe blurring and complex camera motion. To address these issues,\nwe propose Event-assisted 3D Deblur Reconstruction with Gaussian Splatting\n(EaDeblur-GS), which integrates event camera data to enhance the robustness of\n3DGS against motion blur. By employing an Adaptive Deviation Estimator (ADE)\nnetwork to estimate Gaussian center deviations and using novel loss functions,\nEaDeblur-GS achieves sharp 3D reconstructions in real-time, demonstrating\nperformance comparable to state-of-the-art methods.\n", "date": "2024-07-18", "categories": []}
{"id": "2409.02406", "title": "Hadamard Row-Wise Generation Algorithm", "abstract": "  In this paper, we introduce an efficient algorithm for generating specific\nHadamard rows, addressing the memory demands of pre-computing the entire\nmatrix. Leveraging Sylvester's recursive construction, our method generates the\nrequired $i$-th row on demand, significantly reducing computational resources.\nThe algorithm uses the Kronecker product to construct the desired row from the\nbinary representation of the index, without creating the full matrix. This\napproach is particularly useful for single-pixel imaging systems that need only\none row at a time.\n", "date": "2024-09-04", "categories": []}
{"id": "2409.02716", "title": "LIPIDS: Learning-based Illumination Planning In Discretized (Light)\n  Space for Photometric Stereo", "abstract": "  Photometric stereo is a powerful method for obtaining per-pixel surface\nnormals from differently illuminated images of an object. While several methods\naddress photometric stereo with different image (or light) counts ranging from\none to two to a hundred, very few focus on learning optimal lighting\nconfiguration. Finding an optimal configuration is challenging due to the vast\nnumber of possible lighting directions. Moreover, exhaustively sampling all\npossibilities is impractical due to time and resource constraints. Photometric\nstereo methods have demonstrated promising performance on existing datasets,\nwhich feature limited light directions sparsely sampled from the light space.\nTherefore, can we optimally utilize these datasets for illumination planning?\nIn this work, we introduce LIPIDS - Learning-based Illumination Planning In\nDiscretized light Space to achieve minimal and optimal lighting configurations\nfor photometric stereo under arbitrary light distribution. We propose a Light\nSampling Network (LSNet) that optimizes lighting direction for a fixed number\nof lights by minimizing the normal loss through a normal regression network.\nThe learned light configurations can directly estimate surface normals during\ninference, even using an off-the-shelf photometric stereo method. Extensive\nqualitative and quantitative analyses on synthetic and real-world datasets show\nthat photometric stereo under learned lighting configurations through LIPIDS\neither surpasses or is nearly comparable to existing illumination planning\nmethods across different photometric stereo backbones.\n", "date": "2024-09-01", "categories": []}
{"id": "2409.00028", "title": "Pupil-Adaptive 3D Holography Beyond Coherent Depth-of-Field", "abstract": "  Recent holographic display approaches propelled by deep learning have shown\nremarkable success in enabling high-fidelity holographic projections. However,\nthese displays have still not been able to demonstrate realistic focus cues,\nand a major gap still remains between the defocus effects possible with a\ncoherent light-based holographic display and those exhibited by incoherent\nlight in the real world. Moreover, existing methods have not considered the\neffects of the observer's eye pupil size variations on the perceived quality of\n3D projections, especially on the defocus blur due to varying depth-of-field of\nthe eye.\n  In this work, we propose a framework that bridges the gap between the\ncoherent depth-of-field of holographic displays and what is seen in the real\nworld due to incoherent light. To this end, we investigate the effect of\nvarying shape and motion of the eye pupil on the quality of holographic\nprojections, and devise a method that changes the depth-of-the-field of\nholographic projections dynamically in a pupil-adaptive manner. Specifically,\nwe introduce a learning framework that adjusts the receptive fields on-the-go\nbased on the current state of the observer's eye pupil to produce image effects\nthat otherwise are not possible in current computer-generated holography\napproaches. We validate the proposed method both in simulations and on an\nexperimental prototype holographic display, and demonstrate significant\nimprovements in the depiction of depth-of-field effects, outperforming existing\napproaches both qualitatively and quantitatively by at least 5 dB in peak\nsignal-to-noise ratio.\n", "date": "2024-08-17", "categories": []}
{"id": "2409.01274", "title": "DAVIDE: Depth-Aware Video Deblurring", "abstract": "  Video deblurring aims at recovering sharp details from a sequence of blurry\nframes. Despite the proliferation of depth sensors in mobile phones and the\npotential of depth information to guide deblurring, depth-aware deblurring has\nreceived only limited attention. In this work, we introduce the 'Depth-Aware\nVIdeo DEblurring' (DAVIDE) dataset to study the impact of depth information in\nvideo deblurring. The dataset comprises synchronized blurred, sharp, and depth\nvideos. We investigate how the depth information should be injected into the\nexisting deep RGB video deblurring models, and propose a strong baseline for\ndepth-aware video deblurring. Our findings reveal the significance of depth\ninformation in video deblurring and provide insights into the use cases where\ndepth cues are beneficial. In addition, our results demonstrate that while the\ndepth improves deblurring performance, this effect diminishes when models are\nprovided with a longer temporal context. Project page:\nhttps://germanftv.github.io/DAVIDE.github.io/ .\n", "date": "2024-09-02", "categories": []}
{"id": "2409.02056", "title": "F2former: When Fractional Fourier Meets Deep Wiener Deconvolution and\n  Selective Frequency Transformer for Image Deblurring", "abstract": "  Recent progress in image deblurring techniques focuses mainly on operating in\nboth frequency and spatial domains using the Fourier transform (FT) properties.\nHowever, their performance is limited due to the dependency of FT on stationary\nsignals and its lack of capability to extract spatial-frequency properties. In\nthis paper, we propose a novel approach based on the Fractional Fourier\nTransform (FRFT), a unified spatial-frequency representation leveraging both\nspatial and frequency components simultaneously, making it ideal for processing\nnon-stationary signals like images. Specifically, we introduce a Fractional\nFourier Transformer (F2former), where we combine the classical fractional\nFourier based Wiener deconvolution (F2WD) as well as a multi-branch\nencoder-decoder transformer based on a new fractional frequency aware\ntransformer block (F2TB). We design F2TB consisting of a fractional frequency\naware self-attention (F2SA) to estimate element-wise product attention based on\nimportant frequency components and a novel feed-forward network based on\nfrequency division multiplexing (FM-FFN) to refine high and low frequency\nfeatures separately for efficient latent clear image restoration. Experimental\nresults for the cases of both motion deblurring as well as defocus deblurring\nshow that the performance of our proposed method is superior to other\nstate-of-the-art (SOTA) approaches.\n", "date": "2024-09-03", "categories": []}
{"id": "2209.12101", "title": "3D Reconstruction using Structured Light from off-the-shelf components", "abstract": "  The coordinate measuring machine(CMM) has been the benchmark of accuracy in\nmeasuring solid objects from nearly past 50 years or more. However with the\nadvent of 3D scanning technology, the accuracy and the density of point cloud\ngenerated has taken over. In this project we not only compare the different\nalgorithms that can be used in a 3D scanning software, but also create our own\n3D scanner from off-the-shelf components like camera and projector. Our\nobjective has been : 1. To develop a prototype for 3D scanner to achieve a\nsystem that performs at optimal accuracy over a wide typology of objects. 2. To\nminimise the cost using off-the-shelf components. 3. To reach very close to the\naccuracy of CMM.\n", "date": "2022-09-24", "categories": []}
{"id": "2309.01159", "title": "An Asynchronous Linear Filter Architecture for Hybrid Event-Frame\n  Cameras", "abstract": "  Event cameras are ideally suited to capture High Dynamic Range (HDR) visual\ninformation without blur but provide poor imaging capability for static or\nslowly varying scenes. Conversely, conventional image sensors measure absolute\nintensity of slowly changing scenes effectively but do poorly on HDR or quickly\nchanging scenes. In this paper, we present an asynchronous linear filter\narchitecture, fusing event and frame camera data, for HDR video reconstruction\nand spatial convolution that exploits the advantages of both sensor modalities.\nThe key idea is the introduction of a state that directly encodes the\nintegrated or convolved image information and that is updated asynchronously as\neach event or each frame arrives from the camera. The state can be read-off\nas-often-as and whenever required to feed into subsequent vision modules for\nreal-time robotic systems. Our experimental results are evaluated on both\npublicly available datasets with challenging lighting conditions and fast\nmotions, along with a new dataset with HDR reference that we provide. The\nproposed AKF pipeline outperforms other state-of-the-art methods in both\nabsolute intensity error (69.4% reduction) and image similarity indexes\n(average 35.5% improvement). We also demonstrate the integration of image\nconvolution with linear spatial kernels Gaussian, Sobel, and Laplacian as an\napplication of our architecture.\n", "date": "2023-09-03", "categories": []}
{"id": "2408.17060", "title": "Efficient Image Restoration through Low-Rank Adaptation and Stable\n  Diffusion XL", "abstract": "  In this study, we propose an enhanced image restoration model, SUPIR, based\non the integration of two low-rank adaptive (LoRA) modules with the Stable\nDiffusion XL (SDXL) framework. Our method leverages the advantages of LoRA to\nfine-tune SDXL models, thereby significantly improving image restoration\nquality and efficiency. We collect 2600 high-quality real-world images, each\nwith detailed descriptive text, for training the model. The proposed method is\nevaluated on standard benchmarks and achieves excellent performance,\ndemonstrated by higher peak signal-to-noise ratio (PSNR), lower learned\nperceptual image patch similarity (LPIPS), and higher structural similarity\nindex measurement (SSIM) scores. These results underscore the effectiveness of\ncombining LoRA with SDXL for advanced image restoration tasks, highlighting the\npotential of our approach in generating high-fidelity restored images.\n", "date": "2024-08-30", "categories": []}
{"id": "2302.02688", "title": "HyperSLICE: HyperBand optimized Spiral for Low-latency Interactive\n  Cardiac Examination", "abstract": "  PURPOSE: Interactive cardiac magnetic resonance imaging is used for fast scan\nplanning and MR guided interventions. However, the requirement for real-time\nacquisition and near real-time visualization constrains the achievable\nspatio-temporal resolution. This study aims to improve interactive imaging\nresolution through optimization of undersampled spiral sampling and leveraging\nof deep learning for low-latency reconstruction (deep artifact suppression).\nMETHODS: A variable density spiral trajectory was parametrized and optimized\nvia HyperBand to provide the best candidate trajectory for rapid deep artifact\nsuppression. Training data consisted of 692 breath-held CINEs. The developed\ninteractive sequence was tested in simulations and prospectively in 13 subjects\n(10 for image evaluation, 2 during catheterization, 1 during exercise). In the\nprospective study, the optimized framework -HyperSLICE- was compared to\nconventional Cartesian real-time, and breath-hold CINE imaging in terms\nquantitative and qualitative image metrics. Statistical differences were tested\nusing Friedman chi-squared tests with post-hoc Nemenyi test (p<0.05). RESULTS:\nIn simulations the NRMSE, pSNR, SSIM and LAPE were all statistically\nsignificantly higher using optimized spiral compared to radial and uniform\nspiral sampling, particularly after scan plan changes (SSIM: 0.71 vs 0.45 and\n0.43). Prospectively, HyperSLICE enabled a higher spatial and temporal\nresolution than conventional Cartesian real-time imaging. The pipeline was\ndemonstrated in patients during catheter pull back showing sufficiently fast\nreconstruction for interactive imaging. CONCLUSION: HyperSLICE enables high\nspatial and temporal resolution interactive imaging. Optimizing the spiral\nsampling enabled better overall image quality and superior handling of image\ntransitions compared to radial and uniform spiral trajectories.\n", "date": "2023-02-06", "categories": []}
{"id": "2403.06831", "title": "HDRTransDC: High Dynamic Range Image Reconstruction with Transformer\n  Deformation Convolution", "abstract": "  High Dynamic Range (HDR) imaging aims to generate an artifact-free HDR image\nwith realistic details by fusing multi-exposure Low Dynamic Range (LDR) images.\nCaused by large motion and severe under-/over-exposure among input LDR images,\nHDR imaging suffers from ghosting artifacts and fusion distortions. To address\nthese critical issues, we propose an HDR Transformer Deformation Convolution\n(HDRTransDC) network to generate high-quality HDR images, which consists of the\nTransformer Deformable Convolution Alignment Module (TDCAM) and the Dynamic\nWeight Fusion Block (DWFB). To solve the ghosting artifacts, the proposed TDCAM\nextracts long-distance content similar to the reference feature in the entire\nnon-reference features, which can accurately remove misalignment and fill the\ncontent occluded by moving objects. For the purpose of eliminating fusion\ndistortions, we propose DWFB to spatially adaptively select useful information\nacross frames to effectively fuse multi-exposed features. Extensive experiments\nshow that our method quantitatively and qualitatively achieves state-of-the-art\nperformance.\n", "date": "2024-03-11", "categories": []}
{"id": "2408.16117", "title": "Alternating Direction Method of Multipliers for Negative Binomial Model\n  with The Weighted Difference of Anisotropic and Isotropic Total Variation", "abstract": "  In many applications such as medical imaging, the measurement data represent\ncounts of photons hitting a detector. Such counts in low-photon settings are\noften modeled using a Poisson distribution. However, this model assumes that\nthe mean and variance of the signal's noise distribution are equal. For\noverdispersed data where the variance is greater than the mean, the negative\nbinomial distribution is a more appropriate statistical model. In this paper,\nwe propose an optimization approach for recovering images corrupted by\noverdispersed Poisson noise. In particular, we incorporate a weighted\nanisotropic-isotropic total variation regularizer, which avoids staircasing\nartifacts that are introduced by a regular total variation penalty. We use an\nalternating direction method of multipliers, where each subproblem has a\nclosed-form solution. Numerical experiments demonstrate the effectiveness of\nour proposed approach, especially in very photon-limited settings.\n", "date": "2024-08-28", "categories": []}
{"id": "2408.16150", "title": "Single-Photon 3D Imaging with Equi-Depth Photon Histograms", "abstract": "  Single-photon cameras present a promising avenue for high-resolution 3D\nimaging. They have ultra-high sensitivity -- down to individual photons -- and\ncan record photon arrival times with extremely high (sub-nanosecond)\nresolution. Single-photon 3D cameras estimate the round-trip time of a laser\npulse by forming equi-width (EW) histograms of detected photon timestamps.\nAcquiring and transferring such EW histograms requires high bandwidth and\nin-pixel memory, making SPCs less attractive in resource-constrained settings\nsuch as mobile devices and AR/VR headsets. In this work we propose a 3D sensing\ntechnique based on equi-depth (ED) histograms. ED histograms compress timestamp\ndata more efficiently than EW histograms, reducing the bandwidth requirement.\nMoreover, to reduce the in-pixel memory requirement, we propose a lightweight\nalgorithm to estimate ED histograms in an online fashion without explicitly\nstoring the photon timestamps. This algorithm is amenable to future in-pixel\nimplementations. We propose algorithms that process ED histograms to perform 3D\ncomputer-vision tasks of estimating scene distance maps and performing visual\nodometry under challenging conditions such as high ambient light. Our work\npaves the way towards lower bandwidth and reduced in-pixel memory requirements\nfor SPCs, making them attractive for resource-constrained 3D vision\napplications. Project page:\n$\\href{https://www.computational.camera/pedh}{https://www.computational.camera/pedh}$\n", "date": "2024-08-28", "categories": []}
{"id": "2408.16622", "title": "Sparse Signal Reconstruction for Overdispersed Low-photon Count\n  Biomedical Imaging Using $\\ell_p$ Total Variation", "abstract": "  The negative binomial model, which generalizes the Poisson distribution\nmodel, can be found in applications involving low-photon signal recovery,\nincluding medical imaging. Recent studies have explored several regularization\nterms for the negative binomial model, such as the $\\ell_p$ quasi-norm with $0\n< p < 1$, $\\ell_1$ norm, and the total variation (TV) quasi-seminorm for\npromoting sparsity in signal recovery. These penalty terms have been shown to\nimprove image reconstruction outcomes. In this paper, we investigate the\n$\\ell_p$ quasi-seminorm, both isotropic and anisotropic $\\ell_p$ TV\nquasi-seminorms, within the framework of the negative binomial statistical\nmodel. This problem can be formulated as an optimization problem, which we\nsolve using a gradient-based approach. We present comparisons between the\nnegative binomial and Poisson statistical models using the $\\ell_p$ TV\nquasi-seminorm as well as common penalty terms. Our experimental results\nhighlight the efficacy of the proposed method.\n", "date": "2024-08-29", "categories": []}
{"id": "2309.01949", "title": "Variational Bayesian Imaging with an Efficient Surrogate Score-based\n  Prior", "abstract": "  We propose a surrogate function for efficient yet principled use of\nscore-based priors in Bayesian imaging. We consider ill-posed inverse imaging\nproblems in which one aims for a clean image posterior given incomplete or\nnoisy measurements. Since the measurements do not uniquely determine a true\nimage, a prior is needed to constrain the solution space. Recent work turned\nscore-based diffusion models into principled priors for solving ill-posed\nimaging problems by appealing to an ODE-based log-probability function.\nHowever, evaluating the ODE is computationally inefficient and inhibits\nposterior estimation of high-dimensional images. Our proposed surrogate prior\nis based on the evidence lower bound of a score-based diffusion model. We\ndemonstrate the surrogate prior on variational inference for efficient\napproximate posterior sampling of large images. Compared to the exact prior in\nprevious work, our surrogate accelerates optimization of the variational image\ndistribution by at least two orders of magnitude. We also find that our\nprincipled approach gives more accurate posterior estimation than\nnon-variational diffusion-based approaches that involve hyperparameter-tuning\nat inference. Our work establishes a practical path forward for using\nscore-based diffusion models as general-purpose image priors.\n", "date": "2023-09-05", "categories": []}
{"id": "2310.10835", "title": "Provable Probabilistic Imaging using Score-Based Generative Priors", "abstract": "  Estimating high-quality images while also quantifying their uncertainty are\ntwo desired features in an image reconstruction algorithm for solving ill-posed\ninverse problems. In this paper, we propose plug-and-play Monte Carlo (PMC) as\na principled framework for characterizing the space of possible solutions to a\ngeneral inverse problem. PMC is able to incorporate expressive score-based\ngenerative priors for high-quality image reconstruction while also performing\nuncertainty quantification via posterior sampling. In particular, we develop\ntwo PMC algorithms that can be viewed as the sampling analogues of the\ntraditional plug-and-play priors (PnP) and regularization by denoising (RED)\nalgorithms. To improve the sampling efficiency, we introduce weighted annealing\ninto these PMC algorithms, further developing two additional annealed PMC\nalgorithms (APMC). We establish a theoretical analysis for characterizing the\nconvergence behavior of PMC algorithms. Our analysis provides non-asymptotic\nstationarity guarantees in terms of the Fisher information, fully compatible\nwith the joint presence of weighted annealing, potentially non-log-concave\nlikelihoods, and imperfect score networks. We demonstrate the performance of\nthe PMC algorithms on multiple representative inverse problems with both linear\nand nonlinear forward models. Experimental results show that PMC significantly\nimproves reconstruction quality and enables high-fidelity uncertainty\nquantification.\n", "date": "2023-10-16", "categories": []}
{"id": "2311.13134", "title": "Lightweight High-Speed Photography Built on Coded Exposure and Implicit\n  Neural Representation of Videos", "abstract": "  The demand for compact cameras capable of recording high-speed scenes with\nhigh resolution is steadily increasing. However, achieving such capabilities\noften entails high bandwidth requirements, resulting in bulky, heavy systems\nunsuitable for low-capacity platforms. To address this challenge, leveraging a\ncoded exposure setup to encode a frame sequence into a blurry snapshot and\nsubsequently retrieve the latent sharp video presents a lightweight solution.\nNevertheless, restoring motion from blur remains a formidable challenge due to\nthe inherent ill-posedness of motion blur decomposition, the intrinsic\nambiguity in motion direction, and the diverse motions present in natural\nvideos. In this study, we propose a novel approach to address these challenges\nby combining the classical coded exposure imaging technique with the emerging\nimplicit neural representation for videos. We strategically embed motion\ndirection cues into the blurry image during the imaging process. Additionally,\nwe develop a novel implicit neural representation based blur decomposition\nnetwork to sequentially extract the latent video frames from the blurry image,\nleveraging the embedded motion direction cues. To validate the effectiveness\nand efficiency of our proposed framework, we conduct extensive experiments\nusing benchmark datasets and real-captured blurry images. The results\ndemonstrate that our approach significantly outperforms existing methods in\nterms of both quality and flexibility. The code for our work is available at\n.https://github.com/zhihongz/BDINR\n", "date": "2023-11-22", "categories": []}
{"id": "2305.03330", "title": "Solution existence, uniqueness, and stability of discrete basis\n  sinograms in multispectral CT", "abstract": "  This work investigates conditions for quantitative image reconstruction in\nmultispectral computed tomography (MSCT), which remains a topic of active\nresearch. In MSCT, one seeks to obtain from data the spatial distribution of\nlinear attenuation coefficient, referred to as a virtual monochromatic image\n(VMI), at a given X-ray energy, within the subject imaged. As a VMI is\ndecomposed often into a linear combination of basis images with known\ndecomposition coefficients, the reconstruction of a VMI is thus tantamount to\nthat of the basis images. An empirical, but highly effective, two-step\ndata-domain-decomposition (DDD) method has been developed and used widely for\nquantitative image reconstruction in MSCT. In the two-step DDD method, step (1)\nestimates the so-called basis sinogram from data through solving a nonlinear\ntransform, whereas step (2) reconstructs basis images from their basis\nsinograms estimated. Subsequently, a VMI can readily be obtained from the\nlinear combination of basis images reconstructed. As step (2) involves the\ninversion of a straightforward linear system, step (1) is the key component of\nthe DDD method in which a nonlinear system needs to be inverted for estimating\nthe basis sinograms from data. In this work, we consider a {\\it discrete} form\nof the nonlinear system in step (1), and then carry out theoretical and\nnumerical analyses of conditions on the existence, uniqueness, and stability of\na solution to the discrete nonlinear system for accurately estimating the\ndiscrete basis sinograms, leading to quantitative reconstruction of VMIs in\nMSCT.\n", "date": "2023-05-05", "categories": []}
{"id": "2408.15069", "title": "Geometric Artifact Correction for Symmetric Multi-Linear Trajectory CT:\n  Theory, Method, and Generalization", "abstract": "  For extending CT field-of-view to perform non-destructive testing, the\nSymmetric Multi-Linear trajectory Computed Tomography (SMLCT) has been\ndeveloped as a successful example of non-standard CT scanning modes. However,\ninevitable geometric errors can cause severe artifacts in the reconstructed\nimages. The existing calibration method for SMLCT is both crude and\ninefficient. It involves reconstructing hundreds of images by exhaustively\nsubstituting each potential error, and then manually identifying the images\nwith the fewest geometric artifacts to estimate the final geometric errors for\ncalibration. In this paper, we comprehensively and efficiently address the\nchallenging geometric artifacts in SMLCT, , and the corresponding works mainly\ninvolve theory, method, and generalization. In particular, after identifying\nsensitive parameters and conducting some theory analysis of geometric\nartifacts, we summarize several key properties between sensitive geometric\nparameters and artifact characteristics. Then, we further construct\nmathematical relationships that relate sensitive geometric errors to the pixel\noffsets of reconstruction images with artifact characteristics. To accurately\nextract pixel bias, we innovatively adapt the Generalized Cross-Correlation\nwith Phase Transform (GCC-PHAT) algorithm, commonly used in sound processing,\nfor our image registration task for each paired symmetric LCT. This adaptation\nleads to the design of a highly efficient rigid translation registration\nmethod. Simulation and physical experiments have validated the excellent\nperformance of this work. Additionally, our results demonstrate significant\ngeneralization to common rotated CT and a variant of SMLCT.\n", "date": "2024-08-27", "categories": []}
{"id": "2311.14280", "title": "Latent Diffusion Prior Enhanced Deep Unfolding for Snapshot Spectral\n  Compressive Imaging", "abstract": "  Snapshot compressive spectral imaging reconstruction aims to reconstruct\nthree-dimensional spatial-spectral images from a single-shot two-dimensional\ncompressed measurement. Existing state-of-the-art methods are mostly based on\ndeep unfolding structures but have intrinsic performance bottlenecks: $i$) the\nill-posed problem of dealing with heavily degraded measurement, and $ii$) the\nregression loss-based reconstruction models being prone to recover images with\nfew details. In this paper, we introduce a generative model, namely the latent\ndiffusion model (LDM), to generate degradation-free prior to enhance the\nregression-based deep unfolding method. Furthermore, to overcome the large\ncomputational cost challenge in LDM, we propose a lightweight model to generate\nknowledge priors in deep unfolding denoiser, and integrate these priors to\nguide the reconstruction process for compensating high-quality spectral signal\ndetails. Numeric and visual comparisons on synthetic and real-world datasets\nillustrate the superiority of our proposed method in both reconstruction\nquality and computational efficiency. Code will be released.\n", "date": "2023-11-24", "categories": []}
{"id": "2403.14324", "title": "Neural Network-Based Processing and Reconstruction of Compromised\n  Biophotonic Image Data", "abstract": "  The integration of deep learning techniques with biophotonic setups has\nopened new horizons in bioimaging. A compelling trend in this field involves\ndeliberately compromising certain measurement metrics to engineer better\nbioimaging tools in terms of cost, speed, and form-factor, followed by\ncompensating for the resulting defects through the utilization of deep learning\nmodels trained on a large amount of ideal, superior or alternative data. This\nstrategic approach has found increasing popularity due to its potential to\nenhance various aspects of biophotonic imaging. One of the primary motivations\nfor employing this strategy is the pursuit of higher temporal resolution or\nincreased imaging speed, critical for capturing fine dynamic biological\nprocesses. This approach also offers the prospect of simplifying hardware\nrequirements/complexities, thereby making advanced imaging standards more\naccessible in terms of cost and/or size. This article provides an in-depth\nreview of the diverse measurement aspects that researchers intentionally impair\nin their biophotonic setups, including the point spread function,\nsignal-to-noise ratio, sampling density, and pixel resolution. By deliberately\ncompromising these metrics, researchers aim to not only recuperate them through\nthe application of deep learning networks, but also bolster in return other\ncrucial parameters, such as the field-of-view, depth-of-field, and\nspace-bandwidth product. Here, we discuss various biophotonic methods that have\nsuccessfully employed this strategic approach. These techniques span broad\napplications and showcase the versatility and effectiveness of deep learning in\nthe context of compromised biophotonic data. Finally, by offering our\nperspectives on the future possibilities of this rapidly evolving concept, we\nhope to motivate our readers to explore novel ways of balancing hardware\ncompromises with compensation via AI.\n", "date": "2024-03-21", "categories": []}
{"id": "2408.13782", "title": "Batch-FPM: Random batch-update multi-parameter physical Fourier\n  ptychography neural network", "abstract": "  Fourier Ptychographic Microscopy (FPM) is a computational imaging technique\nthat enables high-resolution imaging over a large field of view. However, its\napplication in the biomedical field has been limited due to the long image\nreconstruction time and poor noise robustness. In this paper, we propose a fast\nand robust FPM reconstruction method based on physical neural networks with\nbatch update stochastic gradient descent (SGD) optimization strategy, capable\nof achieving attractive results with low single-to-noise ratio and correcting\nmultiple system parameters simultaneously. Our method leverages a random batch\noptimization approach, breaks away from the fixed sequential iterative order\nand gives greater attention to high-frequency information. The proposed method\nhas better convergence performance even for low signal-to-noise ratio data\nsets, such as low exposure time dark-field images. As a result, it can greatly\nincrease the image recording and result reconstruction speed without any\nadditional hardware modifications. By utilizing advanced deep learning\noptimizers and perform parallel computational scheme, our method enhances GPU\ncomputational efficiency, significantly reducing reconstruction costs.\nExperimental results demonstrate that our method achieves near real-time\ndigital refocusing of a 1024 x 1024 pixels region of interest on consumer-grade\nGPUs. This approach significantly improves temporal resolution (by reducing the\nexposure time of dark-field images), noise resistance, and reconstruction\nspeed, and therefore can efficiently promote the practical application of FPM\nin clinical diagnostics, digital pathology, and biomedical research, etc. In\naddition, we believe our algorithm scheme can help researchers quickly validate\nand implement FPM-related ideas. We invite requests for the full code via\nemail.\n", "date": "2024-08-25", "categories": []}
{"id": "2408.13832", "title": "A Low-dose CT Reconstruction Network Based on TV-regularized OSEM\n  Algorithm", "abstract": "  Low-dose computed tomography (LDCT) offers significant advantages in reducing\nthe potential harm to human bodies. However, reducing the X-ray dose in CT\nscanning often leads to severe noise and artifacts in the reconstructed images,\nwhich might adversely affect diagnosis. By utilizing the expectation\nmaximization (EM) algorithm, statistical priors could be combined with\nartificial priors to improve LDCT reconstruction quality. However, conventional\nEM-based regularization methods adopt an alternating solving strategy, i.e.\nfull reconstruction followed by image-regularization, resulting in\nover-smoothing and slow convergence. In this paper, we propose to integrate TV\nregularization into the ``M''-step of the EM algorithm, thus achieving\neffective and efficient regularization. Besides, by employing the\nChambolle-Pock (CP) algorithm and the ordered subset (OS) strategy, we propose\nthe OSEM-CP algorithm for LDCT reconstruction, in which both reconstruction and\nregularization are conducted view-by-view. Furthermore, by unrolling OSEM-CP,\nwe propose an end-to-end reconstruction neural network (NN), named OSEM-CPNN,\nwith remarkable performance and efficiency that achieves high-quality\nreconstructions in just one full-view iteration. Experiments on different\nmodels and datasets demonstrate our methods' outstanding performance compared\nto traditional and state-of-the-art deep-learning methods.\n", "date": "2024-08-25", "categories": []}
{"id": "2402.17745", "title": "Low-light phase retrieval with implicit generative priors", "abstract": "  Phase retrieval (PR) is fundamentally important in scientific imaging and is\ncrucial for nanoscale techniques like coherent diffractive imaging (CDI). Low\nradiation dose imaging is essential for applications involving\nradiation-sensitive samples. However, most PR methods struggle in low-dose\nscenarios due to high shot noise. Recent advancements in optical data\nacquisition setups, such as in-situ CDI, have shown promise for low-dose\nimaging, but they rely on a time series of measurements, making them unsuitable\nfor single-image applications. Similarly, data-driven phase retrieval\ntechniques are not easily adaptable to data-scarce situations. Zero-shot deep\nlearning methods based on pre-trained and implicit generative priors have been\neffective in various imaging tasks but have shown limited success in PR. In\nthis work, we propose low-dose deep image prior (LoDIP), which combines in-situ\nCDI with the power of implicit generative priors to address single-image\nlow-dose phase retrieval. Quantitative evaluations demonstrate LoDIP's superior\nperformance in this task and its applicability to real experimental scenarios.\n", "date": "2024-02-27", "categories": []}
{"id": "2408.11840", "title": "Joint PET-MRI Reconstruction with Diffusion Stochastic Differential\n  Model", "abstract": "  PET suffers from a low signal-to-noise ratio. Meanwhile, the k-space data\nacquisition process in MRI is time-consuming by PET-MRI systems. We aim to\naccelerate MRI and improve PET image quality. This paper proposed a novel joint\nreconstruction model by diffusion stochastic differential equations based on\nlearning the joint probability distribution of PET and MRI. Compare the results\nunderscore the qualitative and quantitative improvements our model brings to\nPET and MRI reconstruction, surpassing the current state-of-the-art\nmethodologies. Joint PET-MRI reconstruction is a challenge in the PET-MRI\nsystem. This studies focused on the relationship extends beyond edges. In this\nstudy, PET is generated from MRI by learning joint probability distribution as\nthe relationship.\n", "date": "2024-08-07", "categories": []}
{"id": "2408.12100", "title": "A Unified Plug-and-Play Algorithm with Projected Landweber Operator for\n  Split Convex Feasibility Problems", "abstract": "  In recent years Plug-and-Play (PnP) methods have achieved state-of-the-art\nperformance in inverse imaging problems by replacing proximal operators with\ndenoisers. Based on the proximal gradient method, some theoretical results of\nPnP have appeared, where appropriate step size is crucial for convergence\nanalysis. However, in practical applications, applying PnP methods with\ntheoretically guaranteed step sizes is difficult, and these algorithms are\nlimited to Gaussian noise. In this paper,from a perspective of split convex\nfeasibility problems (SCFP), an adaptive PnP algorithm with Projected Landweber\nOperator (PnP-PLO) is proposed to address these issues. Numerical experiments\non image deblurring, super-resolution, and compressed sensing MRI experiments\nillustrate that PnP-PLO with theoretical guarantees outperforms\nstate-of-the-art methods such as RED and RED-PRO.\n", "date": "2024-08-22", "categories": []}
{"id": "2408.12316", "title": "Unrolled Decomposed Unpaired Learning for Controllable Low-Light Video\n  Enhancement", "abstract": "  Obtaining pairs of low/normal-light videos, with motions, is more challenging\nthan still images, which raises technical issues and poses the technical route\nof unpaired learning as a critical role. This paper makes endeavors in the\ndirection of learning for low-light video enhancement without using paired\nground truth. Compared to low-light image enhancement, enhancing low-light\nvideos is more difficult due to the intertwined effects of noise, exposure, and\ncontrast in the spatial domain, jointly with the need for temporal coherence.\nTo address the above challenge, we propose the Unrolled Decomposed Unpaired\nNetwork (UDU-Net) for enhancing low-light videos by unrolling the optimization\nfunctions into a deep network to decompose the signal into spatial and\ntemporal-related factors, which are updated iteratively. Firstly, we formulate\nlow-light video enhancement as a Maximum A Posteriori estimation (MAP) problem\nwith carefully designed spatial and temporal visual regularization. Then, via\nunrolling the problem, the optimization of the spatial and temporal constraints\ncan be decomposed into different steps and updated in a stage-wise manner. From\nthe spatial perspective, the designed Intra subnet leverages unpair prior\ninformation from expert photography retouched skills to adjust the statistical\ndistribution. Additionally, we introduce a novel mechanism that integrates\nhuman perception feedback to guide network optimization, suppressing\nover/under-exposure conditions. Meanwhile, to address the issue from the\ntemporal perspective, the designed Inter subnet fully exploits temporal cues in\nprogressive optimization, which helps achieve improved temporal consistency in\nenhancement results. Consequently, the proposed method achieves superior\nperformance to state-of-the-art methods in video illumination, noise\nsuppression, and temporal consistency across outdoor and indoor scenes.\n", "date": "2024-08-22", "categories": []}
{"id": "2308.03139", "title": "Unfolded proximal neural networks for robust image Gaussian denoising", "abstract": "  A common approach to solve inverse imaging problems relies on finding a\nmaximum a posteriori (MAP) estimate of the original unknown image, by solving a\nminimization problem. In thiscontext, iterative proximal algorithms are widely\nused, enabling to handle non-smooth functions and linear operators. Recently,\nthese algorithms have been paired with deep learning strategies, to further\nimprove the estimate quality. In particular, proximal neural networks (PNNs)\nhave been introduced, obtained by unrolling a proximal algorithm as for finding\na MAP estimate, but over a fixed number of iterations, with learned linear\noperators and parameters. As PNNs are based on optimization theory, they are\nvery flexible, and can be adapted to any image restoration task, as soon as a\nproximal algorithm can solve it. They further have much lighter architectures\nthan traditional networks. In this article we propose a unified framework to\nbuild PNNs for the Gaussian denoising task, based on both the dual-FB and the\nprimal-dual Chambolle-Pock algorithms. We further show that accelerated\ninertial versions of these algorithms enable skip connections in the associated\nNN layers. We propose different learning strategies for our PNN framework, and\ninvestigate their robustness (Lipschitz property) and denoising efficiency.\nFinally, we assess the robustness of our PNNs when plugged in a\nforward-backward algorithm for an image deblurring problem.\n", "date": "2023-08-06", "categories": []}
{"id": "2408.09731", "title": "Reconstruct Spine CT from Biplanar X-Rays via Diffusion Learning", "abstract": "  Intraoperative CT imaging serves as a crucial resource for surgical guidance;\nhowever, it may not always be readily accessible or practical to implement. In\nscenarios where CT imaging is not an option, reconstructing CT scans from\nX-rays can offer a viable alternative. In this paper, we introduce an\ninnovative method for 3D CT reconstruction utilizing biplanar X-rays. Distinct\nfrom previous research that relies on conventional image generation techniques,\nour approach leverages a conditional diffusion process to tackle the task of\nreconstruction. More precisely, we employ a diffusion-based probabilistic model\ntrained to produce 3D CT images based on orthogonal biplanar X-rays. To improve\nthe structural integrity of the reconstructed images, we incorporate a novel\nprojection loss function. Experimental results validate that our proposed\nmethod surpasses existing state-of-the-art benchmarks in both visual image\nquality and multiple evaluative metrics. Specifically, our technique achieves a\nhigher Structural Similarity Index (SSIM) of 0.83, a relative increase of 10\\%,\nand a lower Fr\\'echet Inception Distance (FID) of 83.43, which represents a\nrelative decrease of 25\\%.\n", "date": "2024-08-19", "categories": []}
{"id": "2408.11682", "title": "LiFCal: Online Light Field Camera Calibration via Bundle Adjustment", "abstract": "  We propose LiFCal, a novel geometric online calibration pipeline for\nMLA-based light field cameras. LiFCal accurately determines model parameters\nfrom a moving camera sequence without precise calibration targets, integrating\narbitrary metric scaling constraints. It optimizes intrinsic parameters of the\nlight field camera model, the 3D coordinates of a sparse set of scene points\nand camera poses in a single bundle adjustment defined directly on micro image\npoints.\n  We show that LiFCal can reliably and repeatably calibrate a focused plenoptic\ncamera using different input sequences, providing intrinsic camera parameters\nextremely close to state-of-the-art methods, while offering two main\nadvantages: it can be applied in a target-free scene, and it is implemented\nonline in a complete and continuous pipeline.\n  Furthermore, we demonstrate the quality of the obtained camera parameters in\ndownstream tasks like depth estimation and SLAM.\n  Webpage: https://lifcal.github.io/\n", "date": "2024-08-21", "categories": []}
{"id": "2303.14353", "title": "DiracDiffusion: Denoising and Incremental Reconstruction with Assured\n  Data-Consistency", "abstract": "  Diffusion models have established new state of the art in a multitude of\ncomputer vision tasks, including image restoration. Diffusion-based inverse\nproblem solvers generate reconstructions of exceptional visual quality from\nheavily corrupted measurements. However, in what is widely known as the\nperception-distortion trade-off, the price of perceptually appealing\nreconstructions is often paid in declined distortion metrics, such as PSNR.\nDistortion metrics measure faithfulness to the observation, a crucial\nrequirement in inverse problems. In this work, we propose a novel framework for\ninverse problem solving, namely we assume that the observation comes from a\nstochastic degradation process that gradually degrades and noises the original\nclean image. We learn to reverse the degradation process in order to recover\nthe clean image. Our technique maintains consistency with the original\nmeasurement throughout the reverse process, and allows for great flexibility in\ntrading off perceptual quality for improved distortion metrics and sampling\nspeedup via early-stopping. We demonstrate the efficiency of our method on\ndifferent high-resolution datasets and inverse problems, achieving great\nimprovements over other state-of-the-art diffusion-based methods with respect\nto both perceptual and distortion metrics.\n", "date": "2023-03-25", "categories": []}
{"id": "2408.10653", "title": "UIE-UnFold: Deep Unfolding Network with Color Priors and Vision\n  Transformer for Underwater Image Enhancement", "abstract": "  Underwater image enhancement (UIE) plays a crucial role in various marine\napplications, but it remains challenging due to the complex underwater\nenvironment. Current learning-based approaches frequently lack explicit\nincorporation of prior knowledge about the physical processes involved in\nunderwater image formation, resulting in limited optimization despite their\nimpressive enhancement results. This paper proposes a novel deep unfolding\nnetwork (DUN) for UIE that integrates color priors and inter-stage feature\ntransformation to improve enhancement performance. The proposed DUN model\ncombines the iterative optimization and reliability of model-based methods with\nthe flexibility and representational power of deep learning, offering a more\nexplainable and stable solution compared to existing learning-based UIE\napproaches. The proposed model consists of three key components: a Color Prior\nGuidance Block (CPGB) that establishes a mapping between color channels of\ndegraded and original images, a Nonlinear Activation Gradient Descent Module\n(NAGDM) that simulates the underwater image degradation process, and an Inter\nStage Feature Transformer (ISF-Former) that facilitates feature exchange\nbetween different network stages. By explicitly incorporating color priors and\nmodeling the physical characteristics of underwater image formation, the\nproposed DUN model achieves more accurate and reliable enhancement results.\nExtensive experiments on multiple underwater image datasets demonstrate the\nsuperiority of the proposed model over state-of-the-art methods in both\nquantitative and qualitative evaluations. The proposed DUN-based approach\noffers a promising solution for UIE, enabling more accurate and reliable\nscientific analysis in marine research. The code is available at\nhttps://github.com/CXH-Research/UIE-UnFold.\n", "date": "2024-08-20", "categories": []}
{"id": "2403.12354", "title": "Sim2Real in Reconstructive Spectroscopy: Deep Learning with Augmented\n  Device-Informed Data Simulation", "abstract": "  This work proposes a deep learning (DL)-based framework, namely Sim2Real, for\nspectral signal reconstruction in reconstructive spectroscopy, focusing on\nefficient data sampling and fast inference time. The work focuses on the\nchallenge of reconstructing real-world spectral signals under the extreme\nsetting where only device-informed simulated data are available for training.\nSuch device-informed simulated data are much easier to collect than real-world\ndata but exhibit large distribution shifts from their real-world counterparts.\nTo leverage such simulated data effectively, a hierarchical data augmentation\nstrategy is introduced to mitigate the adverse effects of this domain shift,\nand a corresponding neural network for the spectral signal reconstruction with\nour augmented data is designed. Experiments using a real dataset measured from\nour spectrometer device demonstrate that Sim2Real achieves significant speed-up\nduring the inference while attaining on-par performance with the\nstate-of-the-art optimization-based methods.\n", "date": "2024-03-19", "categories": []}
{"id": "2302.02504", "title": "Motion-compensated MR CINE reconstruction with reconstruction-driven\n  motion estimation", "abstract": "  In cardiac CINE, motion-compensated MR reconstruction (MCMR) is an effective\napproach to address highly undersampled acquisitions by incorporating motion\ninformation between frames. In this work, we propose a novel perspective for\naddressing the MCMR problem and a more integrated and efficient solution to the\nMCMR field. Contrary to state-of-the-art (SOTA) MCMR methods which break the\noriginal problem into two sub-optimization problems, i.e. motion estimation and\nreconstruction, we formulate this problem as a single entity with one single\noptimization. Our approach is unique in that the motion estimation is directly\ndriven by the ultimate goal, reconstruction, but not by the canonical\nmotion-warping loss (similarity measurement between motion-warped images and\ntarget images). We align the objectives of motion estimation and\nreconstruction, eliminating the drawbacks of artifacts-affected motion\nestimation and therefore error-propagated reconstruction. Further, we can\ndeliver high-quality reconstruction and realistic motion without applying any\nregularization/smoothness loss terms, circumventing the non-trivial weighting\nfactor tuning. We evaluate our method on two datasets: 1) an in-house acquired\n2D CINE dataset for the retrospective study and 2) the public OCMR cardiac\ndataset for the prospective study. The conducted experiments indicate that the\nproposed MCMR framework can deliver artifact-free motion estimation and\nhigh-quality MR images even for imaging accelerations up to 20x, outperforming\nSOTA non-MCMR and MCMR methods in both qualitative and quantitative evaluation\nacross all experiments. The code is available at\nhttps://github.com/JZPeterPan/MCMR-Recon-Driven-Motion.\n", "date": "2023-02-05", "categories": []}
{"id": "2312.01677", "title": "Multi-task Image Restoration Guided By Robust DINO Features", "abstract": "  Multi-task image restoration has gained significant interest due to its\ninherent versatility and efficiency compared to its single-task counterpart.\nHowever, performance decline is observed with an increase in the number of\ntasks, primarily attributed to the restoration model's challenge in handling\ndifferent tasks with distinct natures at the same time. Thus, a perspective\nemerged aiming to explore the degradation-insensitive semantic commonalities\namong different degradation tasks. In this paper, we observe that the features\nof DINOv2 can effectively model semantic information and are independent of\ndegradation factors. Motivated by this observation, we propose\n\\mbox{\\textbf{DINO-IR}}, a multi-task image restoration approach leveraging\nrobust features extracted from DINOv2 to solve multi-task image restoration\nsimultaneously. We first propose a pixel-semantic fusion (PSF) module to\ndynamically fuse DINOV2's shallow features containing pixel-level information\nand deep features containing degradation-independent semantic information. To\nguide the restoration model with the features of DINOv2, we develop a\nDINO-Restore adaption and fusion module to adjust the channel of fused features\nfrom PSF and then integrate them with the features from the restoration model.\nBy formulating these modules into a unified deep model, we propose a DINO\nperception contrastive loss to constrain the model training. Extensive\nexperimental results demonstrate that our DINO-IR performs favorably against\nexisting multi-task image restoration approaches in various tasks by a large\nmargin. The source codes and trained models will be made available.\n", "date": "2023-12-04", "categories": []}
{"id": "2408.08087", "title": "ColorMamba: Towards High-quality NIR-to-RGB Spectral Translation with\n  Mamba", "abstract": "  Translating NIR to the visible spectrum is challenging due to cross-domain\ncomplexities. Current models struggle to balance a broad receptive field with\ncomputational efficiency, limiting practical use. Although the Selective\nStructured State Space Model, especially the improved version, Mamba, excels in\ngenerative tasks by capturing long-range dependencies with linear complexity,\nits default approach of converting 2D images into 1D sequences neglects local\ncontext. In this work, we propose a simple but effective backbone, dubbed\nColorMamba, which first introduces Mamba into spectral translation tasks. To\nexplore global long-range dependencies and local context for efficient spectral\ntranslation, we introduce learnable padding tokens to enhance the distinction\nof image boundaries and prevent potential confusion within the sequence model.\nFurthermore, local convolutional enhancement and agent attention are designed\nto improve the vanilla Mamba. Moreover, we exploit the HSV color to provide\nmulti-scale guidance in the reconstruction process for more accurate spectral\ntranslation. Extensive experiments show that our ColorMamba achieves a 1.02\nimprovement in terms of PSNR compared with the state-of-the-art method. Our\ncode is available at https://github.com/AlexYangxx/ColorMamba.\n", "date": "2024-08-15", "categories": []}
{"id": "2212.00014", "title": "Attentional Ptycho-Tomography (APT) for three-dimensional nanoscale\n  X-ray imaging with minimal data acquisition and computation time", "abstract": "  Noninvasive X-ray imaging of nanoscale three-dimensional objects, e.g.\nintegrated circuits (ICs), generally requires two types of scanning:\nptychographic, which is translational and returns estimates of complex\nelectromagnetic field through ICs; and tomographic scanning, which collects\ncomplex field projections from multiple angles. Here, we present Attentional\nPtycho-Tomography (APT), an approach trained to provide accurate\nreconstructions of ICs despite incomplete measurements, using a dramatically\nreduced amount of angular scanning. Training process includes regularizing\npriors based on typical IC patterns and the physics of X-ray propagation. We\ndemonstrate that APT with 12-time reduced angles achieves fidelity comparable\nto the gold standard with the original set of angles. With the same set of\nreduced angles, APT also outperforms baseline reconstruction methods. In our\nexperiments, APT achieves 108-time aggregate reduction in data acquisition and\ncomputation without compromising quality. We expect our physics-assisted\nmachine learning framework could also be applied to other branches of nanoscale\nimaging.\n", "date": "2022-11-29", "categories": []}
{"id": "2304.04597", "title": "Accelerated deep self-supervised ptycho-laminography for\n  three-dimensional nanoscale imaging of integrated circuits", "abstract": "  Three-dimensional inspection of nanostructures such as integrated circuits is\nimportant for security and reliability assurance. Two scanning operations are\nrequired: ptychographic to recover the complex transmissivity of the specimen;\nand rotation of the specimen to acquire multiple projections covering the 3D\nspatial frequency domain. Two types of rotational scanning are possible:\ntomographic and laminographic. For flat, extended samples, for which the full\n180 degree coverage is not possible, the latter is preferable because it\nprovides better coverage of the 3D spatial frequency domain compared to\nlimited-angle tomography. It is also because the amount of attenuation through\nthe sample is approximately the same for all projections. However, both\ntechniques are time consuming because of extensive acquisition and computation\ntime. Here, we demonstrate the acceleration of ptycho-laminographic\nreconstruction of integrated circuits with 16-times fewer angular samples and\n4.67-times faster computation by using a physics-regularized deep\nself-supervised learning architecture. We check the fidelity of our\nreconstruction against a densely sampled reconstruction that uses full scanning\nand no learning. As already reported elsewhere [Zhou and Horstmeyer, Opt.\nExpress, 28(9), pp. 12872-12896], we observe improvement of reconstruction\nquality even over the densely sampled reconstruction, due to the ability of the\nself-supervised learning kernel to fill the missing cone.\n", "date": "2023-04-10", "categories": []}
{"id": "2408.07541", "title": "DifuzCam: Replacing Camera Lens with a Mask and a Diffusion Model", "abstract": "  The flat lensless camera design reduces the camera size and weight\nsignificantly. In this design, the camera lens is replaced by another optical\nelement that interferes with the incoming light. The image is recovered from\nthe raw sensor measurements using a reconstruction algorithm. Yet, the quality\nof the reconstructed images is not satisfactory. To mitigate this, we propose\nutilizing a pre-trained diffusion model with a control network and a learned\nseparable transformation for reconstruction. This allows us to build a\nprototype flat camera with high-quality imaging, presenting state-of-the-art\nresults in both terms of quality and perceptuality. We demonstrate its ability\nto leverage also textual descriptions of the captured scene to further enhance\nreconstruction. Our reconstruction method which leverages the strong\ncapabilities of a pre-trained diffusion model can be used in other imaging\nsystems for improved reconstruction results.\n", "date": "2024-08-14", "categories": []}
{"id": "2408.07580", "title": "Theoretical and Practical Progress in Hyperspectral Pixel Unmixing with\n  Large Spectral Libraries from a Sparse Perspective", "abstract": "  Hyperspectral unmixing is the process of determining the presence of\nindividual materials and their respective abundances from an observed pixel\nspectrum. Unmixing is a fundamental process in hyperspectral image analysis,\nand is growing in importance as increasingly large spectral libraries are\ncreated and used. Unmixing is typically done with ordinary least squares (OLS)\nregression. However, unmixing with large spectral libraries where the materials\npresent in a pixel are not a priori known, solving for the coefficients in OLS\nrequires inverting a non-invertible matrix from a large spectral library. A\nnumber of regression methods are available that can produce a numerical\nsolution using regularization, but with considerably varied effectiveness.\nAlso, simple methods that are unpopular in the statistics literature (i.e.\nstep-wise regression) are used with some level of effectiveness in\nhyperspectral analysis. In this paper, we provide a thorough performance\nevaluation of the methods considered, evaluating methods based on how often\nthey select the correct materials in the models. Investigated methods include\nordinary least squares regression, non-negative least squares regression, ridge\nregression, lasso regression, step-wise regression and Bayesian model\naveraging. We evaluated these unmixing approaches using multiple criteria:\nincorporation of non-negative abundances, model size, accurate mineral\ndetection and root mean squared error (RMSE). We provide a taxonomy of the\nregression methods, showing that most methods can be understood as Bayesian\nmethods with specific priors. We conclude that methods that can be derived with\npriors that correspond to the phenomenology of hyperspectral imagery outperform\nthose with priors that are optimal for prediction performance under the\nassumptions of ordinary least squares linear regression.\n", "date": "2024-08-14", "categories": []}
{"id": "2408.05683", "title": "Single Image Dehazing Using Scene Depth Ordering", "abstract": "  Images captured in hazy weather generally suffer from quality degradation,\nand many dehazing methods have been developed to solve this problem. However,\nsingle image dehazing problem is still challenging due to its ill-posed nature.\nIn this paper, we propose a depth order guided single image dehazing method,\nwhich utilizes depth order in hazy images to guide the dehazing process to\nachieve a similar depth perception in corresponding dehazing results. The\nconsistency of depth perception ensures that the regions that look farther or\ncloser in hazy images also appear farther or closer in the corresponding\ndehazing results, and thus effectively avoid the undesired visual effects. To\nachieve this goal, a simple yet effective strategy is proposed to extract the\ndepth order in hazy images, which offers a reference for depth perception in\nhazy weather. Additionally, a depth order embedded transformation model is\ndevised, which performs transmission estimation under the guidance of depth\norder to realize an unchanged depth order in the dehazing results. The\nextracted depth order provides a powerful global constraint for the dehazing\nprocess, which contributes to the efficient utilization of global information,\nthereby bringing an overall improvement in restoration quality. Extensive\nexperiments demonstrate that the proposed method can better recover potential\nstructure and vivid color with higher computational efficiency than the\nstate-of-the-art dehazing methods.\n", "date": "2024-08-11", "categories": []}
{"id": "2408.05923", "title": "Image Denoising Using Green Channel Prior", "abstract": "  Image denoising is an appealing and challenging task, in that noise\nstatistics of real-world observations may vary with local image contents and\ndifferent image channels. Specifically, the green channel usually has twice the\nsampling rate in raw data. To handle noise variances and leverage such\nchannel-wise prior information, we propose a simple and effective green channel\nprior-based image denoising (GCP-ID) method, which integrates GCP into the\nclassic patch-based denoising framework. Briefly, we exploit the green channel\nto guide the search for similar patches, which aims to improve the patch\ngrouping quality and encourage sparsity in the transform domain. The grouped\nimage patches are then reformulated into RGGB arrays to explicitly characterize\nthe density of green samples. Furthermore, to enhance the adaptivity of GCP-ID\nto various image contents, we cast the noise estimation problem into a\nclassification task and train an effective estimator based on convolutional\nneural networks (CNNs). Experiments on real-world datasets demonstrate the\ncompetitive performance of the proposed GCP-ID method for image and video\ndenoising applications in both raw and sRGB spaces. Our code is available at\nhttps://github.com/ZhaomingKong/GCP-ID.\n", "date": "2024-08-12", "categories": []}
{"id": "2408.06014", "title": "A Sharpness Based Loss Function for Removing Out-of-Focus Blur", "abstract": "  The success of modern Deep Neural Network (DNN) approaches can be attributed\nto the use of complex optimization criteria beyond standard losses such as mean\nabsolute error (MAE) or mean squared error (MSE). In this work, we propose a\nnovel method of utilising a no-reference sharpness metric Q introduced by Zhu\nand Milanfar for removing out-of-focus blur from images. We also introduce a\nnovel dataset of real-world out-of-focus images for assessing restoration\nmodels. Our fine-tuned method produces images with a 7.5 % increase in\nperceptual quality (LPIPS) as compared to a standard model trained only on MAE.\nFurthermore, we observe a 6.7 % increase in Q (reflecting sharper restorations)\nand 7.25 % increase in PSNR over most state-of-the-art (SOTA) methods.\n", "date": "2024-08-12", "categories": []}
{"id": "2408.04171", "title": "Rotation center identification based on geometric relationships for\n  rotary motion deblurring", "abstract": "  Non-blind rotary motion deblurring (RMD) aims to recover the latent clear\nimage from a rotary motion blurred (RMB) image. The rotation center is a\ncrucial input parameter in non-blind RMD methods. Existing methods directly\nestimate the rotation center from the RMB image. However they always suffer\nsignificant errors, and the performance of RMD is limited. For the assembled\nimaging systems, the position of the rotation center remains fixed. Leveraging\nthis prior knowledge, we propose a geometric-based method for rotation center\nidentification and analyze its error range. Furthermore, we construct a RMB\nimaging system. The experiment demonstrates that our method achieves less than\n1-pixel error along a single axis (x-axis or y-axis). We utilize the\nconstructed imaging system to capture real RMB images, and experimental results\nshow that our method can help existing RMD approaches yield better RMD images.\n", "date": "2024-08-08", "categories": []}
{"id": "2408.04227", "title": "Physical prior guided cooperative learning framework for joint\n  turbulence degradation estimation and infrared video restoration", "abstract": "  Infrared imaging and turbulence strength measurements are in widespread\ndemand in many fields. This paper introduces a Physical Prior Guided\nCooperative Learning (P2GCL) framework to jointly enhance atmospheric\nturbulence strength estimation and infrared image restoration. P2GCL involves a\ncyclic collaboration between two models, i.e., a TMNet measures turbulence\nstrength and outputs the refractive index structure constant (Cn2) as a\nphysical prior, a TRNet conducts infrared image sequence restoration based on\nCn2 and feeds the restored images back to the TMNet to boost the measurement\naccuracy. A novel Cn2-guided frequency loss function and a physical constraint\nloss are introduced to align the training process with physical theories.\nExperiments demonstrate P2GCL achieves the best performance for both turbulence\nstrength estimation (improving Cn2 MAE by 0.0156, enhancing R2 by 0.1065) and\nimage restoration (enhancing PSNR by 0.2775 dB), validating the significant\nimpact of physical prior guided cooperative learning.\n", "date": "2024-08-08", "categories": []}
{"id": "2302.01089", "title": "Curriculum Learning for ab initio Deep Learned Refractive Optics", "abstract": "  Deep optical optimization has recently emerged as a new paradigm for\ndesigning computational imaging systems using only the output image as the\nobjective. However, it has been limited to either simple optical systems\nconsisting of a single element such as a diffractive optical element (DOE) or\nmetalens, or the fine-tuning of compound lenses from good initial designs. Here\nwe present a DeepLens design method based on curriculum learning, which is able\nto learn optical designs of compound lenses ab initio from randomly initialized\nsurfaces without human intervention, therefore overcoming the need for a good\ninitial design. We demonstrate the effectiveness of our approach by fully\nautomatically designing both classical imaging lenses and a large field-of-view\nextended depth-of-field computational lens in a cellphone-style form factor,\nwith highly aspheric surfaces and a short back focal length.\n", "date": "2023-02-02", "categories": []}
{"id": "2408.03327", "title": "Reconstruction of the shape of irregular rough particles from their\n  interferometric images using a convolutional neural network", "abstract": "  We have developed a convolutional neural network (CNN) to reconstruct the\nshape of irregular rough particles from their interferometric images. The CNN\nis based on a UNET architecture with residual block modules. The database has\nbeen constructed using the experimental patterns generated by perfectly known\npseudo-particles programmed on a Digital Micromirror Device (DMD) and under\nlaser illumination. The CNN has been trained on a basis of 18000 experimental\ninterferometric images using the AUSTRAL super computer (at CRIANN in\nNormandy). The CNN is tested in the case of centrosymmetric (stick, cross,\ndendrite) and non-centrosymmetric (like T, Y or L) particles. The size and the\n3D orientation of the programmed particles are random. The different shapes are\nreconstructed by the CNN with good accuracy. Using three angles of view, the 3D\nreconstruction of particles from three reconstructed faces can be further done.\n", "date": "2024-07-19", "categories": []}
{"id": "2308.10269", "title": "Domain Reduction Strategy for Non Line of Sight Imaging", "abstract": "  This paper presents a novel optimization-based method for non-line-of-sight\n(NLOS) imaging that aims to reconstruct hidden scenes under general setups with\nsignificantly reduced reconstruction time. In NLOS imaging, the visible\nsurfaces of the target objects are notably sparse. To mitigate unnecessary\ncomputations arising from empty regions, we design our method to render the\ntransients through partial propagations from a continuously sampled set of\npoints from the hidden space. Our method is capable of accurately and\nefficiently modeling the view-dependent reflectance using surface normals,\nwhich enables us to obtain surface geometry as well as albedo. In this\npipeline, we propose a novel domain reduction strategy to eliminate superfluous\ncomputations in empty regions. During the optimization process, our domain\nreduction procedure periodically prunes the empty regions from our sampling\ndomain in a coarse-to-fine manner, leading to substantial improvement in\nefficiency. We demonstrate the effectiveness of our method in various NLOS\nscenarios with sparse scanning patterns. Experiments conducted on both\nsynthetic and real-world data support the efficacy in general NLOS scenarios,\nand the improved efficiency of our method compared to the previous\noptimization-based solutions. Our code is available at\nhttps://github.com/hyunbo9/domain-reduction-strategy.\n", "date": "2023-08-20", "categories": []}
{"id": "2405.01726", "title": "SSUMamba: Spatial-Spectral Selective State Space Model for Hyperspectral\n  Image Denoising", "abstract": "  Denoising is a crucial preprocessing step for hyperspectral images (HSIs) due\nto noise arising from intra-imaging mechanisms and environmental factors.\nLong-range spatial-spectral correlation modeling is beneficial for HSI\ndenoising but often comes with high computational complexity. Based on the\nstate space model (SSM), Mamba is known for its remarkable long-range\ndependency modeling capabilities and computational efficiency. Building on\nthis, we introduce a memory-efficient spatial-spectral UMamba (SSUMamba) for\nHSI denoising, with the spatial-spectral continuous scan (SSCS) Mamba being the\ncore component. SSCS Mamba alternates the row, column, and band in six\ndifferent orders to generate the sequence and uses the bidirectional SSM to\nexploit long-range spatial-spectral dependencies. In each order, the images are\nrearranged between adjacent scans to ensure spatial-spectral continuity.\nAdditionally, 3D convolutions are embedded into the SSCS Mamba to enhance local\nspatial-spectral modeling. Experiments demonstrate that SSUMamba achieves\nsuperior denoising results with lower memory consumption per batch compared to\ntransformer-based methods. The source code is available at\nhttps://github.com/lronkitty/SSUMamba.\n", "date": "2024-05-02", "categories": []}
{"id": "2403.14837", "title": "Osmosis: RGBD Diffusion Prior for Underwater Image Restoration", "abstract": "  Underwater image restoration is a challenging task because of water effects\nthat increase dramatically with distance. This is worsened by lack of ground\ntruth data of clean scenes without water. Diffusion priors have emerged as\nstrong image restoration priors. However, they are often trained with a dataset\nof the desired restored output, which is not available in our case. We also\nobserve that using only color data is insufficient, and therefore augment the\nprior with a depth channel. We train an unconditional diffusion model prior on\nthe joint space of color and depth, using standard RGBD datasets of natural\noutdoor scenes in air. Using this prior together with a novel guidance method\nbased on the underwater image formation model, we generate posterior samples of\nclean images, removing the water effects. Even though our prior did not see any\nunderwater images during training, our method outperforms state-of-the-art\nbaselines for image restoration on very challenging scenes. Our code, models\nand data are available on the project website.\n", "date": "2024-03-21", "categories": []}
{"id": "2407.13211", "title": "Research on Image Super-Resolution Reconstruction Mechanism based on\n  Convolutional Neural Network", "abstract": "  Super-resolution reconstruction techniques entail the utilization of software\nalgorithms to transform one or more sets of low-resolution images captured from\nthe same scene into high-resolution images. In recent years, considerable\nadvancement has been observed in the domain of single-image super-resolution\nalgorithms, particularly those based on deep learning techniques. Nevertheless,\nthe extraction of image features and nonlinear mapping methods in the\nreconstruction process remain challenging for existing algorithms. These issues\nresult in the network architecture being unable to effectively utilize the\ndiverse range of information at different levels. The loss of high-frequency\ndetails is significant, and the final reconstructed image features are overly\nsmooth, with a lack of fine texture details. This negatively impacts the\nsubjective visual quality of the image. The objective is to recover\nhigh-quality, high-resolution images from low-resolution images. In this work,\nan enhanced deep convolutional neural network model is employed, comprising\nmultiple convolutional layers, each of which is configured with specific\nfilters and activation functions to effectively capture the diverse features of\nthe image. Furthermore, a residual learning strategy is employed to accelerate\ntraining and enhance the convergence of the network, while sub-pixel\nconvolutional layers are utilized to refine the high-frequency details and\ntextures of the image. The experimental analysis demonstrates the superior\nperformance of the proposed model on multiple public datasets when compared\nwith the traditional bicubic interpolation method and several other\nlearning-based super-resolution methods. Furthermore, it proves the model's\nefficacy in maintaining image edges and textures.\n", "date": "2024-07-18", "categories": []}
{"id": "2408.00470", "title": "Image Super-Resolution with Taylor Expansion Approximation and Large\n  Field Reception", "abstract": "  Self-similarity techniques are booming in blind super-resolution (SR) due to\naccurate estimation of the degradation types involved in low-resolution images.\nHowever, high-dimensional matrix multiplication within self-similarity\ncomputation prohibitively consumes massive computational costs. We find that\nthe high-dimensional attention map is derived from the matrix multiplication\nbetween Query and Key, followed by a softmax function. This softmax makes the\nmatrix multiplication between Query and Key inseparable, posing a great\nchallenge in simplifying computational complexity. To address this issue, we\nfirst propose a second-order Taylor expansion approximation (STEA) to separate\nthe matrix multiplication of Query and Key, resulting in the complexity\nreduction from $\\mathcal{O}(N^2)$ to $\\mathcal{O}(N)$. Then, we design a\nmulti-scale large field reception (MLFR) to compensate for the performance\ndegradation caused by STEA. Finally, we apply these two core designs to\nlaboratory and real-world scenarios by constructing LabNet and RealNet,\nrespectively. Extensive experimental results tested on five synthetic datasets\ndemonstrate that our LabNet sets a new benchmark in qualitative and\nquantitative evaluations. Tested on the RealWorld38 dataset, our RealNet\nachieves superior visual quality over existing methods. Ablation studies\nfurther verify the contributions of STEA and MLFR towards both LabNet and\nRealNet frameworks.\n", "date": "2024-08-01", "categories": []}
{"id": "2408.00695", "title": "Accelerating Full Waveform Inversion By Transfer Learning", "abstract": "  Full waveform inversion (FWI) is a powerful tool for reconstructing material\nfields based on sparsely measured data obtained by wave propagation. For\nspecific problems, discretizing the material field with a neural network (NN)\nimproves the robustness and reconstruction quality of the corresponding\noptimization problem. We call this method NN-based FWI. Starting from an\ninitial guess, the weights of the NN are iteratively updated to fit the\nsimulated wave signals to the sparsely measured data set. For gradient-based\noptimization, a suitable choice of the initial guess, i.e., a suitable NN\nweight initialization, is crucial for fast and robust convergence.\n  In this paper, we introduce a novel transfer learning approach to further\nimprove NN-based FWI. This approach leverages supervised pretraining to provide\na better NN weight initialization, leading to faster convergence of the\nsubsequent optimization problem. Moreover, the inversions yield physically more\nmeaningful local minima. The network is pretrained to predict the unknown\nmaterial field using the gradient information from the first iteration of\nconventional FWI. In our computational experiments on two-dimensional domains,\nthe training data set consists of reference simulations with arbitrarily\npositioned elliptical voids of different shapes and orientations. We compare\nthe performance of the proposed transfer learning NN-based FWI with three other\nmethods: conventional FWI, NN-based FWI without pretraining and conventional\nFWI with an initial guess predicted from the pretrained NN. Our results show\nthat transfer learning NN-based FWI outperforms the other methods in terms of\nconvergence speed and reconstruction quality.\n", "date": "2024-08-01", "categories": []}
{"id": "2312.00125", "title": "Scalable Bayesian uncertainty quantification with data-driven priors for\n  radio interferometric imaging", "abstract": "  Next-generation radio interferometers like the Square Kilometer Array have\nthe potential to unlock scientific discoveries thanks to their unprecedented\nangular resolution and sensitivity. One key to unlocking their potential\nresides in handling the deluge and complexity of incoming data. This challenge\nrequires building radio interferometric imaging methods that can cope with the\nmassive data sizes and provide high-quality image reconstructions with\nuncertainty quantification (UQ). This work proposes a method coined QuantifAI\nto address UQ in radio-interferometric imaging with data-driven (learned)\npriors for high-dimensional settings. Our model, rooted in the Bayesian\nframework, uses a physically motivated model for the likelihood. The model\nexploits a data-driven convex prior, which can encode complex information\nlearned implicitly from simulations and guarantee the log-concavity of the\nposterior. We leverage probability concentration phenomena of high-dimensional\nlog-concave posteriors that let us obtain information about the posterior,\navoiding MCMC sampling techniques. We rely on convex optimisation methods to\ncompute the MAP estimation, which is known to be faster and better scale with\ndimension than MCMC sampling strategies. Our method allows us to compute local\ncredible intervals, i.e., Bayesian error bars, and perform hypothesis testing\nof structure on the reconstructed image. In addition, we propose a novel\nblazing-fast method to compute pixel-wise uncertainties at different scales. We\ndemonstrate our method by reconstructing radio-interferometric images in a\nsimulated setting and carrying out fast and scalable UQ, which we validate with\nMCMC sampling. Our method shows an improved image quality and more meaningful\nuncertainties than the benchmark method based on a sparsity-promoting prior.\nQuantifAI's source code: https://github.com/astro-informatics/QuantifAI.\n", "date": "2023-11-30", "categories": []}
{"id": "2407.19248", "title": "Mamba-UIE: Enhancing Underwater Images with Physical Model Constraint", "abstract": "  In underwater image enhancement (UIE), convolutional neural networks (CNN)\nhave inherent limitations in modeling long-range dependencies and are less\neffective in recovering global features. While Transformers excel at modeling\nlong-range dependencies, their quadratic computational complexity with\nincreasing image resolution presents significant efficiency challenges.\nAdditionally, most supervised learning methods lack effective physical model\nconstraint, which can lead to insufficient realism and overfitting in generated\nimages. To address these issues, we propose a physical model constraint-based\nunderwater image enhancement framework, Mamba-UIE. Specifically, we decompose\nthe input image into four components: underwater scene radiance, direct\ntransmission map, backscatter transmission map, and global background light.\nThese components are reassembled according to the revised underwater image\nformation model, and the reconstruction consistency constraint is applied\nbetween the reconstructed image and the original image, thereby achieving\neffective physical constraint on the underwater image enhancement process. To\ntackle the quadratic computational complexity of Transformers when handling\nlong sequences, we introduce the Mamba-UIE network based on linear complexity\nstate space models. By incorporating the Mamba in Convolution block, long-range\ndependencies are modeled at both the channel and spatial levels, while the CNN\nbackbone is retained to recover local features and details. Extensive\nexperiments on three public datasets demonstrate that our proposed Mamba-UIE\noutperforms existing state-of-the-art methods, achieving a PSNR of 27.13 and an\nSSIM of 0.93 on the UIEB dataset. Our method is available at\nhttps://github.com/zhangsong1213/Mamba-UIE.\n", "date": "2024-07-27", "categories": []}
{"id": "2405.07842", "title": "Ground-based image deconvolution with Swin Transformer UNet", "abstract": "  As ground-based all-sky astronomical surveys will gather millions of images\nin the coming years, a critical requirement emerges for the development of fast\ndeconvolution algorithms capable of efficiently improving the spatial\nresolution of these images. By successfully recovering clean and\nhigh-resolution images from these surveys, the objective is to deepen the\nunderstanding of galaxy formation and evolution through accurate photometric\nmeasurements. We introduce a two-step deconvolution framework using a Swin\nTransformer architecture. Our study reveals that the deep learning-based\nsolution introduces a bias, constraining the scope of scientific analysis. To\naddress this limitation, we propose a novel third step relying on the active\ncoefficients in the sparsity wavelet framework. We conducted a performance\ncomparison between our deep learning-based method and Firedec, a classical\ndeconvolution algorithm, based on an analysis of a subset of the EDisCS cluster\nsamples. We demonstrate the advantage of our method in terms of resolution\nrecovery, generalisation to different noise properties, and computational\nefficiency. The analysis of this cluster sample not only allowed us to assess\nthe efficiency of our method, but it also enabled us to quantify the number of\nclumps within these galaxies in relation to their disc colour. This robust\ntechnique that we propose holds promise for identifying structures in the\ndistant universe through ground-based images.\n", "date": "2024-05-13", "categories": []}
{"id": "2407.20399", "title": "Analysis and Improvement of Rank-Ordered Mean Algorithm in Single-Photon\n  LiDAR", "abstract": "  Depth estimation using a single-photon LiDAR is often solved by a matched\nfilter. It is, however, error-prone in the presence of background noise. A\ncommonly used technique to reject background noise is the rank-ordered mean\n(ROM) filter previously reported by Shin \\textit{et al.} (2015). ROM rejects\nnoisy photon arrival timestamps by selecting only a small range of them around\nthe median statistics within its local neighborhood. Despite the promising\nperformance of ROM, its theoretical performance limit is unknown. In this\npaper, we theoretically characterize the ROM performance by showing that ROM\nfails when the reflectivity drops below a threshold predetermined by the depth\nand signal-to-background ratio, and its accuracy undergoes a phase transition\nat the cutoff. Based on our theory, we propose an improved signal extraction\ntechnique by selecting tight timestamp clusters. Experimental results show that\nthe proposed algorithm improves depth estimation performance over ROM by 3\norders of magnitude at the same signal intensities, and achieves high image\nfidelity at noise levels as high as 17 times that of signal.\n", "date": "2024-07-29", "categories": []}
{"id": "2407.20502", "title": "Restoring Real-World Degraded Events Improves Deblurring Quality", "abstract": "  Due to its high speed and low latency, DVS is frequently employed in motion\ndeblurring. Ideally, high-quality events would adeptly capture intricate motion\ninformation. However, real-world events are generally degraded, thereby\nintroducing significant artifacts into the deblurred results. In response to\nthis challenge, we model the degradation of events and propose RDNet to improve\nthe quality of image deblurring. Specifically, we first analyze the mechanisms\nunderlying degradation and simulate paired events based on that. These paired\nevents are then fed into the first stage of the RDNet for training the\nrestoration model. The events restored in this stage serve as a guide for the\nsecond-stage deblurring process. To better assess the deblurring performance of\ndifferent methods on real-world degraded events, we present a new real-world\ndataset named DavisMCR. This dataset incorporates events with diverse\ndegradation levels, collected by manipulating environmental brightness and\ntarget object contrast. Our experiments are conducted on synthetic datasets\n(GOPRO), real-world datasets (REBlur), and the proposed dataset (DavisMCR). The\nresults demonstrate that RDNet outperforms classical event denoising methods in\nevent restoration. Furthermore, RDNet exhibits better performance in deblurring\ntasks compared to state-of-the-art methods. DavisMCR are available at\nhttps://github.com/Yeeesir/DVS_RDNet.\n", "date": "2024-07-30", "categories": []}
{"id": "2407.20878", "title": "S3PET: Semi-supervised Standard-dose PET Image Reconstruction via\n  Dose-aware Token Swap", "abstract": "  To acquire high-quality positron emission tomography (PET) images while\nreducing the radiation tracer dose, numerous efforts have been devoted to\nreconstructing standard-dose PET (SPET) images from low-dose PET (LPET).\nHowever, the success of current fully-supervised approaches relies on abundant\npaired LPET and SPET images, which are often unavailable in clinic. Moreover,\nthese methods often mix the dose-invariant content with dose level-related\ndose-specific details during reconstruction, resulting in distorted images. To\nalleviate these problems, in this paper, we propose a two-stage Semi-Supervised\nSPET reconstruction framework, namely S3PET, to accommodate the training of\nabundant unpaired and limited paired SPET and LPET images. Our S3PET involves\nan un-supervised pre-training stage (Stage I) to extract representations from\nunpaired images, and a supervised dose-aware reconstruction stage (Stage II) to\nachieve LPET-to-SPET reconstruction by transferring the dose-specific knowledge\nbetween paired images. Specifically, in stage I, two independent dose-specific\nmasked autoencoders (DsMAEs) are adopted to comprehensively understand the\nunpaired SPET and LPET images. Then, in Stage II, the pre-trained DsMAEs are\nfurther finetuned using paired images. To prevent distortions in both content\nand details, we introduce two elaborate modules, i.e., a dose knowledge\ndecouple module to disentangle the respective dose-specific and dose-invariant\nknowledge of LPET and SPET, and a dose-specific knowledge learning module to\ntransfer the dose-specific information from SPET to LPET, thereby achieving\nhigh-quality SPET reconstruction from LPET images. Experiments on two datasets\ndemonstrate that our S3PET achieves state-of-the-art performance quantitatively\nand qualitatively.\n", "date": "2024-07-30", "categories": []}
{"id": "2407.18574", "title": "Learning to Enhance Aperture Phasor Field for Non-Line-of-Sight Imaging", "abstract": "  This paper aims to facilitate more practical NLOS imaging by reducing the\nnumber of samplings and scan areas. To this end, we introduce a phasor-based\nenhancement network that is capable of predicting clean and full measurements\nfrom noisy partial observations. We leverage a denoising autoencoder scheme to\nacquire rich and noise-robust representations in the measurement space. Through\nthis pipeline, our enhancement network is trained to accurately reconstruct\ncomplete measurements from their corrupted and partial counterparts. However,\nwe observe that the \\naive application of denoising often yields degraded and\nover-smoothed results, caused by unnecessary and spurious frequency signals\npresent in measurements. To address this issue, we introduce a phasor-based\npipeline designed to limit the spectrum of our network to the frequency range\nof interests, where the majority of informative signals are detected. The\nphasor wavefronts at the aperture, which are band-limited signals, are employed\nas inputs and outputs of the network, guiding our network to learn from the\nfrequency range of interests and discard unnecessary information. The\nexperimental results in more practical acquisition scenarios demonstrate that\nwe can look around the corners with $16\\times$ or $64\\times$ fewer samplings\nand $4\\times$ smaller apertures. Our code is available at\nhttps://github.com/join16/LEAP.\n", "date": "2024-07-26", "categories": []}
{"id": "2407.19407", "title": "Near-Isotropic Sub-{\\AA}ngstrom 3D Resolution Phase Contrast Imaging\n  Achieved by End-to-End Ptychographic Electron Tomography", "abstract": "  Three-dimensional atomic resolution imaging using transmission electron\nmicroscopes is a unique capability that requires challenging experiments.\nLinear electron tomography methods are limited by the missing wedge effect,\nrequiring a high tilt range. Multislice ptychography can achieve deep\nsub-{\\AA}ngstrom resolution in the transverse direction, but the depth\nresolution is limited to 2 to 3 nanometers. In this paper, we propose and\ndemonstrate an end-to-end approach to reconstructing the electrostatic\npotential volume of the sample directly from the 4D-STEM datasets. End-to-end\nmulti-slice ptychographic tomography recovers several slices at each tomography\ntilt angle and compensates for the missing wedge effect. The algorithm is\ninitially tested in simulation with a Pt@$\\mathrm{Al_2O_3}$ core-shell\nnanoparticle, where both heavy and light atoms are recovered in 3D from an\nunaligned 4D-STEM tilt series with a restricted tilt range of 90 degrees. We\nalso demonstrate the algorithm experimentally, recovering a Te nanoparticle\nwith sub-{\\AA}ngstrom resolution.\n", "date": "2024-07-28", "categories": []}
{"id": "2407.19866", "title": "Deep Image Priors for Magnetic Resonance Fingerprinting with pretrained\n  Bloch-consistent denoising autoencoders", "abstract": "  The estimation of multi-parametric quantitative maps from Magnetic Resonance\nFingerprinting (MRF) compressed sampled acquisitions, albeit successful,\nremains a challenge due to the high underspampling rate and artifacts naturally\noccuring during image reconstruction. Whilst state-of-the-art DL methods can\nsuccessfully address the task, to fully exploit their capabilities they often\nrequire training on a paired dataset, in an area where ground truth is seldom\navailable. In this work, we propose a method that combines a deep image prior\n(DIP) module that, without ground truth and in conjunction with a Bloch\nconsistency enforcing autoencoder, can tackle the problem, resulting in a\nmethod faster and of equivalent or better accuracy than DIP-MRF.\n", "date": "2024-07-29", "categories": []}
{"id": "2403.11035", "title": "Multiplane Quantitative Phase Imaging Using a Wavelength-Multiplexed\n  Diffractive Optical Processor", "abstract": "  Quantitative phase imaging (QPI) is a label-free technique that provides\noptical path length information for transparent specimens, finding utility in\nbiology, materials science, and engineering. Here, we present quantitative\nphase imaging of a 3D stack of phase-only objects using a\nwavelength-multiplexed diffractive optical processor. Utilizing multiple\nspatially engineered diffractive layers trained through deep learning, this\ndiffractive processor can transform the phase distributions of multiple 2D\nobjects at various axial positions into intensity patterns, each encoded at a\nunique wavelength channel. These wavelength-multiplexed patterns are projected\nonto a single field-of-view (FOV) at the output plane of the diffractive\nprocessor, enabling the capture of quantitative phase distributions of input\nobjects located at different axial planes using an intensity-only image sensor.\nBased on numerical simulations, we show that our diffractive processor could\nsimultaneously achieve all-optical quantitative phase imaging across several\ndistinct axial planes at the input by scanning the illumination wavelength. A\nproof-of-concept experiment with a 3D-fabricated diffractive processor further\nvalidated our approach, showcasing successful imaging of two distinct phase\nobjects at different axial positions by scanning the illumination wavelength in\nthe terahertz spectrum. Diffractive network-based multiplane QPI designs can\nopen up new avenues for compact on-chip phase imaging and sensing devices.\n", "date": "2024-03-16", "categories": []}
{"id": "2407.18500", "title": "Revisit Event Generation Model: Self-Supervised Learning of\n  Event-to-Video Reconstruction with Implicit Neural Representations", "abstract": "  Reconstructing intensity frames from event data while maintaining high\ntemporal resolution and dynamic range is crucial for bridging the gap between\nevent-based and frame-based computer vision. Previous approaches have depended\non supervised learning on synthetic data, which lacks interpretability and risk\nover-fitting to the setting of the event simulator. Recently, self-supervised\nlearning (SSL) based methods, which primarily utilize per-frame optical flow to\nestimate intensity via photometric constancy, has been actively investigated.\nHowever, they are vulnerable to errors in the case of inaccurate optical flow.\nThis paper proposes a novel SSL event-to-video reconstruction approach, dubbed\nEvINR, which eliminates the need for labeled data or optical flow estimation.\nOur core idea is to reconstruct intensity frames by directly addressing the\nevent generation model, essentially a partial differential equation (PDE) that\ndescribes how events are generated based on the time-varying brightness\nsignals. Specifically, we utilize an implicit neural representation (INR),\nwhich takes in spatiotemporal coordinate $(x, y, t)$ and predicts intensity\nvalues, to represent the solution of the event generation equation. The INR,\nparameterized as a fully-connected Multi-layer Perceptron (MLP), can be\noptimized with its temporal derivatives supervised by events. To make EvINR\nfeasible for online requisites, we propose several acceleration techniques that\nsubstantially expedite the training process. Comprehensive experiments\ndemonstrate that our EvINR surpasses previous SSL methods by 38% w.r.t. Mean\nSquared Error (MSE) and is comparable or superior to SoTA supervised methods.\nProject page: https://vlislab22.github.io/EvINR/.\n", "date": "2024-07-26", "categories": []}
{"id": "2312.13317", "title": "Deep Hybrid Camera Deblurring for Smartphone Cameras", "abstract": "  Mobile cameras, despite their significant advancements, still have difficulty\nin low-light imaging due to compact sensors and lenses, leading to longer\nexposures and motion blur. Traditional blind deconvolution methods and\nlearning-based deblurring methods can be potential solutions to remove blur.\nHowever, achieving practical performance still remains a challenge. To address\nthis, we propose a learning-based deblurring framework for smartphones,\nutilizing wide and ultra-wide cameras as a hybrid camera system. We\nsimultaneously capture a long-exposure wide image and short-exposure burst\nultra-wide images, and utilize the burst images to deblur the wide image. To\nfully exploit burst ultra-wide images, we present HCDeblur, a practical\ndeblurring framework that includes novel deblurring networks, HC-DNet and\nHC-FNet. HC-DNet utilizes motion information extracted from burst images to\ndeblur a wide image, and HC-FNet leverages burst images as reference images to\nfurther enhance a deblurred output. For training and evaluating the proposed\nmethod, we introduce the HCBlur dataset, which consists of synthetic and\nreal-world datasets. Our experiments demonstrate that HCDeblur achieves\nstate-of-the-art deblurring quality. Code and datasets are available at\nhttps://cg.postech.ac.kr/research/HCDeblur.\n", "date": "2023-12-20", "categories": []}
{"id": "2407.17780", "title": "HF-Fed: Hierarchical based customized Federated Learning Framework for\n  X-Ray Imaging", "abstract": "  In clinical applications, X-ray technology is vital for noninvasive\nexaminations like mammography, providing essential anatomical information.\nHowever, the radiation risk associated with X-ray procedures raises concerns.\nX-ray reconstruction is crucial in medical imaging for detailed visual\nrepresentations of internal structures, aiding diagnosis and treatment without\ninvasive procedures. Recent advancements in deep learning (DL) have shown\npromise in X-ray reconstruction, but conventional DL methods often require\ncentralized aggregation of large datasets, leading to domain shifts and privacy\nissues. To address these challenges, we introduce the Hierarchical\nFramework-based Federated Learning method (HF-Fed) for customized X-ray\nimaging. HF-Fed tackles X-ray imaging optimization by decomposing the problem\ninto local data adaptation and holistic X-ray imaging. It employs a\nhospital-specific hierarchical framework and a shared common imaging network\ncalled Network of Networks (NoN) to acquire stable features from diverse data\ndistributions. The hierarchical hypernetwork extracts domain-specific\nhyperparameters, conditioning the NoN for customized X-ray reconstruction.\nExperimental results demonstrate HF-Fed's competitive performance, offering a\npromising solution for enhancing X-ray imaging without data sharing. This study\nsignificantly contributes to the literature on federated learning in\nhealthcare, providing valuable insights for policymakers and healthcare\nproviders. The source code and pre-trained HF-Fed model are available at\n\\url{https://tisharepo.github.io/Webpage/}.\n", "date": "2024-07-25", "categories": []}
{"id": "2407.18035", "title": "RestoreAgent: Autonomous Image Restoration Agent via Multimodal Large\n  Language Models", "abstract": "  Natural images captured by mobile devices often suffer from multiple types of\ndegradation, such as noise, blur, and low light. Traditional image restoration\nmethods require manual selection of specific tasks, algorithms, and execution\nsequences, which is time-consuming and may yield suboptimal results. All-in-one\nmodels, though capable of handling multiple tasks, typically support only a\nlimited range and often produce overly smooth, low-fidelity outcomes due to\ntheir broad data distribution fitting. To address these challenges, we first\ndefine a new pipeline for restoring images with multiple degradations, and then\nintroduce RestoreAgent, an intelligent image restoration system leveraging\nmultimodal large language models. RestoreAgent autonomously assesses the type\nand extent of degradation in input images and performs restoration through (1)\ndetermining the appropriate restoration tasks, (2) optimizing the task\nsequence, (3) selecting the most suitable models, and (4) executing the\nrestoration. Experimental results demonstrate the superior performance of\nRestoreAgent in handling complex degradation, surpassing human experts.\nFurthermore, the system modular design facilitates the fast integration of new\ntasks and models, enhancing its flexibility and scalability for various\napplications.\n", "date": "2024-07-25", "categories": []}
{"id": "2312.02494", "title": "ReconU-Net: a direct PET image reconstruction using U-Net architecture\n  with back projection-induced skip connection", "abstract": "  [Objective] This study aims to introduce a novel back projection-induced\nU-Net-shaped architecture, called ReconU-Net, for deep learning-based direct\npositron emission tomography (PET) image reconstruction. Additionally, our\nobjective is to analyze the behavior of direct PET image reconstruction and\ngain deeper insights by comparing the proposed ReconU-Net architecture with\nother encoder-decoder architectures without skip connections. [Approach] The\nproposed ReconU-Net architecture uniquely integrates the physical model of the\nback projection operation into the skip connection. This distinctive feature\nfacilitates the effective transfer of intrinsic spatial information from the\ninput sinogram to the reconstructed image via an embedded physical model. The\nproposed ReconU-Net was trained using Monte Carlo simulation data from the\nBrainweb phantom and tested on both simulated and real Hoffman brain phantom\ndata. [Main results] The proposed ReconU-Net method generated a reconstructed\nimage with a more accurate structure compared to other deep learning-based\ndirect reconstruction methods. Further analysis showed that the proposed\nReconU-Net architecture has the ability to transfer features of multiple\nresolutions, especially non-abstract high-resolution information, through skip\nconnections. Despite limited training on simulated data, the proposed\nReconU-Net successfully reconstructed the real Hoffman brain phantom, unlike\nother deep learning-based direct reconstruction methods, which failed to\nproduce a reconstructed image. [Significance] The proposed ReconU-Net can\nimprove the fidelity of direct PET image reconstruction, even when dealing with\nsmall training datasets, by leveraging the synergistic relationship between\ndata-driven modeling and the physics model of the imaging process.\n", "date": "2023-12-05", "categories": []}
{"id": "2407.07503", "title": "Inter and Intra Prior Learning-based Hyperspectral Image Reconstruction\n  Using Snapshot SWIR Metasurface", "abstract": "  Shortwave-infrared(SWIR) spectral information, ranging from 1 {\\mu}m to\n2.5{\\mu}m, overcomes the limitations of traditional color cameras in acquiring\nscene information. However, conventional SWIR hyperspectral imaging systems\nface challenges due to their bulky setups and low acquisition speeds. This work\nintroduces a snapshot SWIR hyperspectral imaging system based on a metasurface\nfilter and a corresponding filter selection method to achieve the lowest\ncorrelation coefficient among these filters. This system offers the advantages\nof compact size and snapshot imaging. We propose a novel inter and intra prior\nlearning unfolding framework to achieve high-quality SWIR hyperspectral image\nreconstruction, which bridges the gap between prior learning and cross-stage\ninformation interaction. Additionally, We design an adaptive feature transfer\nmechanism to adaptively transfer the contextual correlation of multi-scale\nencoder features to prevent detailed information loss in the decoder.\nExperiment results demonstrate that our method can reconstruct hyperspectral\nimages with high speed and superior performance over existing methods.\n", "date": "2024-07-10", "categories": []}
{"id": "2407.15964", "title": "FDWST: Fingerphoto Deblurring using Wavelet Style Transfer", "abstract": "  The challenge of deblurring fingerphoto images, or generating a sharp\nfingerphoto from a given blurry one, is a significant problem in the realm of\ncomputer vision. To address this problem, we propose a fingerphoto deblurring\narchitecture referred to as Fingerphoto Deblurring using Wavelet Style Transfer\n(FDWST), which aims to utilize the information transmission of Style Transfer\ntechniques to deblur fingerphotos. Additionally, we incorporate the Discrete\nWavelet Transform (DWT) for its ability to split images into different\nfrequency bands. By combining these two techniques, we can perform Style\nTransfer over a wide array of wavelet frequency bands, thereby increasing the\nquality and variety of sharpness information transferred from sharp to blurry\nimages. Using this technique, our model was able to drastically increase the\nquality of the generated fingerphotos compared to their originals, and achieve\na peak matching accuracy of 0.9907 when tasked with matching a deblurred\nfingerphoto to its sharp counterpart, outperforming multiple other\nstate-of-the-art deblurring and style transfer techniques.\n", "date": "2024-07-22", "categories": []}
{"id": "2407.16125", "title": "Diffusion Prior-Based Amortized Variational Inference for Noisy Inverse\n  Problems", "abstract": "  Recent studies on inverse problems have proposed posterior samplers that\nleverage the pre-trained diffusion models as powerful priors. These attempts\nhave paved the way for using diffusion models in a wide range of inverse\nproblems. However, the existing methods entail computationally demanding\niterative sampling procedures and optimize a separate solution for each\nmeasurement, which leads to limited scalability and lack of generalization\ncapability across unseen samples. To address these limitations, we propose a\nnovel approach, Diffusion prior-based Amortized Variational Inference (DAVI)\nthat solves inverse problems with a diffusion prior from an amortized\nvariational inference perspective. Specifically, instead of separate\nmeasurement-wise optimization, our amortized inference learns a function that\ndirectly maps measurements to the implicit posterior distributions of\ncorresponding clean data, enabling a single-step posterior sampling even for\nunseen measurements. Extensive experiments on image restoration tasks, e.g.,\nGaussian deblur, 4$\\times$ super-resolution, and box inpainting with two\nbenchmark datasets, demonstrate our approach's superior performance over strong\nbaselines. Code is available at https://github.com/mlvlab/DAVI.\n", "date": "2024-07-23", "categories": []}
{"id": "2407.16308", "title": "SAFNet: Selective Alignment Fusion Network for Efficient HDR Imaging", "abstract": "  Multi-exposure High Dynamic Range (HDR) imaging is a challenging task when\nfacing truncated texture and complex motion. Existing deep learning-based\nmethods have achieved great success by either following the alignment and\nfusion pipeline or utilizing attention mechanism. However, the large\ncomputation cost and inference delay hinder them from deploying on resource\nlimited devices. In this paper, to achieve better efficiency, a novel Selective\nAlignment Fusion Network (SAFNet) for HDR imaging is proposed. After extracting\npyramid features, it jointly refines valuable area masks and cross-exposure\nmotion in selected regions with shared decoders, and then fuses high quality\nHDR image in an explicit way. This approach can focus the model on finding\nvaluable regions while estimating their easily detectable and meaningful\nmotion. For further detail enhancement, a lightweight refine module is\nintroduced which enjoys privileges from previous optical flow, selection masks\nand initial prediction. Moreover, to facilitate learning on samples with large\nmotion, a new window partition cropping method is presented during training.\nExperiments on public and newly developed challenging datasets show that\nproposed SAFNet not only exceeds previous SOTA competitors quantitatively and\nqualitatively, but also runs order of magnitude faster. Code and dataset is\navailable at https://github.com/ltkong218/SAFNet.\n", "date": "2024-07-23", "categories": []}
{"id": "2211.16986", "title": "A Geometric Model for Polarization Imaging on Projective Cameras", "abstract": "  The vast majority of Shape-from-Polarization (SfP) methods work under the\noversimplified assumption of using orthographic cameras. Indeed, it is still\nnot well understood how to project the Stokes vectors when the incoming rays\nare not orthogonal to the image plane. We try to answer this question\npresenting a geometric model describing how a general projective camera\ncaptures the light polarization state. Based on the optical properties of a\ntilted polarizer, our model is implemented as a pre-processing operation acting\non raw images, followed by a per-pixel rotation of the reconstructed normal\nfield. In this way, all the existing SfP methods assuming orthographic cameras\ncan behave like they were designed for projective ones. Moreover, our model is\nconsistent with state-of-the-art forward and inverse renderers (like Mitsuba3\nand ART), intrinsically enforces physical constraints among the captured\nchannels, and handles demosaicing of DoFP sensors. Experiments on existing and\nnew datasets demonstrate the accuracy of the model when applied to commercially\navailable polarimetric cameras.\n", "date": "2022-11-29", "categories": []}
{"id": "2407.14564", "title": "APS-USCT: Ultrasound Computed Tomography on Sparse Data via AI-Physic\n  Synergy", "abstract": "  Ultrasound computed tomography (USCT) is a promising technique that achieves\nsuperior medical imaging reconstruction resolution by fully leveraging waveform\ninformation, outperforming conventional ultrasound methods. Despite its\nadvantages, high-quality USCT reconstruction relies on extensive data\nacquisition by a large number of transducers, leading to increased costs,\ncomputational demands, extended patient scanning times, and manufacturing\ncomplexities. To mitigate these issues, we propose a new USCT method called\nAPS-USCT, which facilitates imaging with sparse data, substantially reducing\ndependence on high-cost dense data acquisition. Our APS-USCT method consists of\ntwo primary components: APS-wave and APS-FWI. The APS-wave component, an\nencoder-decoder system, preprocesses the waveform data, converting sparse data\ninto dense waveforms to augment sample density prior to reconstruction. The\nAPS-FWI component, utilizing the InversionNet, directly reconstructs the speed\nof sound (SOS) from the ultrasound waveform data. We further improve the\nmodel's performance by incorporating Squeeze-and-Excitation (SE) Blocks and\nsource encoding techniques. Testing our method on a breast cancer dataset\nyielded promising results. It demonstrated outstanding performance with an\naverage Structural Similarity Index (SSIM) of 0.8431. Notably, over 82% of\nsamples achieved an SSIM above 0.8, with nearly 61% exceeding 0.85,\nhighlighting the significant potential of our approach in improving USCT image\nreconstruction by efficiently utilizing sparse data.\n", "date": "2024-07-18", "categories": []}
{"id": "2407.14816", "title": "Blind Image Deconvolution by Generative-based Kernel Prior and\n  Initializer via Latent Encoding", "abstract": "  Blind image deconvolution (BID) is a classic yet challenging problem in the\nfield of image processing. Recent advances in deep image prior (DIP) have\nmotivated a series of DIP-based approaches, demonstrating remarkable success in\nBID. However, due to the high non-convexity of the inherent optimization\nprocess, these methods are notorious for their sensitivity to the initialized\nkernel. To alleviate this issue and further improve their performance, we\npropose a new framework for BID that better considers the prior modeling and\nthe initialization for blur kernels, leveraging a deep generative model. The\nproposed approach pre-trains a generative adversarial network-based kernel\ngenerator that aptly characterizes the kernel priors and a kernel initializer\nthat facilitates a well-informed initialization for the blur kernel through\nlatent space encoding. With the pre-trained kernel generator and initializer,\none can obtain a high-quality initialization of the blur kernel, and enable\noptimization within a compact latent kernel manifold. Such a framework results\nin an evident performance improvement over existing DIP-based BID methods.\nExtensive experiments on different datasets demonstrate the effectiveness of\nthe proposed method.\n", "date": "2024-07-20", "categories": []}
{"id": "2407.14868", "title": "Dual High-Order Total Variation Model for Underwater Image Restoration", "abstract": "  Underwater images are typically characterized by color cast, haze, blurring,\nand uneven illumination due to the selective absorption and scattering when\nlight propagates through the water, which limits their practical applications.\nUnderwater image enhancement and restoration (UIER) is one crucial mode to\nimprove the visual quality of underwater images. However, most existing UIER\nmethods concentrate on enhancing contrast and dehazing, rarely pay attention to\nthe local illumination differences within the image caused by illumination\nvariations, thus introducing some undesirable artifacts and unnatural color. To\naddress this issue, an effective variational framework is proposed based on an\nextended underwater image formation model (UIFM). Technically, dual high-order\nregularizations are successfully integrated into the variational model to\nacquire smoothed local ambient illuminance and structure-revealed reflectance\nin a unified manner. In our proposed framework, the weight factors-based color\ncompensation is combined with the color balance to compensate for the\nattenuated color channels and remove the color cast. In particular, the local\nambient illuminance with strong robustness is acquired by performing the local\npatch brightest pixel estimation and an improved gamma correction.\nAdditionally, we design an iterative optimization algorithm relying on the\nalternating direction method of multipliers (ADMM) to accelerate the solution\nof the proposed variational model. Considerable experiments on three real-world\nunderwater image datasets demonstrate that the proposed method outperforms\nseveral state-of-the-art methods with regard to visual quality and quantitative\nassessments. Moreover, the proposed method can also be extended to outdoor\nimage dehazing, low-light image enhancement, and some high-level vision tasks.\nThe code is available at https://github.com/Hou-Guojia/UDHTV.\n", "date": "2024-07-20", "categories": []}
{"id": "2407.14899", "title": "Hyperspectral Unmixing Under Endmember Variability: A Variational\n  Inference Framework", "abstract": "  This work proposes a variational inference (VI) framework for hyperspectral\nunmixing in the presence of endmember variability (HU-EV). An EV-accounted\nnoisy linear mixture model (LMM) is considered, and the presence of outliers is\nalso incorporated into the model. Following the marginalized maximum likelihood\n(MML) principle, a VI algorithmic structure is designed for probabilistic\ninference for HU-EV. Specifically, a patch-wise static endmember assumption is\nemployed to exploit spatial smoothness and to try to overcome the ill-posed\nnature of the HU-EV problem. The design facilitates lightweight, continuous\noptimization-based updates under a variety of endmember priors. Some of the\npriors, such as the Beta prior, were previously used under computationally\nheavy, sampling-based probabilistic HU-EV methods. The effectiveness of the\nproposed framework is demonstrated through synthetic, semi-real, and real-data\nexperiments.\n", "date": "2024-07-20", "categories": []}
{"id": "2407.08377", "title": "Long-range Turbulence Mitigation: A Large-scale Dataset and A\n  Coarse-to-fine Framework", "abstract": "  Long-range imaging inevitably suffers from atmospheric turbulence with severe\ngeometric distortions due to random refraction of light. The further the\ndistance, the more severe the disturbance. Despite existing research has\nachieved great progress in tackling short-range turbulence, there is less\nattention paid to long-range turbulence with significant distortions. To\naddress this dilemma and advance the field, we construct a large-scale real\nlong-range atmospheric turbulence dataset (RLR-AT), including 1500 turbulence\nsequences spanning distances from 1 Km to 13 Km. The advantages of RLR-AT\ncompared to existing ones: turbulence with longer-distances and\nhigher-diversity, scenes with greater-variety and larger-scale. Moreover, most\nexisting work adopts either registration-based or decomposition-based methods\nto address distortions through one-step mitigation. However, they fail to\neffectively handle long-range turbulence due to its significant pixel\ndisplacements. In this work, we propose a coarse-to-fine framework to handle\nsevere distortions, which cooperates dynamic turbulence and static background\npriors (CDSP). On the one hand, we discover the pixel motion statistical prior\nof turbulence, and propose a frequency-aware reference frame for better\nlarge-scale distortion registration, greatly reducing the burden of refinement.\nOn the other hand, we take advantage of the static prior of background, and\npropose a subspace-based low-rank tensor refinement model to eliminate the\nmisalignments inevitably left by registration while well preserving details.\nThe dynamic and static priors complement to each other, facilitating us to\nprogressively mitigate long-range turbulence with severe distortions. Extensive\nexperiments demonstrate that the proposed method outperforms SOTA methods on\ndifferent datasets.\n", "date": "2024-07-11", "categories": []}
{"id": "2407.11946", "title": "Hierarchical Separable Video Transformer for Snapshot Compressive\n  Imaging", "abstract": "  Transformers have achieved the state-of-the-art performance on solving the\ninverse problem of Snapshot Compressive Imaging (SCI) for video, whose\nill-posedness is rooted in the mixed degradation of spatial masking and\ntemporal aliasing. However, previous Transformers lack an insight into the\ndegradation and thus have limited performance and efficiency. In this work, we\ntailor an efficient reconstruction architecture without temporal aggregation in\nearly layers and Hierarchical Separable Video Transformer (HiSViT) as building\nblock. HiSViT is built by multiple groups of Cross-Scale Separable Multi-head\nSelf-Attention (CSS-MSA) and Gated Self-Modulated Feed-Forward Network\n(GSM-FFN) with dense connections, each of which is conducted within a separate\nchannel portions at a different scale, for multi-scale interactions and\nlong-range modeling. By separating spatial operations from temporal ones,\nCSS-MSA introduces an inductive bias of paying more attention within frames\ninstead of between frames while saving computational overheads. GSM-FFN further\nenhances the locality via gated mechanism and factorized spatial-temporal\nconvolutions. Extensive experiments demonstrate that our method outperforms\nprevious methods by $\\!>\\!0.5$ dB with comparable or fewer parameters and\ncomplexity. The source codes and pretrained models are released at\nhttps://github.com/pwangcs/HiSViT.\n", "date": "2024-07-16", "categories": []}
{"id": "2312.11595", "title": "SPIRE: Semantic Prompt-Driven Image Restoration", "abstract": "  Text-driven diffusion models have become increasingly popular for various\nimage editing tasks, including inpainting, stylization, and object replacement.\nHowever, it still remains an open research problem to adopt this\nlanguage-vision paradigm for more fine-level image processing tasks, such as\ndenoising, super-resolution, deblurring, and compression artifact removal. In\nthis paper, we develop SPIRE, a Semantic and restoration Prompt-driven Image\nRestoration framework that leverages natural language as a user-friendly\ninterface to control the image restoration process. We consider the capacity of\nprompt information in two dimensions. First, we use content-related prompts to\nenhance the semantic alignment, effectively alleviating identity ambiguity in\nthe restoration outcomes. Second, our approach is the first framework that\nsupports fine-level instruction through language-based quantitative\nspecification of the restoration strength, without the need for explicit\ntask-specific design. In addition, we introduce a novel fusion mechanism that\naugments the existing ControlNet architecture by learning to rescale the\ngenerative prior, thereby achieving better restoration fidelity. Our extensive\nexperiments demonstrate the superior restoration performance of SPIRE compared\nto the state of the arts, alongside offering the flexibility of text-based\ncontrol over the restoration effects.\n", "date": "2023-12-18", "categories": []}
{"id": "2407.11685", "title": "Deconvolution with a Box", "abstract": "  Deconvolution with a box (square wave) is a key operation for\nsuper-resolution with pixel-shift cameras. In general convolution with a box is\nnot invertible. However, we can obtain perfect reconstructions of sparse\nsignals using convex optimization. We give a direct proof that improves on the\nreconstruction bound that follows from previous results. We also show our bound\nis tight and matches an information theoretic limit.\n", "date": "2024-07-16", "categories": []}
{"id": "2404.10700", "title": "Rawformer: Unpaired Raw-to-Raw Translation for Learnable Camera ISPs", "abstract": "  Modern smartphone camera quality heavily relies on the image signal processor\n(ISP) to enhance captured raw images, utilizing carefully designed modules to\nproduce final output images encoded in a standard color space (e.g., sRGB).\nNeural-based end-to-end learnable ISPs offer promising advancements,\npotentially replacing traditional ISPs with their ability to adapt without\nrequiring extensive tuning for each new camera model, as is often the case for\nnearly every module in traditional ISPs. However, the key challenge with the\nrecent learning-based ISPs is the urge to collect large paired datasets for\neach distinct camera model due to the influence of intrinsic camera\ncharacteristics on the formation of input raw images. This paper tackles this\nchallenge by introducing a novel method for unpaired learning of raw-to-raw\ntranslation across diverse cameras. Specifically, we propose Rawformer, an\nunsupervised Transformer-based encoder-decoder method for raw-to-raw\ntranslation. It accurately maps raw images captured by a certain camera to the\ntarget camera, facilitating the generalization of learnable ISPs to new unseen\ncameras. Our method demonstrates superior performance on real camera datasets,\nachieving higher accuracy compared to previous state-of-the-art techniques, and\npreserving a more robust correlation between the original and translated raw\nimages. The codes and the pretrained models are available at\nhttps://github.com/gosha20777/rawformer.\n", "date": "2024-04-16", "categories": []}
{"id": "2407.09525", "title": "A Deep Learning Framework for Three Dimensional Shape Reconstruction\n  from Phaseless Acoustic Scattering Far-field Data", "abstract": "  The inverse scattering problem is of critical importance in a number of\nfields, including medical imaging, sonar, sensing, non-destructive evaluation,\nand several others. The problem of interest can vary from detecting the shape\nto the constitutive properties of the obstacle. The challenge in both is that\nthis problem is ill-posed, more so when there is limited information. That\nsaid, significant effort has been expended over the years in developing\nsolutions to this problem. Here, we use a different approach, one that is\nfounded on data. Specifically, we develop a deep learning framework for shape\nreconstruction using limited information with single incident wave, single\nfrequency, and phase-less far-field data. This is done by (a) using a compact\nprobabilistic shape latent space, learned by a 3D variational auto-encoder, and\n(b) a convolutional neural network trained to map the acoustic scattering\ninformation to this shape representation. The proposed framework is evaluated\non a synthetic 3D particle dataset, as well as ShapeNet, a popular 3D shape\nrecognition dataset. As demonstrated via a number of results, the proposed\nmethod is able to produce accurate reconstructions for large batches of complex\nscatterer shapes (such as airplanes and automobiles), despite the significant\nvariation present within the data.\n", "date": "2024-06-24", "categories": []}
{"id": "2407.10559", "title": "LIP-CAR: contrast agent reduction by a deep learned inverse problem", "abstract": "  The adoption of contrast agents in medical imaging protocols is crucial for\naccurate and timely diagnosis. While highly effective and characterized by an\nexcellent safety profile, the use of contrast agents has its limitation,\nincluding rare risk of allergic reactions, potential environmental impact and\neconomic burdens on patients and healthcare systems. In this work, we address\nthe contrast agent reduction (CAR) problem, which involves reducing the\nadministered dosage of contrast agent while preserving the visual enhancement.\nThe current literature on the CAR task is based on deep learning techniques\nwithin a fully image processing framework. These techniques digitally simulate\nhigh-dose images from images acquired with a low dose of contrast agent. We\ninvestigate the feasibility of a ``learned inverse problem'' (LIP) approach, as\nopposed to the end-to-end paradigm in the state-of-the-art literature.\n  Specifically, we learn the image-to-image operator that maps high-dose images\nto their corresponding low-dose counterparts, and we frame the CAR task as an\ninverse problem. We then solve this problem through a regularized optimization\nreformulation. Regularization methods are well-established mathematical\ntechniques that offer robustness and explainability. Our approach combines\nthese rigorous techniques with cutting-edge deep learning tools. Numerical\nexperiments performed on pre-clinical medical images confirm the effectiveness\nof this strategy, showing improved stability and accuracy in the simulated\nhigh-dose images.\n", "date": "2024-07-15", "categories": []}
{"id": "2407.10641", "title": "Deep Diffusion Image Prior for Efficient OOD Adaptation in 3D Inverse\n  Problems", "abstract": "  Recent inverse problem solvers that leverage generative diffusion priors have\ngarnered significant attention due to their exceptional quality. However,\nadaptation of the prior is necessary when there exists a discrepancy between\nthe training and testing distributions. In this work, we propose deep diffusion\nimage prior (DDIP), which generalizes the recent adaptation method of SCD by\nintroducing a formal connection to the deep image prior. Under this framework,\nwe propose an efficient adaptation method dubbed D3IP, specified for 3D\nmeasurements, which accelerates DDIP by orders of magnitude while achieving\nsuperior performance. D3IP enables seamless integration of 3D inverse solvers\nand thus leads to coherent 3D reconstruction. Moreover, we show that\nmeta-learning techniques can also be applied to yield even better performance.\nWe show that our method is capable of solving diverse 3D reconstructive tasks\nfrom the generative prior trained only with phantom images that are vastly\ndifferent from the training set, opening up new opportunities of applying\ndiffusion inverse solvers even when training with gold standard data is\nimpossible. Code: https://github.com/HJ-harry/DDIP3D\n", "date": "2024-07-15", "categories": []}
{"id": "2403.01105", "title": "Depth Information Assisted Collaborative Mutual Promotion Network for\n  Single Image Dehazing", "abstract": "  Recovering a clear image from a single hazy image is an open inverse problem.\nAlthough significant research progress has been made, most existing methods\nignore the effect that downstream tasks play in promoting upstream dehazing.\nFrom the perspective of the haze generation mechanism, there is a potential\nrelationship between the depth information of the scene and the hazy image.\nBased on this, we propose a dual-task collaborative mutual promotion framework\nto achieve the dehazing of a single image. This framework integrates depth\nestimation and dehazing by a dual-task interaction mechanism and achieves\nmutual enhancement of their performance. To realize the joint optimization of\nthe two tasks, an alternative implementation mechanism with the difference\nperception is developed. On the one hand, the difference perception between the\ndepth maps of the dehazing result and the ideal image is proposed to promote\nthe dehazing network to pay attention to the non-ideal areas of the dehazing.\nOn the other hand, by improving the depth estimation performance in the\ndifficult-to-recover areas of the hazy image, the dehazing network can\nexplicitly use the depth information of the hazy image to assist the clear\nimage recovery. To promote the depth estimation, we propose to use the\ndifference between the dehazed image and the ground truth to guide the depth\nestimation network to focus on the dehazed unideal areas. It allows dehazing\nand depth estimation to leverage their strengths in a mutually reinforcing\nmanner. Experimental results show that the proposed method can achieve better\nperformance than that of the state-of-the-art approaches.\n", "date": "2024-03-02", "categories": []}
{"id": "2406.10744", "title": "Technique Report of CVPR 2024 PBDL Challenges", "abstract": "  The intersection of physics-based vision and deep learning presents an\nexciting frontier for advancing computer vision technologies. By leveraging the\nprinciples of physics to inform and enhance deep learning models, we can\ndevelop more robust and accurate vision systems. Physics-based vision aims to\ninvert the processes to recover scene properties such as shape, reflectance,\nlight distribution, and medium properties from images. In recent years, deep\nlearning has shown promising improvements for various vision tasks, and when\ncombined with physics-based vision, these approaches can enhance the robustness\nand accuracy of vision systems. This technical report summarizes the outcomes\nof the Physics-Based Vision Meets Deep Learning (PBDL) 2024 challenge, held in\nCVPR 2024 workshop. The challenge consisted of eight tracks, focusing on\nLow-Light Enhancement and Detection as well as High Dynamic Range (HDR)\nImaging. This report details the objectives, methodologies, and results of each\ntrack, highlighting the top-performing solutions and their innovative\napproaches.\n", "date": "2024-06-15", "categories": []}
{"id": "2407.09352", "title": "Imaging Interiors: An Implicit Solution to Electromagnetic Inverse\n  Scattering Problems", "abstract": "  Electromagnetic Inverse Scattering Problems (EISP) have gained wide\napplications in computational imaging. By solving EISP, the internal relative\npermittivity of the scatterer can be non-invasively determined based on the\nscattered electromagnetic fields. Despite previous efforts to address EISP,\nachieving better solutions to this problem has remained elusive, due to the\nchallenges posed by inversion and discretization. This paper tackles those\nchallenges in EISP via an implicit approach. By representing the scatterer's\nrelative permittivity as a continuous implicit representation, our method is\nable to address the low-resolution problems arising from discretization.\nFurther, optimizing this implicit representation within a forward framework\nallows us to conveniently circumvent the challenges posed by inverse\nestimation. Our approach outperforms existing methods on standard benchmark\ndatasets. Project page: https://luo-ziyuan.github.io/Imaging-Interiors\n", "date": "2024-07-12", "categories": []}
{"id": "2305.15078", "title": "UniINR: Event-guided Unified Rolling Shutter Correction, Deblurring, and\n  Interpolation", "abstract": "  Video frames captured by rolling shutter (RS) cameras during fast camera\nmovement frequently exhibit RS distortion and blur simultaneously. Naturally,\nrecovering high-frame-rate global shutter (GS) sharp frames from an RS blur\nframe must simultaneously consider RS correction, deblur, and frame\ninterpolation. A naive way is to decompose the whole process into separate\ntasks and cascade existing methods; however, this results in cumulative errors\nand noticeable artifacts. Event cameras enjoy many advantages, e.g., high\ntemporal resolution, making them potential for our problem. To this end, we\npropose the first and novel approach, named UniINR, to recover arbitrary\nframe-rate sharp GS frames from an RS blur frame and paired events. Our key\nidea is unifying spatial-temporal implicit neural representation (INR) to\ndirectly map the position and time coordinates to color values to address the\ninterlocking degradations. Specifically, we introduce spatial-temporal implicit\nencoding (STE) to convert an RS blur image and events into a spatial-temporal\nrepresentation (STR). To query a specific sharp frame (GS or RS), we embed the\nexposure time into STR and decode the embedded features pixel-by-pixel to\nrecover a sharp frame. Our method features a lightweight model with only 0.38M\nparameters, and it also enjoys high inference efficiency, achieving\n2.83ms/frame in 31 times frame interpolation of an RS blur frame. Extensive\nexperiments show that our method significantly outperforms prior methods. Code\nis available at https://github.com/yunfanLu/UniINR.\n", "date": "2023-05-24", "categories": []}
{"id": "2311.14473", "title": "Joint Diffusion: Mutual Consistency-Driven Diffusion Model for PET-MRI\n  Co-Reconstruction", "abstract": "  Positron Emission Tomography and Magnetic Resonance Imaging (PET-MRI) systems\ncan obtain functional and anatomical scans. PET suffers from a low\nsignal-to-noise ratio. Meanwhile, the k-space data acquisition process in MRI\nis time-consuming. The study aims to accelerate MRI and enhance PET image\nquality. Conventional approaches involve the separate reconstruction of each\nmodality within PET-MRI systems. However, there exists complementary\ninformation among multi-modal images. The complementary information can\ncontribute to image reconstruction. In this study, we propose a novel PET-MRI\njoint reconstruction model employing a mutual consistency-driven diffusion\nmode, namely MC-Diffusion. MC-Diffusion learns the joint probability\ndistribution of PET and MRI for utilizing complementary information. We\nconducted a series of contrast experiments about LPLS, Joint ISAT-net and\nMC-Diffusion by the ADNI dataset. The results underscore the qualitative and\nquantitative improvements achieved by MC-Diffusion, surpassing the\nstate-of-the-art method.\n", "date": "2023-11-24", "categories": []}
{"id": "2407.08149", "title": "Deep Polarization Cues for Single-shot Shape and Subsurface Scattering\n  Estimation", "abstract": "  In this work, we propose a novel learning-based method to jointly estimate\nthe shape and subsurface scattering (SSS) parameters of translucent objects by\nutilizing polarization cues. Although polarization cues have been used in\nvarious applications, such as shape from polarization (SfP), BRDF estimation,\nand reflection removal, their application in SSS estimation has not yet been\nexplored. Our observations indicate that the SSS affects not only the light\nintensity but also the polarization signal. Hence, the polarization signal can\nprovide additional cues for SSS estimation. We also introduce the first\nlarge-scale synthetic dataset of polarized translucent objects for training our\nmodel. Our method outperforms several baselines from the SfP and inverse\nrendering realms on both synthetic and real data, as demonstrated by\nqualitative and quantitative results.\n", "date": "2024-07-11", "categories": []}
{"id": "2407.08252", "title": "Spatially-Variant Degradation Model for Dataset-free Super-resolution", "abstract": "  This paper focuses on the dataset-free Blind Image Super-Resolution (BISR).\nUnlike existing dataset-free BISR methods that focus on obtaining a degradation\nkernel for the entire image, we are the first to explicitly design a\nspatially-variant degradation model for each pixel. Our method also benefits\nfrom having a significantly smaller number of learnable parameters compared to\ndata-driven spatially-variant BISR methods. Concretely, each pixel's\ndegradation kernel is expressed as a linear combination of a learnable\ndictionary composed of a small number of spatially-variant atom kernels. The\ncoefficient matrices of the atom degradation kernels are derived using\nmembership functions of fuzzy set theory. We construct a novel Probabilistic\nBISR model with tailored likelihood function and prior terms. Subsequently, we\nemploy the Monte Carlo EM algorithm to infer the degradation kernels for each\npixel. Our method achieves a significant improvement over other\nstate-of-the-art BISR methods, with an average improvement of 1 dB (2x).Code\nwill be released at https://github.com/shaojieguoECNU/SVDSR.\n", "date": "2024-07-11", "categories": []}
{"id": "2402.12072", "title": "Robustness and Exploration of Variational and Machine Learning\n  Approaches to Inverse Problems: An Overview", "abstract": "  This paper provides an overview of current approaches for solving inverse\nproblems in imaging using variational methods and machine learning. A special\nfocus lies on point estimators and their robustness against adversarial\nperturbations. In this context results of numerical experiments for a\none-dimensional toy problem are provided, showing the robustness of different\napproaches and empirically verifying theoretical guarantees. Another focus of\nthis review is the exploration of the subspace of data-consistent solutions\nthrough explicit guidance to satisfy specific semantic or textural properties.\n", "date": "2024-02-19", "categories": []}
{"id": "2407.06343", "title": "Novel Models for High-Dimensional Imaging: High-Resolution fMRI\n  Acceleration and Quantification", "abstract": "  The goals of functional Magnetic Resonance Imaging (fMRI) include high\nspatial and temporal resolutions with a high signal-to-noise ratio (SNR). To\nsimultaneously improve spatial and temporal resolutions and maintain the high\nSNR advantage of OSSI, we present novel pipelines for fast acquisition and\nhigh-resolution fMRI reconstruction and physics parameter quantification. We\npropose a patch-tensor low-rank model, a physics-based manifold model, and a\nvoxel-wise attention network. With novel models for acquisition and\nreconstruction, we demonstrate that we can improve SNR and resolution\nsimultaneously without compromising scan time. All the proposed models\noutperform other comparison approaches with higher resolution and more\nfunctional information.\n", "date": "2024-07-08", "categories": []}
{"id": "2407.06608", "title": "Iteratively Refined Image Reconstruction with Learned Attentive\n  Regularizers", "abstract": "  We propose a regularization scheme for image reconstruction that leverages\nthe power of deep learning while hinging on classic sparsity-promoting models.\nMany deep-learning-based models are hard to interpret and cumbersome to analyze\ntheoretically. In contrast, our scheme is interpretable because it corresponds\nto the minimization of a series of convex problems. For each problem in the\nseries, a mask is generated based on the previous solution to refine the\nregularization strength spatially. In this way, the model becomes progressively\nattentive to the image structure. For the underlying update operator, we prove\nthe existence of a fixed point. As a special case, we investigate a mask\ngenerator for which the fixed-point iterations converge to a critical point of\nan explicit energy functional. In our experiments, we match the performance of\nstate-of-the-art learned variational models for the solution of inverse\nproblems. Additionally, we offer a promising balance between interpretability,\ntheoretical guarantees, reliability, and performance.\n", "date": "2024-07-09", "categories": []}
{"id": "2407.06727", "title": "Towards Physics-informed Cyclic Adversarial Multi-PSF Lensless Imaging", "abstract": "  Lensless imaging has emerged as a promising field within inverse imaging,\noffering compact, cost-effective solutions with the potential to revolutionize\nthe computational camera market. By circumventing traditional optical\ncomponents like lenses and mirrors, novel approaches like mask-based lensless\nimaging eliminate the need for conventional hardware. However, advancements in\nlensless image reconstruction, particularly those leveraging Generative\nAdversarial Networks (GANs), are hindered by the reliance on data-driven\ntraining processes, resulting in network specificity to the Point Spread\nFunction (PSF) of the imaging system. This necessitates a complete retraining\nfor minor PSF changes, limiting adaptability and generalizability across\ndiverse imaging scenarios. In this paper, we introduce a novel approach to\nmulti-PSF lensless imaging, employing a dual discriminator cyclic adversarial\nframework. We propose a unique generator architecture with a sparse\nconvolutional PSF-aware auxiliary branch, coupled with a forward model\nintegrated into the training loop to facilitate physics-informed learning to\nhandle the substantial domain gap between lensless and lensed images.\nComprehensive performance evaluation and ablation studies underscore the\neffectiveness of our model, offering robust and adaptable lensless image\nreconstruction capabilities. Our method achieves comparable performance to\nexisting PSF-agnostic generative methods for single PSF cases and demonstrates\nresilience to PSF changes without the need for retraining.\n", "date": "2024-07-09", "categories": []}
{"id": "2401.10461", "title": "Learning to Robustly Reconstruct Low-light Dynamic Scenes from Spike\n  Streams", "abstract": "  As a neuromorphic sensor with high temporal resolution, spike camera can\ngenerate continuous binary spike streams to capture per-pixel light intensity.\nWe can use reconstruction methods to restore scene details in high-speed\nscenarios. However, due to limited information in spike streams, low-light\nscenes are difficult to effectively reconstruct. In this paper, we propose a\nbidirectional recurrent-based reconstruction framework, including a\nLight-Robust Representation (LR-Rep) and a fusion module, to better handle such\nextreme conditions. LR-Rep is designed to aggregate temporal information in\nspike streams, and a fusion module is utilized to extract temporal features.\nAdditionally, we have developed a reconstruction benchmark for high-speed\nlow-light scenes. Light sources in the scenes are carefully aligned to\nreal-world conditions. Experimental results demonstrate the superiority of our\nmethod, which also generalizes well to real spike streams. Related codes and\nproposed datasets will be released after publication.\n", "date": "2024-01-19", "categories": []}
{"id": "2404.06657", "title": "Res-U2Net: Untrained Deep Learning for Phase Retrieval and Image\n  Reconstruction", "abstract": "  Conventional deep learning-based image reconstruction methods require a large\namount of training data which can be hard to obtain in practice. Untrained deep\nlearning methods overcome this limitation by training a network to invert a\nphysical model of the image formation process. Here we present a novel\nuntrained Res-U2Net model for phase retrieval. We use the extracted phase\ninformation to determine changes in an object's surface and generate a mesh\nrepresentation of its 3D structure. We compare the performance of Res-U2Net\nphase retrieval against UNet and U2Net using images from the GDXRAY dataset.\n", "date": "2024-04-09", "categories": []}
{"id": "2407.05279", "title": "A Hybrid Registration and Fusion Method for Hyperspectral\n  Super-resolution", "abstract": "  Fusing hyperspectral images (HSIs) with multispectral images (MSIs) has\nbecome a mainstream approach to enhance the spatial resolution of HSIs. Many\nHSI-MSI fusion methods have achieved impressive results. Nevertheless, certain\nchallenges persist, including: (a) A majority of current methods rely on\naccurate registration of HSI and MSI, which can be challenging in real-world\napplications.(b) The obtained HSI-MSI pairs may not be fully utilized. In this\npaper, we propose a hybrid registration and fusion constrained optimization\nmodel named RAF-NLRGS. With respect to challenge (a), the RAF model integrates\nbatch image alignment within the fusion process, facilitating simultaneous\nexecution of image registration and fusion. To address issue (b), the NLRGS\nmodel incorporates a nonconvex low-rank and group-sparse structure, leveraging\ngroup sparsity to effectively harness valuable information embedded in the\nresidual data. Moreover, the NLRGS model can further enhance fusion performance\nbased on the RAF model. Subsequently, the RAF-NLRGS model is solved within the\nframework of Generalized Gauss-Newton (GGN) algorithm and Proximal Alternating\nOptimization (PAO) algorithm. Theoretically, we establish the error bounds for\nthe NLRGS model and the convergence analysis of corresponding algorithms is\nalso presented. Finally, extensive numerical experiments on HSI datasets are\nconducted to verify the effectiveness of our method.\n", "date": "2024-07-07", "categories": []}
{"id": "2306.12992", "title": "Minimalist and High-Quality Panoramic Imaging with PSF-aware\n  Transformers", "abstract": "  High-quality panoramic images with a Field of View (FoV) of 360{\\deg} are\nessential for contemporary panoramic computer vision tasks. However,\nconventional imaging systems come with sophisticated lens designs and heavy\noptical components. This disqualifies their usage in many mobile and wearable\napplications where thin and portable, minimalist imaging systems are desired.\nIn this paper, we propose a Panoramic Computational Imaging Engine (PCIE) to\nachieve minimalist and high-quality panoramic imaging. With less than three\nspherical lenses, a Minimalist Panoramic Imaging Prototype (MPIP) is\nconstructed based on the design of the Panoramic Annular Lens (PAL), but with\nlow-quality imaging results due to aberrations and small image plane size. We\npropose two pipelines, i.e. Aberration Correction (AC) and Super-Resolution and\nAberration Correction (SR&AC), to solve the image quality problems of MPIP,\nwith imaging sensors of small and large pixel size, respectively. To leverage\nthe prior information of the optical system, we propose a Point Spread Function\n(PSF) representation method to produce a PSF map as an additional modality. A\nPSF-aware Aberration-image Recovery Transformer (PART) is designed as a\nuniversal network for the two pipelines, in which the self-attention\ncalculation and feature extraction are guided by the PSF map. We train PART on\nsynthetic image pairs from simulation and put forward the PALHQ dataset to fill\nthe gap of real-world high-quality PAL images for low-level vision. A\ncomprehensive variety of experiments on synthetic and real-world benchmarks\ndemonstrates the impressive imaging results of PCIE and the effectiveness of\nthe PSF representation. We further deliver heuristic experimental findings for\nminimalist and high-quality panoramic imaging. Our dataset and code will be\navailable at https://github.com/zju-jiangqi/PCIE-PART.\n", "date": "2023-06-22", "categories": []}
{"id": "2407.04024", "title": "Adaptive Step-size Perception Unfolding Network with Non-local Hybrid\n  Attention for Hyperspectral Image Reconstruction", "abstract": "  Deep unfolding methods and transformer architecture have recently shown\npromising results in hyperspectral image (HSI) reconstruction. However, there\nstill exist two issues: (1) in the data subproblem, most methods represents the\nstepsize utilizing a learnable parameter. Nevertheless, for different spectral\nchannel, error between features and ground truth is unequal. (2) Transformer\nstruggles to balance receptive field size with pixel-wise detail information.\nTo overcome the aforementioned drawbacks, We proposed an adaptive step-size\nperception unfolding network (ASPUN), a deep unfolding network based on FISTA\nalgorithm, which uses an adaptive step-size perception module to estimate the\nupdate step-size of each spectral channel. In addition, we design a Non-local\nHybrid Attention Transformer(NHAT) module for fully leveraging the receptive\nfield advantage of transformer. By plugging the NLHA into the Non-local\nInformation Aggregation (NLIA) module, the unfolding network can achieve better\nreconstruction results. Experimental results show that our ASPUN is superior to\nthe existing SOTA algorithms and achieves the best performance.\n", "date": "2024-07-04", "categories": []}
{"id": "2407.04162", "title": "Measurement Embedded Schr\\\"odinger Bridge for Inverse Problems", "abstract": "  Score-based diffusion models are frequently employed as structural priors in\ninverse problems. However, their iterative denoising process, initiated from\nGaussian noise, often results in slow inference speeds. The Image-to-Image\nSchr\\\"odinger Bridge (I$^2$SB), which begins with the corrupted image, presents\na promising alternative as a prior for addressing inverse problems. In this\nwork, we introduce the Measurement Embedded Schr\\\"odinger Bridge (MESB). MESB\nestablishes Schr\\\"odinger Bridges between the distribution of corrupted images\nand the distribution of clean images given observed measurements. Based on\noptimal transport theory, we derive the forward and backward processes of MESB.\nThrough validation on diverse inverse problems, our proposed approach exhibits\nsuperior performance compared to existing Schr\\\"odinger Bridge-based inverse\nproblems solvers in both visual quality and quantitative metrics.\n", "date": "2024-05-23", "categories": []}
{"id": "2407.02683", "title": "Generalized Event Cameras", "abstract": "  Event cameras capture the world at high time resolution and with minimal\nbandwidth requirements. However, event streams, which only encode changes in\nbrightness, do not contain sufficient scene information to support a wide\nvariety of downstream tasks. In this work, we design generalized event cameras\nthat inherently preserve scene intensity in a bandwidth-efficient manner. We\ngeneralize event cameras in terms of when an event is generated and what\ninformation is transmitted. To implement our designs, we turn to single-photon\nsensors that provide digital access to individual photon detections; this\nmodality gives us the flexibility to realize a rich space of generalized event\ncameras. Our single-photon event cameras are capable of high-speed,\nhigh-fidelity imaging at low readout rates. Consequently, these event cameras\ncan support plug-and-play downstream inference, without capturing new event\ndatasets or designing specialized event-vision models. As a practical\nimplication, our designs, which involve lightweight and near-sensor-compatible\ncomputations, provide a way to use single-photon sensors without exorbitant\nbandwidth costs.\n", "date": "2024-07-02", "categories": []}
{"id": "2407.02744", "title": "Highly Accelerated MRI via Implicit Neural Representation Guided\n  Posterior Sampling of Diffusion Models", "abstract": "  Reconstructing high-fidelity magnetic resonance (MR) images from\nunder-sampled k-space is a commonly used strategy to reduce scan time. The\nposterior sampling of diffusion models based on the real measurement data holds\nsignificant promise of improved reconstruction accuracy. However, traditional\nposterior sampling methods often lack effective data consistency guidance,\nleading to inaccurate and unstable reconstructions. Implicit neural\nrepresentation (INR) has emerged as a powerful paradigm for solving inverse\nproblems by modeling a signal's attributes as a continuous function of spatial\ncoordinates. In this study, we present a novel posterior sampler for diffusion\nmodels using INR, named DiffINR. The INR-based component incorporates both the\ndiffusion prior distribution and the MRI physical model to ensure high data\nfidelity. DiffINR demonstrates superior performance on experimental datasets\nwith remarkable accuracy, even under high acceleration factors (up to R=12 in\nsingle-channel reconstruction). Notably, our proposed framework can be a\ngeneralizable framework to solve inverse problems in other medical imaging\ntasks.\n", "date": "2024-07-03", "categories": []}
{"id": "2407.03308", "title": "Accelerated Proton Resonance Frequency-based Magnetic Resonance\n  Thermometry by Optimized Deep Learning Method", "abstract": "  Proton resonance frequency (PRF) based MR thermometry is essential for\nfocused ultrasound (FUS) thermal ablation therapies. This work aims to enhance\ntemporal resolution in dynamic MR temperature map reconstruction using an\nimproved deep learning method. The training-optimized methods and five\nclassical neural networks were applied on the 2-fold and 4-fold under-sampling\nk-space data to reconstruct the temperature maps. The enhanced training modules\nincluded offline/online data augmentations, knowledge distillation, and the\namplitude-phase decoupling loss function. The heating experiments were\nperformed by a FUS transducer on phantom and ex vivo tissues, respectively.\nThese data were manually under-sampled to imitate acceleration procedures and\ntrained in our method to get the reconstruction model. The additional dozen or\nso testing datasets were separately obtained for evaluating the real-time\nperformance and temperature accuracy. Acceleration factors of 1.9 and 3.7 were\nfound for 2 times and 4 times k-space under-sampling strategies and the\nResUNet-based deep learning reconstruction performed exceptionally well. In\n2-fold acceleration scenario, the RMSE of temperature map patches provided the\nvalues of 0.888 degree centigrade and 1.145 degree centigrade on phantom and ex\nvivo testing datasets. The DICE value of temperature areas enclosed by 43\ndegree centigrade isotherm was 0.809, and the Bland-Altman analysis showed a\nbias of -0.253 degree centigrade with the apart of plus or minus 2.16 degree\ncentigrade. In 4 times under-sampling case, these evaluating values decreased\nby approximately 10%. This study demonstrates that deep learning-based\nreconstruction can significantly enhance the accuracy and efficiency of MR\nthermometry for clinical FUS thermal therapies.\n", "date": "2024-07-03", "categories": []}
{"id": "2209.01763", "title": "Uformer-ICS: A U-Shaped Transformer for Image Compressive Sensing\n  Service", "abstract": "  Many service computing applications require real-time dataset collection from\nmultiple devices, necessitating efficient sampling techniques to reduce\nbandwidth and storage pressure. Compressive sensing (CS) has found wide-ranging\napplications in image acquisition and reconstruction. Recently, numerous\ndeep-learning methods have been introduced for CS tasks. However, the accurate\nreconstruction of images from measurements remains a significant challenge,\nespecially at low sampling rates. In this paper, we propose Uformer-ICS as a\nnovel U-shaped transformer for image CS tasks by introducing inner\ncharacteristics of CS into transformer architecture. To utilize the uneven\nsparsity distribution of image blocks, we design an adaptive sampling\narchitecture that allocates measurement resources based on the estimated block\nsparsity, allowing the compressed results to retain maximum information from\nthe original image. Additionally, we introduce a multi-channel projection (MCP)\nmodule inspired by traditional CS optimization methods. By integrating the MCP\nmodule into the transformer blocks, we construct projection-based transformer\nblocks, and then form a symmetrical reconstruction model using these blocks and\nresidual convolutional blocks. Therefore, our reconstruction model can\nsimultaneously utilize the local features and long-range dependencies of image,\nand the prior projection knowledge of CS theory.\n  Experimental results demonstrate its significantly better reconstruction\nperformance than state-of-the-art deep learning-based CS methods.\n", "date": "2022-09-05", "categories": []}
{"id": "2407.01559", "title": "Data-driven approaches for electrical impedance tomography image\n  segmentation from partial boundary data", "abstract": "  Electrical impedance tomography (EIT) plays a crucial role in non-invasive\nimaging, with both medical and industrial applications. In this paper, we\npresent three data-driven reconstruction methods for EIT imaging. These three\napproaches were originally submitted to the Kuopio tomography challenge 2023\n(KTC2023). First, we introduce a post-processing approach, which achieved first\nplace at KTC2023. Further, we present a fully learned and a conditional\ndiffusion approach. All three methods are based on a similar neural network as\na backbone and were trained using a synthetically generated data set, providing\nwith an opportunity for a fair comparison of these different data-driven\nreconstruction methods.\n", "date": "2024-05-06", "categories": []}
{"id": "2407.00944", "title": "Diffusion Transformer Model With Compact Prior for Low-dose PET\n  Reconstruction", "abstract": "  Positron emission tomography (PET) is an advanced medical imaging technique\nthat plays a crucial role in non-invasive clinical diagnosis. However, while\nreducing radiation exposure through low-dose PET scans is beneficial for\npatient safety, it often results in insufficient statistical data. This\nscarcity of data poses significant challenges for accurately reconstructing\nhigh-quality images, which are essential for reliable diagnostic outcomes. In\nthis research, we propose a diffusion transformer model (DTM) guided by joint\ncompact prior (JCP) to enhance the reconstruction quality of low-dose PET\nimaging. In light of current research findings, we present a pioneering PET\nreconstruction model that integrates diffusion and transformer models for joint\noptimization. This model combines the powerful distribution mapping abilities\nof diffusion models with the capacity of transformers to capture long-range\ndependencies, offering significant advantages for low-dose PET reconstruction.\nAdditionally, the incorporation of the lesion refining block and penalized\nweighted least squares (PWLS) enhance the recovery capability of lesion regions\nand preserves detail information, solving blurring problems in lesion areas and\ntexture details of most deep learning frameworks. Experimental results\ndemonstrate the effectiveness of DTM in enhancing image quality and preserving\ncritical clinical information for low-dose PET scans. Our approach not only\nreduces radiation exposure risks but also provides a more reliable PET imaging\ntool for early disease detection and patient management.\n", "date": "2024-07-01", "categories": []}
{"id": "2407.00972", "title": "FALCON: Frequency Adjoint Link with CONtinuous Density Mask for Fast\n  Single Image Dehazing", "abstract": "  Image dehazing, addressing atmospheric interference like fog and haze,\nremains a pervasive challenge crucial for robust vision applications such as\nsurveillance and remote sensing under adverse visibility. While various\nmethodologies have evolved from early works predicting transmission matrix and\natmospheric light features to deep learning and dehazing networks, they\ninnately prioritize dehazing quality metrics, neglecting the need for real-time\napplicability in time-sensitive domains like autonomous driving. This work\nintroduces FALCON (Frequency Adjoint Link with CONtinuous density mask), a\nsingle-image dehazing system achieving state-of-the-art performance on both\nquality and speed. Particularly, we develop a novel bottleneck module, namely,\nFrequency Adjoint Link, operating in the frequency space to globally expand the\nreceptive field with minimal growth in network size. Further, we leverage the\nunderlying haze distribution based on the atmospheric scattering model via a\nContinuous Density Mask (CDM) which serves as a continuous-valued mask input\nprior and a differentiable auxiliary loss. Comprehensive experiments involving\nmultiple state-of-the-art methods and ablation analysis demonstrate FALCON's\nexceptional performance in both dehazing quality and speed (i.e., >$180\nframes-per-second), quantified by metrics such as FPS, PSNR, and SSIM.\n", "date": "2024-07-01", "categories": []}
{"id": "2407.01014", "title": "An Expectation-Maximization Algorithm for Training Clean Diffusion\n  Models from Corrupted Observations", "abstract": "  Diffusion models excel in solving imaging inverse problems due to their\nability to model complex image priors. However, their reliance on large, clean\ndatasets for training limits their practical use where clean data is scarce. In\nthis paper, we propose EMDiffusion, an expectation-maximization (EM) approach\nto train diffusion models from corrupted observations. Our method alternates\nbetween reconstructing clean images from corrupted data using a known diffusion\nmodel (E-step) and refining diffusion model weights based on these\nreconstructions (M-step). This iterative process leads the learned diffusion\nmodel to gradually converge to the true clean data distribution. We validate\nour method through extensive experiments on diverse computational imaging\ntasks, including random inpainting, denoising, and deblurring, achieving new\nstate-of-the-art performance.\n", "date": "2024-07-01", "categories": []}
{"id": "2304.10839", "title": "Cross-domain Denoising for Low-dose Multi-frame Spiral Computed\n  Tomography", "abstract": "  Computed tomography (CT) has been used worldwide as a non-invasive test to\nassist in diagnosis. However, the ionizing nature of X-ray exposure raises\nconcerns about potential health risks such as cancer. The desire for lower\nradiation doses has driven researchers to improve reconstruction quality.\nAlthough previous studies on low-dose computed tomography (LDCT) denoising have\ndemonstrated the effectiveness of learning-based methods, most were developed\non the simulated data. However, the real-world scenario differs significantly\nfrom the simulation domain, especially when using the multi-slice spiral\nscanner geometry. This paper proposes a two-stage method for the commercially\navailable multi-slice spiral CT scanners that better exploits the complete\nreconstruction pipeline for LDCT denoising across different domains. Our\napproach makes good use of the high redundancy of multi-slice projections and\nthe volumetric reconstructions while leveraging the over-smoothing problem in\nconventional cascaded frameworks caused by aggressive denoising. The dedicated\ndesign also provides a more explicit interpretation of the data flow. Extensive\nexperiments on various datasets showed that the proposed method could remove up\nto 70\\% of noise without compromised spatial resolution, and subjective\nevaluations by two experienced radiologists further supported its superior\nperformance against state-of-the-art methods in clinical practice.\n", "date": "2023-04-21", "categories": []}
{"id": "2406.19560", "title": "Cost-efficient Active Illumination Camera For Hyper-spectral\n  Reconstruction", "abstract": "  Hyper-spectral imaging has recently gained increasing attention for use in\ndifferent applications, including agricultural investigation, ground tracking,\nremote sensing and many other. However, the high cost, large physical size and\ncomplicated operation process stop hyperspectral cameras from being employed\nfor various applications and research fields. In this paper, we introduce a\ncost-efficient, compact and easy to use active illumination camera that may\nbenefit many applications. We developed a fully functional prototype of such\ncamera. With the hope of helping with agricultural research, we tested our\ncamera for plant root imaging. In addition, a U-Net model for spectral\nreconstruction was trained by using a reference hyperspectral camera's data as\nground truth and our camera's data as input. We demonstrated our camera's\nability to obtain additional information over a typical RGB camera. In\naddition, the ability to reconstruct hyperspectral data from multi-spectral\ninput makes our device compatible to models and algorithms developed for\nhyperspectral applications with no modifications required.\n", "date": "2024-06-27", "categories": []}
{"id": "2406.19666", "title": "CSAKD: Knowledge Distillation with Cross Self-Attention for\n  Hyperspectral and Multispectral Image Fusion", "abstract": "  Hyperspectral imaging, capturing detailed spectral information for each\npixel, is pivotal in diverse scientific and industrial applications. Yet, the\nacquisition of high-resolution (HR) hyperspectral images (HSIs) often needs to\nbe addressed due to the hardware limitations of existing imaging systems. A\nprevalent workaround involves capturing both a high-resolution multispectral\nimage (HR-MSI) and a low-resolution (LR) HSI, subsequently fusing them to yield\nthe desired HR-HSI. Although deep learning-based methods have shown promising\nin HR-MSI/LR-HSI fusion and LR-HSI super-resolution (SR), their substantial\nmodel complexities hinder deployment on resource-constrained imaging devices.\nThis paper introduces a novel knowledge distillation (KD) framework for\nHR-MSI/LR-HSI fusion to achieve SR of LR-HSI. Our KD framework integrates the\nproposed Cross-Layer Residual Aggregation (CLRA) block to enhance efficiency\nfor constructing Dual Two-Streamed (DTS) network structure, designed to extract\njoint and distinct features from LR-HSI and HR-MSI simultaneously. To fully\nexploit the spatial and spectral feature representations of LR-HSI and HR-MSI,\nwe propose a novel Cross Self-Attention (CSA) fusion module to adaptively fuse\nthose features to improve the spatial and spectral quality of the reconstructed\nHR-HSI. Finally, the proposed KD-based joint loss function is employed to\nco-train the teacher and student networks. Our experimental results demonstrate\nthat the student model not only achieves comparable or superior LR-HSI SR\nperformance but also significantly reduces the model-size and computational\nrequirements. This marks a substantial advancement over existing\nstate-of-the-art methods. The source code is available at\nhttps://github.com/ming053l/CSAKD.\n", "date": "2024-06-28", "categories": []}
{"id": "2406.18628", "title": "IDA-UIE: An Iterative Framework for Deep Network-based Degradation Aware\n  Underwater Image Enhancement", "abstract": "  Underwater image quality is affected by fluorescence, low illumination,\nabsorption, and scattering. Recent works in underwater image enhancement have\nproposed different deep network architectures to handle these problems. Most of\nthese works have proposed a single network to handle all the challenges. We\nbelieve that deep networks trained for specific conditions deliver better\nperformance than a single network learned from all degradation cases.\nAccordingly, the first contribution of this work lies in the proposal of an\niterative framework where a single dominant degradation condition is identified\nand resolved. This proposal considers the following eight degradation\nconditions -- low illumination, low contrast, haziness, blurred image, presence\nof noise and color imbalance in three different channels. A deep network is\ndesigned to identify the dominant degradation condition. Accordingly, an\nappropriate deep network is selected for degradation condition-specific\nenhancement. The second contribution of this work is the construction of\ndegradation condition specific datasets from good quality images of two\nstandard datasets (UIEB and EUVP). This dataset is used to learn the condition\nspecific enhancement networks. The proposed approach is found to outperform\nnine baseline methods on UIEB and EUVP datasets.\n", "date": "2024-06-26", "categories": []}
{"id": "2406.18927", "title": "RoFIR: Robust Fisheye Image Rectification Framework Impervious to\n  Optical Center Deviation", "abstract": "  Fisheye images are categorized fisheye into central and deviated based on the\noptical center position. Existing rectification methods are limited to central\nfisheye images, while this paper proposes a novel method that extends to\ndeviated fisheye image rectification. The challenge lies in the variant global\ndistortion distribution pattern caused by the random optical center position.\nTo address this challenge, we propose a distortion vector map (DVM) that\nmeasures the degree and direction of local distortion. By learning the DVM, the\nmodel can independently identify local distortions at each pixel without\nrelying on global distortion patterns. The model adopts a pre-training and\nfine-tuning training paradigm. In the pre-training stage, it predicts the\ndistortion vector map and perceives the local distortion features of each\npixel. In the fine-tuning stage, it predicts a pixel-wise flow map for deviated\nfisheye image rectification. We also propose a data augmentation method mixing\ncentral, deviated, and distorted-free images. Such data augmentation promotes\nthe model performance in rectifying both central and deviated fisheye images,\ncompared with models trained on single-type fisheye images. Extensive\nexperiments demonstrate the effectiveness and superiority of the proposed\nmethod.\n", "date": "2024-06-27", "categories": []}
{"id": "2406.19126", "title": "Super-resolution imaging using super-oscillatory diffractive neural\n  networks", "abstract": "  Optical super-oscillation enables far-field super-resolution imaging beyond\ndiffraction limits. However, the existing super-oscillatory lens for the\nspatial super-resolution imaging system still confronts critical limitations in\nperformance due to the lack of a more advanced design method and the limited\ndesign degree of freedom. Here, we propose an optical super-oscillatory\ndiffractive neural network, i.e., SODNN, that can achieve super-resolved\nspatial resolution for imaging beyond the diffraction limit with superior\nperformance over existing methods. SODNN is constructed by utilizing\ndiffractive layers to implement optical interconnections and imaging samples or\nbiological sensors to implement nonlinearity, which modulates the incident\noptical field to create optical super-oscillation effects in 3D space and\ngenerate the super-resolved focal spots. By optimizing diffractive layers with\n3D optical field constraints under an incident wavelength size of $\\lambda$, we\nachieved a super-oscillatory spot with a full width at half maximum of\n0.407$\\lambda$ in the far field distance over 400$\\lambda$ without side-lobes\nover the field of view, having a long depth of field over 10$\\lambda$.\nFurthermore, the SODNN implements a multi-wavelength and multi-focus spot array\nthat effectively avoids chromatic aberrations. Our research work will inspire\nthe development of intelligent optical instruments to facilitate the\napplications of imaging, sensing, perception, etc.\n", "date": "2024-06-27", "categories": []}
{"id": "2406.17869", "title": "Burst Image Super-Resolution with Base Frame Selection", "abstract": "  Burst image super-resolution has been a topic of active research in recent\nyears due to its ability to obtain a high-resolution image by using\ncomplementary information between multiple frames in the burst. In this work,\nwe explore using burst shots with non-uniform exposures to confront real-world\npractical scenarios by introducing a new benchmark dataset, dubbed\nNon-uniformly Exposed Burst Image (NEBI), that includes the burst frames at\nvarying exposure times to obtain a broader range of irradiance and motion\ncharacteristics within a scene. As burst shots with non-uniform exposures\nexhibit varying levels of degradation, fusing information of the burst shots\ninto the first frame as a base frame may not result in optimal image quality.\nTo address this limitation, we propose a Frame Selection Network (FSN) for\nnon-uniform scenarios. This network seamlessly integrates into existing\nsuper-resolution methods in a plug-and-play manner with low computational\ncosts. The comparative analysis reveals the effectiveness of the nonuniform\nsetting for the practical scenario and our FSN on synthetic-/real- NEBI\ndatasets.\n", "date": "2024-06-25", "categories": []}
{"id": "2406.17970", "title": "Highly Constrained Coded Aperture Imaging Systems Design Via a Knowledge\n  Distillation Approach", "abstract": "  Computational optical imaging (COI) systems have enabled the acquisition of\nhigh-dimensional signals through optical coding elements (OCEs). OCEs encode\nthe high-dimensional signal in one or more snapshots, which are subsequently\ndecoded using computational algorithms. Currently, COI systems are optimized\nthrough an end-to-end (E2E) approach, where the OCEs are modeled as a layer of\na neural network and the remaining layers perform a specific imaging task.\nHowever, the performance of COI systems optimized through E2E is limited by the\nphysical constraints imposed by these systems. This paper proposes a knowledge\ndistillation (KD) framework for the design of highly physically constrained COI\nsystems. This approach employs the KD methodology, which consists of a\nteacher-student relationship, where a high-performance, unconstrained COI\nsystem (the teacher), guides the optimization of a physically constrained\nsystem (the student) characterized by a limited number of snapshots. We\nvalidate the proposed approach, using a binary coded apertures single pixel\ncamera for monochromatic and multispectral image reconstruction. Simulation\nresults demonstrate the superiority of the KD scheme over traditional E2E\noptimization for the designing of highly physically constrained COI systems.\n", "date": "2024-06-25", "categories": []}
{"id": "2406.18079", "title": "MFDNet: Multi-Frequency Deflare Network for Efficient Nighttime Flare\n  Removal", "abstract": "  When light is scattered or reflected accidentally in the lens, flare\nartifacts may appear in the captured photos, affecting the photos' visual\nquality. The main challenge in flare removal is to eliminate various flare\nartifacts while preserving the original content of the image. To address this\nchallenge, we propose a lightweight Multi-Frequency Deflare Network (MFDNet)\nbased on the Laplacian Pyramid. Our network decomposes the flare-corrupted\nimage into low and high-frequency bands, effectively separating the\nillumination and content information in the image. The low-frequency part\ntypically contains illumination information, while the high-frequency part\ncontains detailed content information. So our MFDNet consists of two main\nmodules: the Low-Frequency Flare Perception Module (LFFPM) to remove flare in\nthe low-frequency part and the Hierarchical Fusion Reconstruction Module (HFRM)\nto reconstruct the flare-free image. Specifically, to perceive flare from a\nglobal perspective while retaining detailed information for image restoration,\nLFFPM utilizes Transformer to extract global information while utilizing a\nconvolutional neural network to capture detailed local features. Then HFRM\ngradually fuses the outputs of LFFPM with the high-frequency component of the\nimage through feature aggregation. Moreover, our MFDNet can reduce the\ncomputational cost by processing in multiple frequency bands instead of\ndirectly removing the flare on the input image. Experimental results\ndemonstrate that our approach outperforms state-of-the-art methods in removing\nnighttime flare on real-world and synthetic images from the Flare7K dataset.\nFurthermore, the computational complexity of our model is remarkably low.\n", "date": "2024-06-26", "categories": []}
{"id": "2406.10212", "title": "NeST: Neural Stress Tensor Tomography by leveraging 3D Photoelasticity", "abstract": "  Photoelasticity enables full-field stress analysis in transparent objects\nthrough stress-induced birefringence. Existing techniques are limited to 2D\nslices and require destructively slicing the object. Recovering the internal 3D\nstress distribution of the entire object is challenging as it involves solving\na tensor tomography problem and handling phase wrapping ambiguities. We\nintroduce NeST, an analysis-by-synthesis approach for reconstructing 3D stress\ntensor fields as neural implicit representations from polarization\nmeasurements. Our key insight is to jointly handle phase unwrapping and tensor\ntomography using a differentiable forward model based on Jones calculus. Our\nnon-linear model faithfully matches real captures, unlike prior linear\napproximations. We develop an experimental multi-axis polariscope setup to\ncapture 3D photoelasticity and experimentally demonstrate that NeST\nreconstructs the internal stress distribution for objects with varying shape\nand force conditions. Additionally, we showcase novel applications in stress\nanalysis, such as visualizing photoelastic fringes by virtually slicing the\nobject and viewing photoelastic fringes from unseen viewpoints. NeST paves the\nway for scalable non-destructive 3D photoelastic analysis.\n", "date": "2024-06-14", "categories": []}
{"id": "2406.15716", "title": "Predicting fluorescent labels in label-free microscopy images with\n  pix2pix and adaptive loss in Light My Cells challenge", "abstract": "  Fluorescence labeling is the standard approach to reveal cellular structures\nand other subcellular constituents for microscopy images. However, this\ninvasive procedure may perturb or even kill the cells and the procedure itself\nis highly time-consuming and complex. Recently, in silico labeling has emerged\nas a promising alternative, aiming to use machine learning models to directly\npredict the fluorescently labeled images from label-free microscopy. In this\npaper, we propose a deep learning-based in silico labeling method for the Light\nMy Cells challenge. Built upon pix2pix, our proposed method can be trained\nusing the partially labeled datasets with an adaptive loss. Moreover, we\nexplore the effectiveness of several training strategies to handle different\ninput modalities, such as training them together or separately. The results\nshow that our method achieves promising performance for in silico labeling. Our\ncode is available at https://github.com/MedICL-VU/LightMyCells.\n", "date": "2024-06-22", "categories": []}
{"id": "2406.16083", "title": "Mamba-based Light Field Super-Resolution with Efficient Subspace\n  Scanning", "abstract": "  Transformer-based methods have demonstrated impressive performance in 4D\nlight field (LF) super-resolution by effectively modeling long-range\nspatial-angular correlations, but their quadratic complexity hinders the\nefficient processing of high resolution 4D inputs, resulting in slow inference\nspeed and high memory cost. As a compromise, most prior work adopts a\npatch-based strategy, which fails to leverage the full information from the\nentire input LFs. The recently proposed selective state-space model, Mamba, has\ngained popularity for its efficient long-range sequence modeling. In this\npaper, we propose a Mamba-based Light Field Super-Resolution method, named\nMLFSR, by designing an efficient subspace scanning strategy. Specifically, we\ntokenize 4D LFs into subspace sequences and conduct bi-directional scanning on\neach subspace. Based on our scanning strategy, we then design the Mamba-based\nGlobal Interaction (MGI) module to capture global information and the local\nSpatial- Angular Modulator (SAM) to complement local details. Additionally, we\nintroduce a Transformer-to-Mamba (T2M) loss to further enhance overall\nperformance. Extensive experiments on public benchmarks demonstrate that MLFSR\nsurpasses CNN-based models and rivals Transformer-based methods in performance\nwhile maintaining higher efficiency. With quicker inference speed and reduced\nmemory demand, MLFSR facilitates full-image processing of high-resolution 4D\nLFs with enhanced performance.\n", "date": "2024-06-23", "categories": []}
{"id": "2406.14976", "title": "CoCPF: Coordinate-based Continuous Projection Field for Ill-Posed\n  Inverse Problem in Imaging", "abstract": "  Sparse-view computed tomography (SVCT) reconstruction aims to acquire CT\nimages based on sparsely-sampled measurements. It allows the subjects exposed\nto less ionizing radiation, reducing the lifetime risk of developing cancers.\nRecent researches employ implicit neural representation (INR) techniques to\nreconstruct CT images from a single SV sinogram. However, due to ill-posedness,\nthese INR-based methods may leave considerable ``holes'' (i.e., unmodeled\nspaces) in their fields, leading to sub-optimal results. In this paper, we\npropose the Coordinate-based Continuous Projection Field (CoCPF), which aims to\nbuild hole-free representation fields for SVCT reconstruction, achieving better\nreconstruction quality. Specifically, to fill the holes, CoCPF first employs\nthe stripe-based volume sampling module to broaden the sampling regions of\nRadon transformation from rays (1D space) to stripes (2D space), which can well\ncover the internal regions between SV projections. Then, by feeding the\nsampling regions into the proposed differentiable rendering modules, the holes\ncan be jointly optimized during training, reducing the ill-posed levels. As a\nresult, CoCPF can accurately estimate the internal measurements between SV\nprojections (i.e., DV sinograms), producing high-quality CT images after\nre-projection. Extensive experiments on simulated and real projection datasets\ndemonstrate that CoCPF outperforms state-of-the-art methods for 2D and 3D SVCT\nreconstructions under various projection numbers and geometries, yielding\nfine-grained details and fewer artifacts. Our code will be publicly available.\n", "date": "2024-06-21", "categories": []}
{"id": "2210.09248", "title": "Provable Phase Retrieval with Mirror Descent", "abstract": "  In this paper, we consider the problem of phase retrieval, which consists of\nrecovering an $n$-dimensional real vector from the magnitude of its $m$ linear\nmeasurements. We propose a mirror descent (or Bregman gradient descent)\nalgorithm based on a wisely chosen Bregman divergence, hence allowing to remove\nthe classical global Lipschitz continuity requirement on the gradient of the\nnon-convex phase retrieval objective to be minimized. We apply the mirror\ndescent for two random measurements: the \\iid standard Gaussian and those\nobtained by multiple structured illuminations through Coded Diffraction\nPatterns (CDP). For the Gaussian case, we show that when the number of\nmeasurements $m$ is large enough, then with high probability, for almost all\ninitializers, the algorithm recovers the original vector up to a global sign\nchange. For both measurements, the mirror descent exhibits a local linear\nconvergence behaviour with a dimension-independent convergence rate. Our\ntheoretical results are finally illustrated with various numerical experiments,\nincluding an application to the reconstruction of images in precision optics.\n", "date": "2022-10-17", "categories": []}
{"id": "2403.00289", "title": "Optimization of array encoding for ultrasound imaging", "abstract": "  Objective: The transmit encoding model for synthetic aperture imaging is a\nrobust and flexible framework for understanding the effects of acoustic\ntransmission on ultrasound image reconstruction. Our objective is to use\nmachine learning (ML) to construct scanning sequences, parameterized by time\ndelays and apodization weights, that produce high-quality B-mode images.\nApproach: We use a custom ML model in PyTorch with simulated RF data from Field\nII to probe the space of possible encoding sequences for those that minimize a\nloss function that describes image quality. This approach is made\ncomputationally feasible by a novel formulation of the derivative for\ndelay-and-sum beamforming. Main Results: When trained for a specified\nexperimental setting (imaging domain, hardware restrictions, etc.), our ML\nmodel produces optimized encoding sequences that, when deployed in the REFoCUS\nimaging framework, improve a number of standard quality metrics over\nconventional sequences including resolution, field of view, and contrast. We\ndemonstrate these results experimentally on both wire targets and a\ntissue-mimicking phantom. Significance: This work demonstrates that the set of\ncommonly used encoding schemes represent only a narrow subset of those\navailable. Additionally, it demonstrates the value for ML tasks in synthetic\ntransmit aperture imaging to consider the beamformer within the model, instead\nof purely as a post-processing step.\n", "date": "2024-03-01", "categories": []}
{"id": "2405.10754", "title": "Stable Phase Retrieval with Mirror Descent", "abstract": "  In this paper, we aim to reconstruct an n-dimensional real vector from m\nphaseless measurements corrupted by an additive noise. We extend the noiseless\nframework developed in [15], based on mirror descent (or Bregman gradient\ndescent), to deal with noisy measurements and prove that the procedure is\nstable to (small enough) additive noise. In the deterministic case, we show\nthat mirror descent converges to a critical point of the phase retrieval\nproblem, and if the algorithm is well initialized and the noise is small\nenough, the critical point is near the true vector up to a global sign change.\nWhen the measurements are i.i.d Gaussian and the signal-to-noise ratio is large\nenough, we provide global convergence guarantees that ensure that with high\nprobability, mirror descent converges to a global minimizer near the true\nvector (up to a global sign change), as soon as the number of measurements m is\nlarge enough. The sample complexity bound can be improved if a spectral method\nis used to provide a good initial guess. We complement our theoretical study\nwith several numerical results showing that mirror descent is both a\ncomputationally and statistically efficient scheme to solve the phase retrieval\nproblem.\n", "date": "2024-05-17", "categories": []}
{"id": "2406.12463", "title": "LFMamba: Light Field Image Super-Resolution with State Space Model", "abstract": "  Recent years have witnessed significant advancements in light field image\nsuper-resolution (LFSR) owing to the progress of modern neural networks.\nHowever, these methods often face challenges in capturing long-range\ndependencies (CNN-based) or encounter quadratic computational complexities\n(Transformer-based), which limit their performance. Recently, the State Space\nModel (SSM) with selective scanning mechanism (S6), exemplified by Mamba, has\nemerged as a superior alternative in various vision tasks compared to\ntraditional CNN- and Transformer-based approaches, benefiting from its\neffective long-range sequence modeling capability and linear-time complexity.\nTherefore, integrating S6 into LFSR becomes compelling, especially considering\nthe vast data volume of 4D light fields. However, the primary challenge lies in\n\\emph{designing an appropriate scanning method for 4D light fields that\neffectively models light field features}. To tackle this, we employ SSMs on the\ninformative 2D slices of 4D LFs to fully explore spatial contextual\ninformation, complementary angular information, and structure information. To\nachieve this, we carefully devise a basic SSM block characterized by an\nefficient SS2D mechanism that facilitates more effective and efficient feature\nlearning on these 2D slices. Based on the above two designs, we further\nintroduce an SSM-based network for LFSR termed LFMamba. Experimental results on\nLF benchmarks demonstrate the superior performance of LFMamba. Furthermore,\nextensive ablation studies are conducted to validate the efficacy and\ngeneralization ability of our proposed method. We expect that our LFMamba shed\nlight on effective representation learning of LFs with state space models.\n", "date": "2024-06-18", "categories": []}
{"id": "2402.16683", "title": "Re-Envisioning Numerical Information Field Theory (NIFTy.re): A Library\n  for Gaussian Processes and Variational Inference", "abstract": "  Imaging is the process of transforming noisy, incomplete data into a space\nthat humans can interpret. NIFTy is a Bayesian framework for imaging and has\nalready successfully been applied to many fields in astrophysics. Previous\ndesign decisions held the performance and the development of methods in NIFTy\nback. We present a rewrite of NIFTy, coined NIFTy.re, which reworks the\nmodeling principle, extends the inference strategies, and outsources much of\nthe heavy lifting to JAX. The rewrite dramatically accelerates models written\nin NIFTy, lays the foundation for new types of inference machineries, improves\nmaintainability, and enables interoperability between NIFTy and the JAX machine\nlearning ecosystem.\n", "date": "2024-02-26", "categories": []}
{"id": "2406.11711", "title": "OGNI-DC: Robust Depth Completion with Optimization-Guided Neural\n  Iterations", "abstract": "  Depth completion is the task of generating a dense depth map given an image\nand a sparse depth map as inputs. It has important applications in various\ndownstream tasks. In this paper, we present OGNI-DC, a novel framework for\ndepth completion. The key to our method is \"Optimization-Guided Neural\nIterations\" (OGNI). It consists of a recurrent unit that refines a depth\ngradient field and a differentiable depth integrator that integrates the depth\ngradients into a depth map. OGNI-DC exhibits strong generalization,\noutperforming baselines by a large margin on unseen datasets and across various\nsparsity levels. Moreover, OGNI-DC has high accuracy, achieving\nstate-of-the-art performance on the NYUv2 and the KITTI benchmarks. Code is\navailable at https://github.com/princeton-vl/OGNI-DC.\n", "date": "2024-06-17", "categories": []}
{"id": "2406.09768", "title": "Bayesian Conditioned Diffusion Models for Inverse Problems", "abstract": "  Diffusion models have recently been shown to excel in many image\nreconstruction tasks that involve inverse problems based on a forward\nmeasurement operator. A common framework uses task-agnostic unconditional\nmodels that are later post-conditioned for reconstruction, an approach that\ntypically suffers from suboptimal task performance. While task-specific\nconditional models have also been proposed, current methods heuristically\ninject measured data as a naive input channel that elicits sampling\ninaccuracies. Here, we address the optimal conditioning of diffusion models for\nsolving challenging inverse problems that arise during image reconstruction.\nSpecifically, we propose a novel Bayesian conditioning technique for diffusion\nmodels, BCDM, based on score-functions associated with the conditional\ndistribution of desired images given measured data. We rigorously derive the\ntheory to express and train the conditional score-function. Finally, we show\nstate-of-the-art performance in image dealiasing, deblurring, super-resolution,\nand inpainting with the proposed technique.\n", "date": "2024-06-14", "categories": []}
{"id": "2401.08923", "title": "Subwavelength Imaging using a Solid-Immersion Diffractive Optical\n  Processor", "abstract": "  Phase imaging is widely used in biomedical imaging, sensing, and material\ncharacterization, among other fields. However, direct imaging of phase objects\nwith subwavelength resolution remains a challenge. Here, we demonstrate\nsubwavelength imaging of phase and amplitude objects based on all-optical\ndiffractive encoding and decoding. To resolve subwavelength features of an\nobject, the diffractive imager uses a thin, high-index solid-immersion layer to\ntransmit high-frequency information of the object to a spatially-optimized\ndiffractive encoder, which converts/encodes high-frequency information of the\ninput into low-frequency spatial modes for transmission through air. The\nsubsequent diffractive decoder layers (in air) are jointly designed with the\nencoder using deep-learning-based optimization, and communicate with the\nencoder layer to create magnified images of input objects at its output,\nrevealing subwavelength features that would otherwise be washed away due to\ndiffraction limit. We demonstrate that this all-optical collaboration between a\ndiffractive solid-immersion encoder and the following decoder layers in air can\nresolve subwavelength phase and amplitude features of input objects in a highly\ncompact design. To experimentally demonstrate its proof-of-concept, we used\nterahertz radiation and developed a fabrication method for creating monolithic\nmulti-layer diffractive processors. Through these monolithically fabricated\ndiffractive encoder-decoder pairs, we demonstrated phase-to-intensity\ntransformations and all-optically reconstructed subwavelength phase features of\ninput objects by directly transforming them into magnified intensity features\nat the output. This solid-immersion-based diffractive imager, with its compact\nand cost-effective design, can find wide-ranging applications in bioimaging,\nendoscopy, sensing and materials characterization.\n", "date": "2024-01-17", "categories": []}
{"id": "2406.09135", "title": "AdaRevD: Adaptive Patch Exiting Reversible Decoder Pushes the Limit of\n  Image Deblurring", "abstract": "  Despite the recent progress in enhancing the efficacy of image deblurring,\nthe limited decoding capability constrains the upper limit of State-Of-The-Art\n(SOTA) methods. This paper proposes a pioneering work, Adaptive Patch Exiting\nReversible Decoder (AdaRevD), to explore their insufficient decoding\ncapability. By inheriting the weights of the well-trained encoder, we refactor\na reversible decoder which scales up the single-decoder training to\nmulti-decoder training while remaining GPU memory-friendly. Meanwhile, we show\nthat our reversible structure gradually disentangles high-level degradation\ndegree and low-level blur pattern (residual of the blur image and its sharp\ncounterpart) from compact degradation representation. Besides, due to the\nspatially-variant motion blur kernels, different blur patches have various\ndeblurring difficulties. We further introduce a classifier to learn the\ndegradation degree of image patches, enabling them to exit at different\nsub-decoders for speedup. Experiments show that our AdaRevD pushes the limit of\nimage deblurring, e.g., achieving 34.60 dB in PSNR on GoPro dataset.\n", "date": "2024-06-13", "categories": []}
{"id": "2406.09409", "title": "CodedEvents: Optimal Point-Spread-Function Engineering for 3D-Tracking\n  with Event Cameras", "abstract": "  Point-spread-function (PSF) engineering is a well-established computational\nimaging technique that uses phase masks and other optical elements to embed\nextra information (e.g., depth) into the images captured by conventional CMOS\nimage sensors. To date, however, PSF-engineering has not been applied to\nneuromorphic event cameras; a powerful new image sensing technology that\nresponds to changes in the log-intensity of light.\n  This paper establishes theoretical limits (Cram\\'er Rao bounds) on 3D point\nlocalization and tracking with PSF-engineered event cameras. Using these\nbounds, we first demonstrate that existing Fisher phase masks are already\nnear-optimal for localizing static flashing point sources (e.g., blinking\nfluorescent molecules). We then demonstrate that existing designs are\nsub-optimal for tracking moving point sources and proceed to use our theory to\ndesign optimal phase masks and binary amplitude masks for this task. To\novercome the non-convexity of the design problem, we leverage novel implicit\nneural representation based parameterizations of the phase and amplitude masks.\nWe demonstrate the efficacy of our designs through extensive simulations. We\nalso validate our method with a simple prototype.\n", "date": "2024-06-13", "categories": []}
{"id": "2401.07856", "title": "Information hiding cameras: optical concealment of object information\n  into ordinary images", "abstract": "  Data protection methods like cryptography, despite being effective,\ninadvertently signal the presence of secret communication, thereby drawing\nundue attention. Here, we introduce an optical information hiding camera\nintegrated with an electronic decoder, optimized jointly through deep learning.\nThis information hiding-decoding system employs a diffractive optical processor\nas its front-end, which transforms and hides input images in the form of\nordinary-looking patterns that deceive/mislead human observers. This\ninformation hiding transformation is valid for infinitely many combinations of\nsecret messages, all of which are transformed into ordinary-looking output\npatterns, achieved all-optically through passive light-matter interactions\nwithin the optical processor. By processing these ordinary-looking output\nimages, a jointly-trained electronic decoder neural network accurately\nreconstructs the original information hidden within the deceptive output\npattern. We numerically demonstrated our approach by designing an information\nhiding diffractive camera along with a jointly-optimized convolutional decoder\nneural network. The efficacy of this system was demonstrated under various\nlighting conditions and noise levels, showing its robustness. We further\nextended this information hiding camera to multi-spectral operation, allowing\nthe concealment and decoding of multiple images at different wavelengths, all\nperformed simultaneously in a single feed-forward operation. The feasibility of\nour framework was also demonstrated experimentally using THz radiation. This\noptical encoder-electronic decoder-based co-design provides a novel information\nhiding camera interface that is both high-speed and energy-efficient, offering\nan intriguing solution for visual information security.\n", "date": "2024-01-15", "categories": []}
{"id": "2406.07951", "title": "DemosaicFormer: Coarse-to-Fine Demosaicing Network for HybridEVS Camera", "abstract": "  Hybrid Event-Based Vision Sensor (HybridEVS) is a novel sensor integrating\ntraditional frame-based and event-based sensors, offering substantial benefits\nfor applications requiring low-light, high dynamic range, and low-latency\nenvironments, such as smartphones and wearable devices. Despite its potential,\nthe lack of Image signal processing (ISP) pipeline specifically designed for\nHybridEVS poses a significant challenge. To address this challenge, in this\nstudy, we propose a coarse-to-fine framework named DemosaicFormer which\ncomprises coarse demosaicing and pixel correction. Coarse demosaicing network\nis designed to produce a preliminary high-quality estimate of the RGB image\nfrom the HybridEVS raw data while the pixel correction network enhances the\nperformance of image restoration and mitigates the impact of defective pixels.\nOur key innovation is the design of a Multi-Scale Gating Module (MSGM) applying\nthe integration of cross-scale features, which allows feature information to\nflow between different scales. Additionally, the adoption of progressive\ntraining and data augmentation strategies further improves model's robustness\nand effectiveness. Experimental results show superior performance against the\nexisting methods both qualitatively and visually, and our DemosaicFormer\nachieves the best performance in terms of all the evaluation metrics in the\nMIPI 2024 challenge on Demosaic for Hybridevs Camera. The code is available at\nhttps://github.com/QUEAHREN/DemosaicFormer.\n", "date": "2024-06-12", "categories": []}
{"id": "2402.01915", "title": "Robust Inverse Graphics via Probabilistic Inference", "abstract": "  How do we infer a 3D scene from a single image in the presence of corruptions\nlike rain, snow or fog? Straightforward domain randomization relies on knowing\nthe family of corruptions ahead of time. Here, we propose a Bayesian\napproach-dubbed robust inverse graphics (RIG)-that relies on a strong scene\nprior and an uninformative uniform corruption prior, making it applicable to a\nwide range of corruptions. Given a single image, RIG performs posterior\ninference jointly over the scene and the corruption. We demonstrate this idea\nby training a neural radiance field (NeRF) scene prior and using a secondary\nNeRF to represent the corruptions over which we place an uninformative prior.\nRIG, trained only on clean data, outperforms depth estimators and alternative\nNeRF approaches that perform point estimation instead of full inference. The\nresults hold for a number of scene prior architectures based on normalizing\nflows and diffusion models. For the latter, we develop reconstruction-guidance\nwith auxiliary latents (ReGAL)-a diffusion conditioning algorithm that is\napplicable in the presence of auxiliary latent variables such as the\ncorruption. RIG demonstrates how scene priors can be used beyond generation\ntasks.\n", "date": "2024-02-02", "categories": []}
{"id": "2406.07006", "title": "MIPI 2024 Challenge on Few-shot RAW Image Denoising: Methods and Results", "abstract": "  The increasing demand for computational photography and imaging on mobile\nplatforms has led to the widespread development and integration of advanced\nimage sensors with novel algorithms in camera systems. However, the scarcity of\nhigh-quality data for research and the rare opportunity for in-depth exchange\nof views from industry and academia constrain the development of mobile\nintelligent photography and imaging (MIPI). Building on the achievements of the\nprevious MIPI Workshops held at ECCV 2022 and CVPR 2023, we introduce our third\nMIPI challenge including three tracks focusing on novel image sensors and\nimaging algorithms. In this paper, we summarize and review the Few-shot RAW\nImage Denoising track on MIPI 2024. In total, 165 participants were\nsuccessfully registered, and 7 teams submitted results in the final testing\nphase. The developed solutions in this challenge achieved state-of-the-art\nerformance on Few-shot RAW Image Denoising. More details of this challenge and\nthe link to the dataset can be found at https://mipichallenge.org/MIPI2024.\n", "date": "2024-06-11", "categories": []}
{"id": "2406.07551", "title": "Blur-aware Spatio-temporal Sparse Transformer for Video Deblurring", "abstract": "  Video deblurring relies on leveraging information from other frames in the\nvideo sequence to restore the blurred regions in the current frame. Mainstream\napproaches employ bidirectional feature propagation, spatio-temporal\ntransformers, or a combination of both to extract information from the video\nsequence. However, limitations in memory and computational resources\nconstraints the temporal window length of the spatio-temporal transformer,\npreventing the extraction of longer temporal contextual information from the\nvideo sequence. Additionally, bidirectional feature propagation is highly\nsensitive to inaccurate optical flow in blurry frames, leading to error\naccumulation during the propagation process. To address these issues, we\npropose \\textbf{BSSTNet}, \\textbf{B}lur-aware \\textbf{S}patio-temporal\n\\textbf{S}parse \\textbf{T}ransformer Network. It introduces the blur map, which\nconverts the originally dense attention into a sparse form, enabling a more\nextensive utilization of information throughout the entire video sequence.\nSpecifically, BSSTNet (1) uses a longer temporal window in the transformer,\nleveraging information from more distant frames to restore the blurry pixels in\nthe current frame. (2) introduces bidirectional feature propagation guided by\nblur maps, which reduces error accumulation caused by the blur frame. The\nexperimental results demonstrate the proposed BSSTNet outperforms the\nstate-of-the-art methods on the GoPro and DVD datasets.\n", "date": "2024-06-11", "categories": []}
{"id": "2311.09253", "title": "The Perception-Robustness Tradeoff in Deterministic Image Restoration", "abstract": "  We study the behavior of deterministic methods for solving inverse problems\nin imaging. These methods are commonly designed to achieve two goals: (1)\nattaining high perceptual quality, and (2) generating reconstructions that are\nconsistent with the measurements. We provide a rigorous proof that the better a\npredictor satisfies these two requirements, the larger its Lipschitz constant\nmust be, regardless of the nature of the degradation involved. In particular,\nto approach perfect perceptual quality and perfect consistency, the Lipschitz\nconstant of the model must grow to infinity. This implies that such methods are\nnecessarily more susceptible to adversarial attacks. We demonstrate our theory\non single image super-resolution algorithms, addressing both noisy and\nnoiseless settings. We also show how this undesired behavior can be leveraged\nto explore the posterior distribution, thereby allowing the deterministic model\nto imitate stochastic methods.\n", "date": "2023-11-14", "categories": []}
{"id": "2312.13319", "title": "In2SET: Intra-Inter Similarity Exploiting Transformer for Dual-Camera\n  Compressive Hyperspectral Imaging", "abstract": "  Dual-Camera Compressed Hyperspectral Imaging (DCCHI) offers the capability to\nreconstruct 3D Hyperspectral Image (HSI) by fusing compressive and Panchromatic\n(PAN) image, which has shown great potential for snapshot hyperspectral imaging\nin practice. In this paper, we introduce a novel DCCHI reconstruction network,\nthe Intra-Inter Similarity Exploiting Transformer (In2SET). Our key insight is\nto make full use of the PAN image to assist the reconstruction. To this end, we\npropose using the intra-similarity within the PAN image as a proxy for\napproximating the intra-similarity in the original HSI, thereby offering an\nenhanced content prior for more accurate HSI reconstruction. Furthermore, we\naim to align the features from the underlying HSI with those of the PAN image,\nmaintaining semantic consistency and introducing new contextual information for\nthe reconstruction process. By integrating In2SET into a PAN-guided unrolling\nframework, our method substantially enhances the spatial-spectral fidelity and\ndetail of the reconstructed images, providing a more comprehensive and accurate\ndepiction of the scene. Extensive experiments conducted on both real and\nsimulated datasets demonstrate that our approach consistently outperforms\nexisting state-of-the-art methods in terms of reconstruction quality and\ncomputational complexity. Code will be released.\n", "date": "2023-12-20", "categories": []}
{"id": "2406.04859", "title": "Stochastic full waveform inversion with deep generative prior for\n  uncertainty quantification", "abstract": "  To obtain high-resolution images of subsurface structures from seismic data,\nseismic imaging techniques such as Full Waveform Inversion (FWI) serve as\ncrucial tools. However, FWI involves solving a nonlinear and often non-unique\ninverse problem, presenting challenges such as local minima trapping and\ninadequate handling of inherent uncertainties. In addressing these challenges,\nwe propose leveraging deep generative models as the prior distribution of\ngeophysical parameters for stochastic Bayesian inversion. This approach\nintegrates the adjoint state gradient for efficient back-propagation from the\nnumerical solution of partial differential equations. Additionally, we\nintroduce explicit and implicit variational Bayesian inference methods. The\nexplicit method computes variational distribution density using a normalizing\nflow-based neural network, enabling computation of the Bayesian posterior of\nparameters. Conversely, the implicit method employs an inference network\nattached to a pretrained generative model to estimate density, incorporating an\nentropy estimator. Furthermore, we also experimented with the Stein Variational\nGradient Descent (SVGD) method as another variational inference technique,\nusing particles. We compare these variational Bayesian inference methods with\nconventional Markov chain Monte Carlo (McMC) sampling. Each method is able to\nquantify uncertainties and to generate seismic data-conditioned realizations of\nsubsurface geophysical parameters. This framework provides insights into\nsubsurface structures while accounting for inherent uncertainties.\n", "date": "2024-06-07", "categories": []}
{"id": "2402.03412", "title": "See More Details: Efficient Image Super-Resolution by Experts Mining", "abstract": "  Reconstructing high-resolution (HR) images from low-resolution (LR) inputs\nposes a significant challenge in image super-resolution (SR). While recent\napproaches have demonstrated the efficacy of intricate operations customized\nfor various objectives, the straightforward stacking of these disparate\noperations can result in a substantial computational burden, hampering their\npractical utility. In response, we introduce SeemoRe, an efficient SR model\nemploying expert mining. Our approach strategically incorporates experts at\ndifferent levels, adopting a collaborative methodology. At the macro scale, our\nexperts address rank-wise and spatial-wise informative features, providing a\nholistic understanding. Subsequently, the model delves into the subtleties of\nrank choice by leveraging a mixture of low-rank experts. By tapping into\nexperts specialized in distinct key factors crucial for accurate SR, our model\nexcels in uncovering intricate intra-feature details. This collaborative\napproach is reminiscent of the concept of \"see more\", allowing our model to\nachieve an optimal performance with minimal computational costs in efficient\nsettings. The source will be publicly made available at\nhttps://github.com/eduardzamfir/seemoredetails\n", "date": "2024-02-05", "categories": []}
{"id": "2406.03694", "title": "Untrained Neural Nets for Snapshot Compressive Imaging: Theory and\n  Algorithms", "abstract": "  Snapshot compressive imaging (SCI) recovers high-dimensional (3D) data cubes\nfrom a single 2D measurement, enabling diverse applications like video and\nhyperspectral imaging to go beyond standard techniques in terms of acquisition\nspeed and efficiency. In this paper, we focus on SCI recovery algorithms that\nemploy untrained neural networks (UNNs), such as deep image prior (DIP), to\nmodel source structure. Such UNN-based methods are appealing as they have the\npotential of avoiding the computationally intensive retraining required for\ndifferent source models and different measurement scenarios. We first develop a\ntheoretical framework for characterizing the performance of such UNN-based\nmethods. The theoretical framework, on the one hand, enables us to optimize the\nparameters of data-modulating masks, and on the other hand, provides a\nfundamental connection between the number of data frames that can be recovered\nfrom a single measurement to the parameters of the untrained NN. We also employ\nthe recently proposed bagged-deep-image-prior (bagged-DIP) idea to develop SCI\nBagged Deep Video Prior (SCI-BDVP) algorithms that address the common\nchallenges faced by standard UNN solutions. Our experimental results show that\nin video SCI our proposed solution achieves state-of-the-art among UNN methods,\nand in the case of noisy measurements, it even outperforms supervised\nsolutions.\n", "date": "2024-06-06", "categories": []}
{"id": "2406.04318", "title": "Adaptive Sampling of k-Space in Magnetic Resonance for Rapid Pathology\n  Prediction", "abstract": "  Magnetic Resonance (MR) imaging, despite its proven diagnostic utility,\nremains an inaccessible imaging modality for disease surveillance at the\npopulation level. A major factor rendering MR inaccessible is lengthy scan\ntimes. An MR scanner collects measurements associated with the underlying\nanatomy in the Fourier space, also known as the k-space. Creating a\nhigh-fidelity image requires collecting large quantities of such measurements,\nincreasing the scan time. Traditionally to accelerate an MR scan, image\nreconstruction from under-sampled k-space data is the method of choice.\nHowever, recent works show the feasibility of bypassing image reconstruction\nand directly learning to detect disease directly from a sparser learned subset\nof the k-space measurements. In this work, we propose Adaptive Sampling for MR\n(ASMR), a sampling method that learns an adaptive policy to sequentially select\nk-space samples to optimize for target disease detection. On 6 out of 8\npathology classification tasks spanning the Knee, Brain, and Prostate MR scans,\nASMR reaches within 2% of the performance of a fully sampled classifier while\nusing only 8% of the k-space, as well as outperforming prior state-of-the-art\nwork in k-space sampling such as EMRT, LOUPE, and DPS.\n", "date": "2024-06-06", "categories": []}
{"id": "2312.02197", "title": "Test-Time Degradation Adaptation for Open-Set Image Restoration", "abstract": "  In contrast to close-set scenarios that restore images from a predefined set\nof degradations, open-set image restoration aims to handle the unknown\ndegradations that were unforeseen during the pretraining phase, which is\nless-touched as far as we know. This work study this challenging problem and\nreveal its essence as unidentified distribution shifts between the test and\ntraining data. Recently, test-time adaptation has emerged as a fundamental\nmethod to address this inherent disparities. Inspired by it, we propose a\ntest-time degradation adaptation framework for open-set image restoration,\nwhich consists of three components, \\textit{i.e.}, i) a pre-trained and\ndegradation-agnostic diffusion model for generating clean images, ii) a\ntest-time degradation adapter adapts the unknown degradations based on the\ninput image during the testing phase, and iii) the adapter-guided image\nrestoration guides the model through the adapter to produce the corresponding\nclean image. Through experiments on multiple degradations, we show that our\nmethod achieves comparable even better performance than those task-specific\nmethods. The code is available at\nhttps://github.com/XLearning-SCU/2024-ICML-TAO.\n", "date": "2023-12-02", "categories": []}
{"id": "2306.11034", "title": "Learning-based sound speed estimation and aberration correction in\n  linear-array photoacoustic imaging", "abstract": "  Photoacoustic (PA) image reconstruction involves acoustic inversion that\nnecessitates the specification of the speed of sound (SoS) within the medium of\npropagation. Due to the lack of information on the spatial distribution of the\nSoS within heterogeneous soft tissue, a homogeneous SoS distribution (such as\n1540 m/s) is typically assumed in PA image reconstruction, similar to that of\nultrasound (US) imaging. Failure to compensate the SoS variations leads to\naberration artefacts, deteriorating the image quality. Various methods have\nbeen proposed to address this issue, but they usually involve complex hardware\nand/or time-consuming algorithms, hindering clinical translation. In this work,\nwe introduce a deep learning framework for SoS estimation and subsequent\naberration correction in a dual-modal PA/US imaging system exploiting a\nclinical US probe. As the acquired PA and US images were inherently\nco-registered, the estimated SoS distribution from US channel data using a deep\nneural network was incorporated for accurate PA image reconstruction. The\nframework comprised an initial pre-training stage based on digital phantoms,\nwhich was further enhanced through transfer learning using physical phantom\ndata and associated SoS maps obtained from measurements. This framework\nachieved a root mean square error of 10.2 m/s and 15.2 m/s for SoS estimation\non digital and physical phantoms, respectively and structural similarity index\nmeasures of up to 0.86 for PA reconstructions as compared to the conventional\napproach of 0.69. A maximum of 1.2 times improvement in signal-to-noise ratio\nof PA images was further demonstrated with a human volunteer study. Our results\nshow that the proposed framework could be valuable in various clinical and\npreclinical applications to enhance PA image reconstruction.\n", "date": "2023-06-19", "categories": []}
{"id": "2311.16518", "title": "SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution", "abstract": "  Owe to the powerful generative priors, the pre-trained text-to-image (T2I)\ndiffusion models have become increasingly popular in solving the real-world\nimage super-resolution problem. However, as a consequence of the heavy quality\ndegradation of input low-resolution (LR) images, the destruction of local\nstructures can lead to ambiguous image semantics. As a result, the content of\nreproduced high-resolution image may have semantic errors, deteriorating the\nsuper-resolution performance. To address this issue, we present a\nsemantics-aware approach to better preserve the semantic fidelity of generative\nreal-world image super-resolution. First, we train a degradation-aware prompt\nextractor, which can generate accurate soft and hard semantic prompts even\nunder strong degradation. The hard semantic prompts refer to the image tags,\naiming to enhance the local perception ability of the T2I model, while the soft\nsemantic prompts compensate for the hard ones to provide additional\nrepresentation information. These semantic prompts encourage the T2I model to\ngenerate detailed and semantically accurate results. Furthermore, during the\ninference process, we integrate the LR images into the initial sampling noise\nto mitigate the diffusion model's tendency to generate excessive random\ndetails. The experiments show that our method can reproduce more realistic\nimage details and hold better the semantics. The source code of our method can\nbe found at https://github.com/cswry/SeeSR.\n", "date": "2023-11-27", "categories": []}
{"id": "2403.02611", "title": "A Unified Framework for Microscopy Defocus Deblur with Multi-Pyramid\n  Transformer and Contrastive Learning", "abstract": "  Defocus blur is a persistent problem in microscope imaging that poses harm to\npathology interpretation and medical intervention in cell microscopy and\nmicroscope surgery. To address this problem, a unified framework including the\nmulti-pyramid transformer (MPT) and extended frequency contrastive\nregularization (EFCR) is proposed to tackle two outstanding challenges in\nmicroscopy deblur: longer attention span and data deficiency. The MPT employs\nan explicit pyramid structure at each network stage that integrates the\ncross-scale window attention (CSWA), the intra-scale channel attention (ISCA),\nand the feature-enhancing feed-forward network (FEFN) to capture long-range\ncross-scale spatial interaction and global channel context. The EFCR addresses\nthe data deficiency problem by exploring latent deblur signals from different\nfrequency bands. It also enables deblur knowledge transfer to learn\ncross-domain information from extra data, improving deblur performance for\nlabeled and unlabeled data. Extensive experiments and downstream task\nvalidation show the framework achieves state-of-the-art performance across\nmultiple datasets. Project page: https://github.com/PieceZhang/MPT-CataBlur.\n", "date": "2024-03-05", "categories": []}
{"id": "2406.01630", "title": "Equivariant amortized inference of poses for cryo-EM", "abstract": "  Cryo-EM is a vital technique for determining 3D structure of biological\nmolecules such as proteins and viruses. The cryo-EM reconstruction problem is\nchallenging due to the high noise levels, the missing poses of particles, and\nthe computational demands of processing large datasets. A promising solution to\nthese challenges lies in the use of amortized inference methods, which have\nshown particular efficacy in pose estimation for large datasets. However, these\nmethods also encounter convergence issues, often necessitating sophisticated\ninitialization strategies or engineered solutions for effective convergence.\nBuilding upon the existing cryoAI pipeline, which employs a symmetric loss\nfunction to address convergence problems, this work explores the emergence and\npersistence of these issues within the pipeline. Additionally, we explore the\nimpact of equivariant amortized inference on enhancing convergence. Our\ninvestigations reveal that, when applied to simulated data, a pipeline\nincorporating an equivariant encoder not only converges faster and more\nfrequently than the standard approach but also demonstrates superior\nperformance in terms of pose estimation accuracy and the resolution of the\nreconstructed volume. Notably, $D_4$-equivariant encoders make the symmetric\nloss superfluous and, therefore, allow for a more efficient reconstruction\npipeline.\n", "date": "2024-06-01", "categories": []}
{"id": "2406.01994", "title": "3D Imaging of Complex Specular Surfaces by Fusing Polarimetric and\n  Deflectometric Information", "abstract": "  Accurate and fast 3D imaging of specular surfaces still poses major\nchallenges for state-of-the-art optical measurement principles. Frequently used\nmethods, such as phase-measuring deflectometry (PMD) or shape-from-polarization\n(SfP), rely on strong assumptions about the measured objects, limiting their\ngeneralizability in broader application areas like medical imaging, industrial\ninspection, virtual reality, or cultural heritage analysis. In this paper, we\nintroduce a measurement principle that utilizes a novel technique to\neffectively encode and decode the information contained in a light field\nreflected off a specular surface. We combine polarization cues from SfP with\ngeometric information obtained from PMD to resolve all arising ambiguities in\nthe 3D measurement. Moreover, our approach removes the unrealistic orthographic\nimaging assumption for SfP, which significantly improves the respective\nresults. We showcase our new technique by demonstrating single-shot and\nmulti-shot measurements on complex-shaped specular surfaces, displaying an\nevaluated accuracy of surface normals below $0.6^\\circ$.\n", "date": "2024-06-04", "categories": []}
{"id": "2403.12063", "title": "Consistency Model is an Effective Posterior Sample Approximation for\n  Diffusion Inverse Solvers", "abstract": "  Diffusion Inverse Solvers (DIS) are designed to sample from the conditional\ndistribution $p_{\\theta}(X_0|y)$, with a predefined diffusion model\n$p_{\\theta}(X_0)$, an operator $f(\\cdot)$, and a measurement $y=f(x'_0)$\nderived from an unknown image $x'_0$. Existing DIS estimate the conditional\nscore function by evaluating $f(\\cdot)$ with an approximated posterior sample\ndrawn from $p_{\\theta}(X_0|X_t)$. However, most prior approximations rely on\nthe posterior means, which may not lie in the support of the image\ndistribution, thereby potentially diverge from the appearance of genuine\nimages. Such out-of-support samples may significantly degrade the performance\nof the operator $f(\\cdot)$, particularly when it is a neural network. In this\npaper, we introduces a novel approach for posterior approximation that\nguarantees to generate valid samples within the support of the image\ndistribution, and also enhances the compatibility with neural network-based\noperators $f(\\cdot)$. We first demonstrate that the solution of the Probability\nFlow Ordinary Differential Equation (PF-ODE) with an initial value $x_t$ yields\nan effective posterior sample $p_{\\theta}(X_0|X_t=x_t)$. Based on this\nobservation, we adopt the Consistency Model (CM), which is distilled from\nPF-ODE, for posterior sampling. Furthermore, we design a novel family of DIS\nusing only CM. Through extensive experiments, we show that our proposed method\nfor posterior sample approximation substantially enhance the effectiveness of\nDIS for neural network operators $f(\\cdot)$ (e.g., in semantic segmentation).\nAdditionally, our experiments demonstrate the effectiveness of the new CM-based\ninversion techniques. The source code is provided in the supplementary\nmaterial.\n", "date": "2024-02-09", "categories": []}
{"id": "2405.15475", "title": "Efficient Degradation-aware Any Image Restoration", "abstract": "  Reconstructing missing details from degraded low-quality inputs poses a\nsignificant challenge. Recent progress in image restoration has demonstrated\nthe efficacy of learning large models capable of addressing various\ndegradations simultaneously. Nonetheless, these approaches introduce\nconsiderable computational overhead and complex learning paradigms, limiting\ntheir practical utility. In response, we propose \\textit{DaAIR}, an efficient\nAll-in-One image restorer employing a Degradation-aware Learner (DaLe) in the\nlow-rank regime to collaboratively mine shared aspects and subtle nuances\nacross diverse degradations, generating a degradation-aware embedding. By\ndynamically allocating model capacity to input degradations, we realize an\nefficient restorer integrating holistic and specific learning within a unified\nmodel. Furthermore, DaAIR introduces a cost-efficient parameter update\nmechanism that enhances degradation awareness while maintaining computational\nefficiency. Extensive comparisons across five image degradations demonstrate\nthat our DaAIR outperforms both state-of-the-art All-in-One models and\ndegradation-specific counterparts, affirming our efficacy and practicality. The\nsource will be publicly made available at https://eduardzamfir.github.io/daair/\n", "date": "2024-05-24", "categories": []}
{"id": "2406.00449", "title": "Dual Hyperspectral Mamba for Efficient Spectral Compressive Imaging", "abstract": "  Deep unfolding methods have made impressive progress in restoring 3D\nhyperspectral images (HSIs) from 2D measurements through convolution neural\nnetworks or Transformers in spectral compressive imaging. However, they cannot\nefficiently capture long-range dependencies using global receptive fields,\nwhich significantly limits their performance in HSI reconstruction. Moreover,\nthese methods may suffer from local context neglect if we directly utilize\nMamba to unfold a 2D feature map as a 1D sequence for modeling global\nlong-range dependencies. To address these challenges, we propose a novel Dual\nHyperspectral Mamba (DHM) to explore both global long-range dependencies and\nlocal contexts for efficient HSI reconstruction. After learning informative\nparameters to estimate degradation patterns of the CASSI system, we use them to\nscale the linear projection and offer noise level for the denoiser (i.e., our\nproposed DHM). Specifically, our DHM consists of multiple dual hyperspectral S4\nblocks (DHSBs) to restore original HSIs. Particularly, each DHSB contains a\nglobal hyperspectral S4 block (GHSB) to model long-range dependencies across\nthe entire high-resolution HSIs using global receptive fields, and a local\nhyperspectral S4 block (LHSB) to address local context neglect by establishing\nstructured state-space sequence (S4) models within local windows. Experiments\nverify the benefits of our DHM for HSI reconstruction. The source codes and\nmodels will be available at https://github.com/JiahuaDong/DHM.\n", "date": "2024-06-01", "categories": []}
{"id": "2406.00683", "title": "Exploiting Frequency Correlation for Hyperspectral Image Reconstruction", "abstract": "  Deep priors have emerged as potent methods in hyperspectral image (HSI)\nreconstruction. While most methods emphasize space-domain learning using image\nspace priors like non-local similarity, frequency-domain learning using image\nfrequency priors remains neglected, limiting the reconstruction capability of\nnetworks. In this paper, we first propose a Hyperspectral Frequency Correlation\n(HFC) prior rooted in in-depth statistical frequency analyses of existent HSI\ndatasets. Leveraging the HFC prior, we subsequently establish the frequency\ndomain learning composed of a Spectral-wise self-Attention of Frequency (SAF)\nand a Spectral-spatial Interaction of Frequency (SIF) targeting low-frequency\nand high-frequency components, respectively. The outputs of SAF and SIF are\nadaptively merged by a learnable gating filter, thus achieving a thorough\nexploitation of image frequency priors. Integrating the frequency domain\nlearning and the existing space domain learning, we finally develop the\nCorrelation-driven Mixing Domains Transformer (CMDT) for HSI reconstruction.\nExtensive experiments highlight that our method surpasses various\nstate-of-the-art (SOTA) methods in reconstruction quality and computational\nefficiency.\n", "date": "2024-06-02", "categories": []}
{"id": "2406.00859", "title": "Streaming quanta sensors for online, high-performance imaging and vision", "abstract": "  Recently quanta image sensors (QIS) -- ultra-fast, zero-read-noise binary\nimage sensors -- have demonstrated remarkable imaging capabilities in many\nchallenging scenarios. Despite their potential, the adoption of these sensors\nis severely hampered by (a) high data rates and (b) the need for new\ncomputational pipelines to handle the unconventional raw data. We introduce a\nsimple, low-bandwidth computational pipeline to address these challenges. Our\napproach is based on a novel streaming representation with a small memory\nfootprint, efficiently capturing intensity information at multiple temporal\nscales. Updating the representation requires only 16 floating-point\noperations/pixel, which can be efficiently computed online at the native frame\nrate of the binary frames. We use a neural network operating on this\nrepresentation to reconstruct videos in real-time (10-30 fps). We illustrate\nwhy such representation is well-suited for these emerging sensors, and how it\noffers low latency and high frame rate while retaining flexibility for\ndownstream computer vision. Our approach results in significant data bandwidth\nreductions ~100X and real-time image reconstruction and computer vision --\n$10^4$-$10^5$ reduction in computation than existing state-of-the-art approach\nwhile maintaining comparable quality. To the best of our knowledge, our\napproach is the first to achieve online, real-time image reconstruction on QIS.\n", "date": "2024-06-02", "categories": []}
{"id": "2406.01028", "title": "LLEMamba: Low-Light Enhancement via Relighting-Guided Mamba with Deep\n  Unfolding Network", "abstract": "  Transformer-based low-light enhancement methods have yielded promising\nperformance by effectively capturing long-range dependencies in a global\ncontext. However, their elevated computational demand limits the scalability of\nmultiple iterations in deep unfolding networks, and hence they have difficulty\nin flexibly balancing interpretability and distortion. To address this issue,\nwe propose a novel Low-Light Enhancement method via relighting-guided Mamba\nwith a deep unfolding network (LLEMamba), whose theoretical interpretability\nand fidelity are guaranteed by Retinex optimization and Mamba deep priors,\nrespectively. Specifically, our LLEMamba first constructs a Retinex model with\ndeep priors, embedding the iterative optimization process based on the\nAlternating Direction Method of Multipliers (ADMM) within a deep unfolding\nnetwork. Unlike Transformer, to assist the deep unfolding framework with\nmultiple iterations, the proposed LLEMamba introduces a novel Mamba\narchitecture with lower computational complexity, which not only achieves\nlight-dependent global visual context for dark images during reflectance\nrelight but also optimizes to obtain more stable closed-form solutions.\nExperiments on the benchmarks show that LLEMamba achieves superior quantitative\nevaluations and lower distortion visual results compared to existing\nstate-of-the-art methods.\n", "date": "2024-06-03", "categories": []}
{"id": "2405.20910", "title": "Predicting ptychography probe positions using single-shot phase\n  retrieval neural network", "abstract": "  Ptychography is a powerful imaging technique that is used in a variety of\nfields, including materials science, biology, and nanotechnology. However, the\naccuracy of the reconstructed ptychography image is highly dependent on the\naccuracy of the recorded probe positions which often contain errors. These\nerrors are typically corrected jointly with phase retrieval through numerical\noptimization approaches. When the error accumulates along the scan path or when\nthe error magnitude is large, these approaches may not converge with\nsatisfactory result. We propose a fundamentally new approach for ptychography\nprobe position prediction for data with large position errors, where a neural\nnetwork is used to make single-shot phase retrieval on individual diffraction\npatterns, yielding the object image at each scan point. The pairwise offsets\namong these images are then found using a robust image registration method, and\nthe results are combined to yield the complete scan path by constructing and\nsolving a linear equation. We show that our method can achieve good position\nprediction accuracy for data with large and accumulating errors on the order of\n$10^2$ pixels, a magnitude that often makes optimization-based algorithms fail\nto converge. For ptychography instruments without sophisticated position\ncontrol equipment such as interferometers, our method is of significant\npractical potential.\n", "date": "2024-05-31", "categories": []}
{"id": "2405.19516", "title": "Enabling Visual Recognition at Radio Frequency", "abstract": "  This paper introduces PanoRadar, a novel RF imaging system that brings RF\nresolution close to that of LiDAR, while providing resilience against\nconditions challenging for optical signals. Our LiDAR-comparable 3D imaging\nresults enable, for the first time, a variety of visual recognition tasks at\nradio frequency, including surface normal estimation, semantic segmentation,\nand object detection. PanoRadar utilizes a rotating single-chip mmWave radar,\nalong with a combination of novel signal processing and machine learning\nalgorithms, to create high-resolution 3D images of the surroundings. Our system\naccurately estimates robot motion, allowing for coherent imaging through a\ndense grid of synthetic antennas. It also exploits the high azimuth resolution\nto enhance elevation resolution using learning-based methods. Furthermore,\nPanoRadar tackles 3D learning via 2D convolutions and addresses challenges due\nto the unique characteristics of RF signals. Our results demonstrate\nPanoRadar's robust performance across 12 buildings.\n", "date": "2024-05-29", "categories": []}
{"id": "2401.16779", "title": "All-optical complex field imaging using diffractive processors", "abstract": "  Complex field imaging, which captures both the amplitude and phase\ninformation of input optical fields or objects, can offer rich structural\ninsights into samples, such as their absorption and refractive index\ndistributions. However, conventional image sensors are intensity-based and\ninherently lack the capability to directly measure the phase distribution of a\nfield. This limitation can be overcome using interferometric or holographic\nmethods, often supplemented by iterative phase retrieval algorithms, leading to\na considerable increase in hardware complexity and computational demand. Here,\nwe present a complex field imager design that enables snapshot imaging of both\nthe amplitude and quantitative phase information of input fields using an\nintensity-based sensor array without any digital processing. Our design\nutilizes successive deep learning-optimized diffractive surfaces that are\nstructured to collectively modulate the input complex field, forming two\nindependent imaging channels that perform amplitude-to-amplitude and\nphase-to-intensity transformations between the input and output planes within a\ncompact optical design, axially spanning ~100 wavelengths. The intensity\ndistributions of the output fields at these two channels on the sensor plane\ndirectly correspond to the amplitude and quantitative phase profiles of the\ninput complex field, eliminating the need for any digital image reconstruction\nalgorithms. We experimentally validated the efficacy of our complex field\ndiffractive imager designs through 3D-printed prototypes operating at the\nterahertz spectrum, with the output amplitude and phase channel images closely\naligning with our numerical simulations. We envision that this complex field\nimager will have various applications in security, biomedical imaging, sensing\nand material science, among others.\n", "date": "2024-01-30", "categories": []}
{"id": "2405.19079", "title": "On the Influence of Smoothness Constraints in Computed Tomography Motion\n  Compensation", "abstract": "  Computed tomography (CT) relies on precise patient immobilization during\nimage acquisition. Nevertheless, motion artifacts in the reconstructed images\ncan persist. Motion compensation methods aim to correct such artifacts\npost-acquisition, often incorporating temporal smoothness constraints on the\nestimated motion patterns. This study analyzes the influence of a spline-based\nmotion model within an existing rigid motion compensation algorithm for\ncone-beam CT on the recoverable motion frequencies. Results demonstrate that\nthe choice of motion model crucially influences recoverable frequencies. The\noptimization-based motion compensation algorithm is able to accurately fit the\nspline nodes for frequencies almost up to the node-dependent theoretical limit\naccording to the Nyquist-Shannon theorem. Notably, a higher node count does not\ncompromise reconstruction performance for slow motion patterns, but can extend\nthe range of recoverable high frequencies for the investigated algorithm.\nEventually, the optimal motion model is dependent on the imaged anatomy,\nclinical use case, and scanning protocol and should be tailored carefully to\nthe expected motion frequency spectrum to ensure accurate motion compensation.\n", "date": "2024-05-29", "categories": []}
{"id": "2310.10123", "title": "AutoDIR: Automatic All-in-One Image Restoration with Latent Diffusion", "abstract": "  We present AutoDIR, an innovative all-in-one image restoration system\nincorporating latent diffusion. AutoDIR excels in its ability to automatically\nidentify and restore images suffering from a range of unknown degradations.\nAutoDIR offers intuitive open-vocabulary image editing, empowering users to\ncustomize and enhance images according to their preferences. Specifically,\nAutoDIR consists of two key stages: a Blind Image Quality Assessment (BIQA)\nstage based on a semantic-agnostic vision-language model which automatically\ndetects unknown image degradations for input images, an All-in-One Image\nRestoration (AIR) stage utilizes structural-corrected latent diffusion which\nhandles multiple types of image degradations. Extensive experimental evaluation\ndemonstrates that AutoDIR outperforms state-of-the-art approaches for a wider\nrange of image restoration tasks. The design of AutoDIR also enables flexible\nuser control (via text prompt) and generalization to new tasks as a foundation\nmodel of image restoration. Project is available at:\n\\url{https://jiangyitong.github.io/AutoDIR_webpage/}.\n", "date": "2023-10-16", "categories": []}
{"id": "2405.17696", "title": "Physics-guided Full Waveform Inversion using Encoder-Solver\n  Convolutional Neural Networks", "abstract": "  Full Waveform Inversion (FWI) is an inverse problem for estimating the wave\nvelocity distribution in a given domain, based on observed data on the\nboundaries. The inversion is computationally demanding because we are required\nto solve multiple forward problems, either in time or frequency domains, to\nsimulate data that are then iteratively fitted to the observed data. We\nconsider FWI in the frequency domain, where the Helmholtz equation is used as a\nforward model, and its repeated solution is the main computational bottleneck\nof the inversion process. To ease this cost, we integrate a learning process of\nan encoder-solver preconditioner that is based on convolutional neural networks\n(CNNs). The encoder-solver is trained to effectively precondition the\ndiscretized Helmholtz operator given velocity medium parameters. Then, by\nre-training the CNN between the iterations of the optimization process, the\nencoder-solver is adapted to the iteratively evolving velocity medium as part\nof the inversion. Without retraining, the performance of the solver\ndeteriorates as the medium changes. Using our light retraining procedures, we\nobtain the forward simulations effectively throughout the process. We\ndemonstrate our approach to solving FWI problems using 2D geophysical models\nwith high-frequency data.\n", "date": "2024-05-27", "categories": []}
{"id": "2404.19534", "title": "MIPI 2024 Challenge on Nighttime Flare Removal: Methods and Results", "abstract": "  The increasing demand for computational photography and imaging on mobile\nplatforms has led to the widespread development and integration of advanced\nimage sensors with novel algorithms in camera systems. However, the scarcity of\nhigh-quality data for research and the rare opportunity for in-depth exchange\nof views from industry and academia constrain the development of mobile\nintelligent photography and imaging (MIPI). Building on the achievements of the\nprevious MIPI Workshops held at ECCV 2022 and CVPR 2023, we introduce our third\nMIPI challenge including three tracks focusing on novel image sensors and\nimaging algorithms. In this paper, we summarize and review the Nighttime Flare\nRemoval track on MIPI 2024. In total, 170 participants were successfully\nregistered, and 14 teams submitted results in the final testing phase. The\ndeveloped solutions in this challenge achieved state-of-the-art performance on\nNighttime Flare Removal. More details of this challenge and the link to the\ndataset can be found at https://mipi-challenge.org/MIPI2024/.\n", "date": "2024-04-30", "categories": []}
{"id": "2405.15817", "title": "Rethinking the Elementary Function Fusion for Single-Image Dehazing", "abstract": "  This paper addresses the limitations of physical models in the current field\nof image dehazing by proposing an innovative dehazing network (CL2S). Building\non the DM2F model, it identifies issues in its ablation experiments and\nreplaces the original logarithmic function model with a trigonometric (sine)\nmodel. This substitution aims to better fit the complex and variable\ndistribution of haze. The approach also integrates the atmospheric scattering\nmodel and other elementary functions to enhance dehazing performance.\nExperimental results demonstrate that CL2S achieves outstanding performance on\nmultiple dehazing datasets, particularly in maintaining image details and color\nauthenticity. Additionally, systematic ablation experiments supplementing DM2F\nvalidate the concerns raised about DM2F and confirm the necessity and\neffectiveness of the functional components in the proposed CL2S model. Our code\nis available at \\url{https://github.com/YesianRohn/CL2S}, where the\ncorresponding pre-trained models can also be accessed.\n", "date": "2024-05-23", "categories": []}
{"id": "2405.17167", "title": "Partitioned Hankel-based Diffusion Models for Few-shot Low-dose CT\n  Reconstruction", "abstract": "  Low-dose computed tomography (LDCT) plays a vital role in clinical\napplications by mitigating radiation risks. Nevertheless, reducing radiation\ndoses significantly degrades image quality. Concurrently, common deep learning\nmethods demand extensive data, posing concerns about privacy, cost, and time\nconstraints. Consequently, we propose a few-shot low-dose CT reconstruction\nmethod using Partitioned Hankel-based Diffusion (PHD) models. During the prior\nlearning stage, the projection data is first transformed into multiple\npartitioned Hankel matrices. Structured tensors are then extracted from these\nmatrices to facilitate prior learning through multiple diffusion models. In the\niterative reconstruction stage, an iterative stochastic differential equation\nsolver is employed along with data consistency constraints to update the\nacquired projection data. Furthermore, penalized weighted least-squares and\ntotal variation techniques are introduced to enhance the resulting image\nquality. The results approximate those of normal-dose counterparts, validating\nPHD model as an effective and practical model for reducing artifacts and noise\nwhile preserving image quality.\n", "date": "2024-05-27", "categories": []}
{"id": "2311.10701", "title": "SpACNN-LDVAE: Spatial Attention Convolutional Latent Dirichlet\n  Variational Autoencoder for Hyperspectral Pixel Unmixing", "abstract": "  The hyperspectral pixel unmixing aims to find the underlying materials\n(endmembers) and their proportions (abundances) in pixels of a hyperspectral\nimage. This work extends the Latent Dirichlet Variational Autoencoder (LDVAE)\npixel unmixing scheme by taking into account local spatial context while\nperforming pixel unmixing. The proposed method uses an isotropic convolutional\nneural network with spatial attention to encode pixels as a dirichlet\ndistribution over endmembers. We have evaluated our model on Samson, Hydice\nUrban, Cuprite, and OnTech-HSI-Syn-21 datasets. Our model also leverages the\ntransfer learning paradigm for Cuprite Dataset, where we train the model on\nsynthetic data and evaluate it on the real-world data. The results suggest that\nincorporating spatial context improves both endmember extraction and abundance\nestimation.\n", "date": "2023-11-17", "categories": []}
{"id": "2210.10605", "title": "Provably Convergent Plug & Play Linearized ADMM, applied to Deblurring\n  Spatially Varying Kernels", "abstract": "  Plug & Play methods combine proximal algorithms with denoiser priors to solve\ninverse problems. These methods rely on the computability of the proximal\noperator of the data fidelity term. In this paper, we propose a Plug & Play\nframework based on linearized ADMM that allows us to bypass the computation of\nintractable proximal operators. We demonstrate the convergence of the algorithm\nand provide results on restoration tasks such as super-resolution and\ndeblurring with non-uniform blur.\n", "date": "2022-10-19", "categories": []}
{"id": "2303.11217", "title": "Inverse problem regularization with hierarchical variational\n  autoencoders", "abstract": "  In this paper, we propose to regularize ill-posed inverse problems using a\ndeep hierarchical variational autoencoder (HVAE) as an image prior. The\nproposed method synthesizes the advantages of i) denoiser-based Plug \\& Play\napproaches and ii) generative model based approaches to inverse problems.\nFirst, we exploit VAE properties to design an efficient algorithm that benefits\nfrom convergence guarantees of Plug-and-Play (PnP) methods. Second, our\napproach is not restricted to specialized datasets and the proposed PnP-HVAE\nmodel is able to solve image restoration problems on natural images of any\nsize. Our experiments show that the proposed PnP-HVAE method is competitive\nwith both SOTA denoiser-based PnP approaches, and other SOTA restoration\nmethods based on generative models.\n", "date": "2023-03-20", "categories": []}
{"id": "2308.02958", "title": "K-band: Self-supervised MRI Reconstruction via Stochastic Gradient\n  Descent over K-space Subsets", "abstract": "  Although deep learning (DL) methods are powerful for solving inverse\nproblems, their reliance on high-quality training data is a major hurdle. This\nis significant in high-dimensional (dynamic/volumetric) magnetic resonance\nimaging (MRI), where acquisition of high-resolution fully sampled k-space data\nis impractical. We introduce a novel mathematical framework, dubbed k-band,\nthat enables training DL models using only partial, limited-resolution k-space\ndata. Specifically, we introduce training with stochastic gradient descent\n(SGD) over k-space subsets. In each training iteration, rather than using the\nfully sampled k-space for computing gradients, we use only a small k-space\nportion. This concept is compatible with different sampling strategies; here we\ndemonstrate the method for k-space \"bands\", which have limited resolution in\none dimension and can hence be acquired rapidly. We prove analytically that our\nmethod stochastically approximates the gradients computed in a fully-supervised\nsetup, when two simple conditions are met: (i) the limited-resolution axis is\nchosen randomly-uniformly for every new scan, hence k-space is fully covered\nacross the entire training set, and (ii) the loss function is weighed with a\nmask, derived here analytically, which facilitates accurate reconstruction of\nhigh-resolution details. Numerical experiments with raw MRI data indicate that\nk-band outperforms two other methods trained on limited-resolution data and\nperforms comparably to state-of-the-art (SoTA) methods trained on\nhigh-resolution data. k-band hence obtains SoTA performance, with the advantage\nof training using only limited-resolution data. This work hence introduces a\npractical, easy-to-implement, self-supervised training framework, which\ninvolves fast acquisition and self-supervised reconstruction and offers\ntheoretical guarantees.\n", "date": "2023-08-05", "categories": []}
{"id": "2312.01831", "title": "Equivariant plug-and-play image reconstruction", "abstract": "  Plug-and-play algorithms constitute a popular framework for solving inverse\nimaging problems that rely on the implicit definition of an image prior via a\ndenoiser. These algorithms can leverage powerful pre-trained denoisers to solve\na wide range of imaging tasks, circumventing the necessity to train models on a\nper-task basis. Unfortunately, plug-and-play methods often show unstable\nbehaviors, hampering their promise of versatility and leading to suboptimal\nquality of reconstructed images. In this work, we show that enforcing\nequivariance to certain groups of transformations (rotations, reflections,\nand/or translations) on the denoiser strongly improves the stability of the\nalgorithm as well as its reconstruction quality. We provide a theoretical\nanalysis that illustrates the role of equivariance on better performance and\nstability. We present a simple algorithm that enforces equivariance on any\nexisting denoiser by simply applying a random transformation to the input of\nthe denoiser and the inverse transformation to the output at each iteration of\nthe algorithm. Experiments on multiple imaging modalities and denoising\nnetworks show that the equivariant plug-and-play algorithm improves both the\nreconstruction performance and the stability compared to their non-equivariant\ncounterparts.\n", "date": "2023-12-04", "categories": []}
{"id": "2405.12759", "title": "Cross-spectral Gated-RGB Stereo Depth Estimation", "abstract": "  Gated cameras flood-illuminate a scene and capture the time-gated impulse\nresponse of a scene. By employing nanosecond-scale gates, existing sensors are\ncapable of capturing mega-pixel gated images, delivering dense depth improving\non today's LiDAR sensors in spatial resolution and depth precision. Although\ngated depth estimation methods deliver a million of depth estimates per frame,\ntheir resolution is still an order below existing RGB imaging methods. In this\nwork, we combine high-resolution stereo HDR RCCB cameras with gated imaging,\nallowing us to exploit depth cues from active gating, multi-view RGB and\nmulti-view NIR sensing -- multi-view and gated cues across the entire spectrum.\nThe resulting capture system consists only of low-cost CMOS sensors and\nflood-illumination. We propose a novel stereo-depth estimation method that is\ncapable of exploiting these multi-modal multi-view depth cues, including the\nactive illumination that is measured by the RCCB camera when removing the\nIR-cut filter. The proposed method achieves accurate depth at long ranges,\noutperforming the next best existing method by 39% for ranges of 100 to 220m in\nMAE on accumulated LiDAR ground-truth. Our code, models and datasets are\navailable at https://light.princeton.edu/gatedrccbstereo/ .\n", "date": "2024-05-21", "categories": []}
{"id": "2312.10299", "title": "Image Restoration Through Generalized Ornstein-Uhlenbeck Bridge", "abstract": "  Diffusion models exhibit powerful generative capabilities enabling noise\nmapping to data via reverse stochastic differential equations. However, in\nimage restoration, the focus is on the mapping relationship from low-quality to\nhigh-quality images. Regarding this issue, we introduce the Generalized\nOrnstein-Uhlenbeck Bridge (GOUB) model. By leveraging the natural\nmean-reverting property of the generalized OU process and further eliminating\nthe variance of its steady-state distribution through the Doob's h-transform,\nwe achieve diffusion mappings from point to point enabling the recovery of\nhigh-quality images from low-quality ones. Moreover, we unravel the fundamental\nmathematical essence shared by various bridge models, all of which are special\ninstances of GOUB and empirically demonstrate the optimality of our proposed\nmodels. Additionally, we present the corresponding Mean-ODE model adept at\ncapturing both pixel-level details and structural perceptions. Experimental\noutcomes showcase the state-of-the-art performance achieved by both models\nacross diverse tasks, including inpainting, deraining, and super-resolution.\nCode is available at \\url{https://github.com/Hammour-steak/GOUB}.\n", "date": "2023-12-16", "categories": []}
{"id": "2405.11823", "title": "Stereo-Knowledge Distillation from dpMV to Dual Pixels for Light Field\n  Video Reconstruction", "abstract": "  Dual pixels contain disparity cues arising from the defocus blur. This\ndisparity information is useful for many vision tasks ranging from autonomous\ndriving to 3D creative realism. However, directly estimating disparity from\ndual pixels is less accurate. This work hypothesizes that distilling\nhigh-precision dark stereo knowledge, implicitly or explicitly, to efficient\ndual-pixel student networks enables faithful reconstructions. This dark\nknowledge distillation should also alleviate stereo-synchronization setup and\ncalibration costs while dramatically increasing parameter and inference time\nefficiency. We collect the first and largest 3-view dual-pixel video dataset,\ndpMV, to validate our explicit dark knowledge distillation hypothesis. We show\nthat these methods outperform purely monocular solutions, especially in\nchallenging foreground-background separation regions using faithful guidance\nfrom dual pixels. Finally, we demonstrate an unconventional use case unlocked\nby dpMV and implicit dark knowledge distillation from an ensemble of teachers\nfor Light Field (LF) video reconstruction. Our LF video reconstruction method\nis the fastest and most temporally consistent to date. It remains competitive\nin reconstruction fidelity while offering many other essential properties like\nhigh parameter efficiency, implicit disocclusion handling, zero-shot\ncross-dataset transfer, geometrically consistent inference on higher\nspatial-angular resolutions, and adaptive baseline control. All source code is\navailable at the anonymous repository https://github.com/Aryan-Garg.\n", "date": "2024-05-20", "categories": []}
{"id": "2405.10357", "title": "RGB Guided ToF Imaging System: A Survey of Deep Learning-based Methods", "abstract": "  Integrating an RGB camera into a ToF imaging system has become a significant\ntechnique for perceiving the real world. The RGB guided ToF imaging system is\ncrucial to several applications, including face anti-spoofing, saliency\ndetection, and trajectory prediction. Depending on the distance of the working\nrange, the implementation schemes of the RGB guided ToF imaging systems are\ndifferent. Specifically, ToF sensors with a uniform field of illumination,\nwhich can output dense depth but have low resolution, are typically used for\nclose-range measurements. In contrast, LiDARs, which emit laser pulses and can\nonly capture sparse depth, are usually employed for long-range detection. In\nthe two cases, depth quality improvement for RGB guided ToF imaging corresponds\nto two sub-tasks: guided depth super-resolution and guided depth completion. In\nlight of the recent significant boost to the field provided by deep learning,\nthis paper comprehensively reviews the works related to RGB guided ToF imaging,\nincluding network structures, learning strategies, evaluation metrics,\nbenchmark datasets, and objective functions. Besides, we present quantitative\ncomparisons of state-of-the-art methods on widely used benchmark datasets.\nFinally, we discuss future trends and the challenges in real applications for\nfurther research.\n", "date": "2024-05-16", "categories": []}
{"id": "2405.09923", "title": "NTIRE 2024 Restore Any Image Model (RAIM) in the Wild Challenge", "abstract": "  In this paper, we review the NTIRE 2024 challenge on Restore Any Image Model\n(RAIM) in the Wild. The RAIM challenge constructed a benchmark for image\nrestoration in the wild, including real-world images with/without reference\nground truth in various scenarios from real applications. The participants were\nrequired to restore the real-captured images from complex and unknown\ndegradation, where generative perceptual quality and fidelity are desired in\nthe restoration result. The challenge consisted of two tasks. Task one employed\nreal referenced data pairs, where quantitative evaluation is available. Task\ntwo used unpaired images, and a comprehensive user study was conducted. The\nchallenge attracted more than 200 registrations, where 39 of them submitted\nresults with more than 400 submissions. Top-ranked methods improved the\nstate-of-the-art restoration performance and obtained unanimous recognition\nfrom all 18 judges. The proposed datasets are available at\nhttps://drive.google.com/file/d/1DqbxUoiUqkAIkExu3jZAqoElr_nu1IXb/view?usp=sharing\nand the homepage of this challenge is at\nhttps://codalab.lisn.upsaclay.fr/competitions/17632.\n", "date": "2024-05-16", "categories": []}
{"id": "2405.06944", "title": "Learning Monocular Depth from Focus with Event Focal Stack", "abstract": "  Depth from Focus estimates depth by determining the moment of maximum focus\nfrom multiple shots at different focal distances, i.e. the Focal Stack.\nHowever, the limited sampling rate of conventional optical cameras makes it\ndifficult to obtain sufficient focus cues during the focal sweep. Inspired by\nbiological vision, the event camera records intensity changes over time in\nextremely low latency, which provides more temporal information for focus time\nacquisition. In this study, we propose the EDFF Network to estimate sparse\ndepth from the Event Focal Stack. Specifically, we utilize the event voxel grid\nto encode intensity change information and project event time surface into the\ndepth domain to preserve per-pixel focal distance information. A\nFocal-Distance-guided Cross-Modal Attention Module is presented to fuse the\ninformation mentioned above. Additionally, we propose a Multi-level Depth\nFusion Block designed to integrate results from each level of a UNet-like\narchitecture and produce the final output. Extensive experiments validate that\nour method outperforms existing state-of-the-art approaches.\n", "date": "2024-05-11", "categories": []}
{"id": "2405.07012", "title": "Incorporating Degradation Estimation in Light Field Spatial\n  Super-Resolution", "abstract": "  Recent advancements in light field super-resolution (SR) have yielded\nimpressive results. In practice, however, many existing methods are limited by\nassuming fixed degradation models, such as bicubic downsampling, which hinders\ntheir robustness in real-world scenarios with complex degradations. To address\nthis limitation, we present LF-DEST, an effective blind Light Field SR method\nthat incorporates explicit Degradation Estimation to handle various degradation\ntypes. LF-DEST consists of two primary components: degradation estimation and\nlight field restoration. The former concurrently estimates blur kernels and\nnoise maps from low-resolution degraded light fields, while the latter\ngenerates super-resolved light fields based on the estimated degradations.\nNotably, we introduce a modulated and selective fusion module that\nintelligently combines degradation representations with image information,\nallowing for effective handling of diverse degradation types. We conduct\nextensive experiments on benchmark datasets, demonstrating that LF-DEST\nachieves superior performance across a variety of degradation scenarios in\nlight field SR.\n", "date": "2024-05-11", "categories": []}
{"id": "2405.05016", "title": "TGTM: TinyML-based Global Tone Mapping for HDR Sensors", "abstract": "  Advanced driver assistance systems (ADAS) relying on multiple cameras are\nincreasingly prevalent in vehicle technology. Yet, conventional imaging sensors\nstruggle to capture clear images in conditions with intense illumination\ncontrast, such as tunnel exits, due to their limited dynamic range. Introducing\nhigh dynamic range (HDR) sensors addresses this issue. However, the process of\nconverting HDR content to a displayable range via tone mapping often leads to\ninefficient computations, when performed directly on pixel data. In this paper,\nwe focus on HDR image tone mapping using a lightweight neural network applied\non image histogram data. Our proposed TinyML-based global tone mapping method,\ntermed as TGTM, operates at 9,000 FLOPS per RGB image of any resolution.\nAdditionally, TGTM offers a generic approach that can be incorporated to any\nclassical tone mapping method. Experimental results demonstrate that TGTM\noutperforms state-of-the-art methods on real HDR camera images by up to 5.85 dB\nhigher PSNR with orders of magnitude less computations.\n", "date": "2024-05-08", "categories": []}
{"id": "2301.06132", "title": "Deep Diversity-Enhanced Feature Representation of Hyperspectral Images", "abstract": "  In this paper, we study the problem of efficiently and effectively embedding\nthe high-dimensional spatio-spectral information of hyperspectral (HS) images,\nguided by feature diversity. Specifically, based on the theoretical formulation\nthat feature diversity is correlated with the rank of the unfolded kernel\nmatrix, we rectify 3D convolution by modifying its topology to enhance the rank\nupper-bound. This modification yields a rank-enhanced spatial-spectral\nsymmetrical convolution set (ReS$^3$-ConvSet), which not only learns diverse\nand powerful feature representations but also saves network parameters.\nAdditionally, we also propose a novel diversity-aware regularization (DA-Reg)\nterm that directly acts on the feature maps to maximize independence among\nelements. To demonstrate the superiority of the proposed ReS$^3$-ConvSet and\nDA-Reg, we apply them to various HS image processing and analysis tasks,\nincluding denoising, spatial super-resolution, and classification. Extensive\nexperiments show that the proposed approaches outperform state-of-the-art\nmethods both quantitatively and qualitatively to a significant extent. The code\nis publicly available at https://github.com/jinnh/ReSSS-ConvSet.\n", "date": "2023-01-15", "categories": []}
{"id": "2405.04867", "title": "MIPI 2024 Challenge on Demosaic for HybridEVS Camera: Methods and\n  Results", "abstract": "  The increasing demand for computational photography and imaging on mobile\nplatforms has led to the widespread development and integration of advanced\nimage sensors with novel algorithms in camera systems. However, the scarcity of\nhigh-quality data for research and the rare opportunity for in-depth exchange\nof views from industry and academia constrain the development of mobile\nintelligent photography and imaging (MIPI). Building on the achievements of the\nprevious MIPI Workshops held at ECCV 2022 and CVPR 2023, we introduce our third\nMIPI challenge including three tracks focusing on novel image sensors and\nimaging algorithms. In this paper, we summarize and review the Nighttime Flare\nRemoval track on MIPI 2024. In total, 170 participants were successfully\nregistered, and 14 teams submitted results in the final testing phase. The\ndeveloped solutions in this challenge achieved state-of-the-art performance on\nNighttime Flare Removal. More details of this challenge and the link to the\ndataset can be found at https://mipi-challenge.org/MIPI2024/.\n", "date": "2024-05-08", "categories": []}
{"id": "2405.02023", "title": "IFNet: Deep Imaging and Focusing for Handheld SAR with Millimeter-wave\n  Signals", "abstract": "  Recent advancements have showcased the potential of handheld millimeter-wave\n(mmWave) imaging, which applies synthetic aperture radar (SAR) principles in\nportable settings. However, existing studies addressing handheld motion errors\neither rely on costly tracking devices or employ simplified imaging models,\nleading to impractical deployment or limited performance. In this paper, we\npresent IFNet, a novel deep unfolding network that combines the strengths of\nsignal processing models and deep neural networks to achieve robust imaging and\nfocusing for handheld mmWave systems. We first formulate the handheld imaging\nmodel by integrating multiple priors about mmWave images and handheld phase\nerrors. Furthermore, we transform the optimization processes into an iterative\nnetwork structure for improved and efficient imaging performance. Extensive\nexperiments demonstrate that IFNet effectively compensates for handheld phase\nerrors and recovers high-fidelity images from severely distorted signals. In\ncomparison with existing methods, IFNet can achieve at least 11.89 dB\nimprovement in average peak signal-to-noise ratio (PSNR) and 64.91% improvement\nin average structural similarity index measure (SSIM) on a real-world dataset.\n", "date": "2024-05-03", "categories": []}
{"id": "2405.02787", "title": "Light Field Spatial Resolution Enhancement Framework", "abstract": "  Light field (LF) imaging captures both angular and spatial light\ndistributions, enabling advanced photographic techniques. However, micro-lens\narray (MLA)- based cameras face a spatial-angular resolution tradeoff due to a\nsingle shared sensor. We propose a novel light field framework for resolution\nenhancement, employing a modular approach. The first module generates a\nhigh-resolution, all-in-focus image. The second module, a texture transformer\nnetwork, enhances the resolution of each light field perspective independently\nusing the output of the first module as a reference image. The final module\nleverages light field regularity to jointly improve resolution across all LF\nimage perspectives. Our approach demonstrates superior performance to existing\nmethods in both qualitative and quantitative evaluations.\n", "date": "2024-05-05", "categories": []}
{"id": "2305.16965", "title": "Accelerating Diffusion Models for Inverse Problems through Shortcut\n  Sampling", "abstract": "  Diffusion models have recently demonstrated an impressive ability to address\ninverse problems in an unsupervised manner. While existing methods primarily\nfocus on modifying the posterior sampling process, the potential of the forward\nprocess remains largely unexplored. In this work, we propose Shortcut Sampling\nfor Diffusion(SSD), a novel approach for solving inverse problems in a\nzero-shot manner. Instead of initiating from random noise, the core concept of\nSSD is to find a specific transitional state that bridges the measurement image\ny and the restored image x. By utilizing the shortcut path of \"input -\ntransitional state - output\", SSD can achieve precise restoration with fewer\nsteps. To derive the transitional state during the forward process, we\nintroduce Distortion Adaptive Inversion. Moreover, we apply back projection as\nadditional consistency constraints during the generation process.\nExperimentally, we demonstrate SSD's effectiveness on multiple representative\nIR tasks. Our method achieves competitive results with only 30 NFEs compared to\nstate-of-the-art zero-shot methods(100 NFEs) and outperforms them with 100 NFEs\nin certain tasks. Code is available at https://github.com/GongyeLiu/SSD\n", "date": "2023-05-26", "categories": []}
{"id": "2403.05452", "title": "The R2D2 deep neural network series paradigm for fast precision imaging\n  in radio astronomy", "abstract": "  Radio-interferometric (RI) imaging entails solving high-resolution\nhigh-dynamic range inverse problems from large data volumes. Recent image\nreconstruction techniques grounded in optimization theory have demonstrated\nremarkable capability for imaging precision, well beyond CLEAN's capability.\nThese range from advanced proximal algorithms propelled by handcrafted\nregularization operators, such as the SARA family, to hybrid plug-and-play\n(PnP) algorithms propelled by learned regularization denoisers, such as AIRI.\nOptimization and PnP structures are however highly iterative, which hinders\ntheir ability to handle the extreme data sizes expected from future\ninstruments. To address this scalability challenge, we introduce a novel deep\nlearning approach, dubbed \"Residual-to-Residual DNN series for high-Dynamic\nrange imaging\". R2D2's reconstruction is formed as a series of residual images,\niteratively estimated as outputs of Deep Neural Networks (DNNs) taking the\nprevious iteration's image estimate and associated data residual as inputs. It\nthus takes a hybrid structure between a PnP algorithm and a learned version of\nthe matching pursuit algorithm that underpins CLEAN. We present a comprehensive\nstudy of our approach, featuring its multiple incarnations distinguished by\ntheir DNN architectures. We provide a detailed description of its training\nprocess, targeting a telescope-specific approach. R2D2's capability to deliver\nhigh precision is demonstrated in simulation, across a variety of image and\nobservation settings using the Very Large Array (VLA). Its reconstruction speed\nis also demonstrated: with only few iterations required to clean data residuals\nat dynamic ranges up to 100000, R2D2 opens the door to fast precision imaging.\nR2D2 codes are available in the BASPLib library on GitHub.\n", "date": "2024-03-08", "categories": []}
{"id": "2405.00666", "title": "RGB$\\leftrightarrow$X: Image decomposition and synthesis using material-\n  and lighting-aware diffusion models", "abstract": "  The three areas of realistic forward rendering, per-pixel inverse rendering,\nand generative image synthesis may seem like separate and unrelated sub-fields\nof graphics and vision. However, recent work has demonstrated improved\nestimation of per-pixel intrinsic channels (albedo, roughness, metallicity)\nbased on a diffusion architecture; we call this the RGB$\\rightarrow$X problem.\nWe further show that the reverse problem of synthesizing realistic images given\nintrinsic channels, X$\\rightarrow$RGB, can also be addressed in a diffusion\nframework.\n  Focusing on the image domain of interior scenes, we introduce an improved\ndiffusion model for RGB$\\rightarrow$X, which also estimates lighting, as well\nas the first diffusion X$\\rightarrow$RGB model capable of synthesizing\nrealistic images from (full or partial) intrinsic channels. Our\nX$\\rightarrow$RGB model explores a middle ground between traditional rendering\nand generative models: we can specify only certain appearance properties that\nshould be followed, and give freedom to the model to hallucinate a plausible\nversion of the rest.\n  This flexibility makes it possible to use a mix of heterogeneous training\ndatasets, which differ in the available channels. We use multiple existing\ndatasets and extend them with our own synthetic and real data, resulting in a\nmodel capable of extracting scene properties better than previous work and of\ngenerating highly realistic images of interior scenes.\n", "date": "2024-05-01", "categories": []}
{"id": "2307.16670", "title": "Conditioning Generative Latent Optimization for Sparse-View CT Image\n  Reconstruction", "abstract": "  Computed Tomography (CT) is a prominent example of Imaging Inverse Problem\nhighlighting the unrivaled performances of data-driven methods in degraded\nmeasurements setups like sparse X-ray projections. Although a significant\nproportion of deep learning approaches benefit from large supervised datasets,\nthey cannot generalize to new experimental setups. In contrast, fully\nunsupervised techniques, most notably using score-based generative models, have\nrecently demonstrated similar or better performances compared to supervised\napproaches while being flexible at test time. However, their use cases are\nlimited as they need considerable amounts of training data to have good\ngeneralization properties. Another unsupervised approach taking advantage of\nthe implicit natural bias of deep convolutional networks, Deep Image Prior, has\nrecently been adapted to solve sparse CT by reparameterizing the reconstruction\nproblem. Although this methodology does not require any training dataset, it\nenforces a weaker prior on the reconstructions when compared to data-driven\nmethods. To fill the gap between these two strategies, we propose an\nunsupervised conditional approach to the Generative Latent Optimization\nframework (cGLO). Similarly to DIP, without any training dataset, cGLO benefits\nfrom the structural bias of a decoder network. However, the prior is further\nreinforced as the effect of a likelihood objective shared between multiple\nslices being reconstructed simultaneously through the same decoder network. In\naddition, the parameters of the decoder may be initialized on an unsupervised,\nand eventually very small, training dataset to enhance the reconstruction. The\nresulting approach is tested on full-dose sparse-view CT using multiple\ntraining dataset sizes and varying numbers of viewing angles.\n", "date": "2023-07-31", "categories": []}
{"id": "2307.13927", "title": "DFR-Net: Density Feature Refinement Network for Image Dehazing Utilizing\n  Haze Density Difference", "abstract": "  In image dehazing task, haze density is a key feature and affects the\nperformance of dehazing methods. However, some of the existing methods lack a\ncomparative image to measure densities, and others create intermediate results\nbut lack the exploitation of their density differences, which can facilitate\nperception of density. To address these deficiencies, we propose a\ndensity-aware dehazing method named Density Feature Refinement Network\n(DFR-Net) that extracts haze density features from density differences and\nleverages density differences to refine density features. In DFR-Net, we first\ngenerate a proposal image that has lower overall density than the hazy input,\nbringing in global density differences. Additionally, the dehazing residual of\nthe proposal image reflects the level of dehazing performance and provides\nlocal density differences that indicate localized hard dehazing or high density\nareas. Subsequently, we introduce a Global Branch (GB) and a Local Branch (LB)\nto achieve density-awareness. In GB, we use Siamese networks for feature\nextraction of hazy inputs and proposal images, and we propose a Global Density\nFeature Refinement (GDFR) module that can refine features by pushing features\nwith different global densities further away. In LB, we explore local density\nfeatures from the dehazing residuals between hazy inputs and proposal images\nand introduce an Intermediate Dehazing Residual Feedforward (IDRF) module to\nupdate local features and pull them closer to clear image features. Sufficient\nexperiments demonstrate that the proposed method achieves results beyond the\nstate-of-the-art methods on various datasets.\n", "date": "2023-07-26", "categories": []}
{"id": "2404.17825", "title": "ODCR: Orthogonal Decoupling Contrastive Regularization for Unpaired\n  Image Dehazing", "abstract": "  Unpaired image dehazing (UID) holds significant research importance due to\nthe challenges in acquiring haze/clear image pairs with identical backgrounds.\nThis paper proposes a novel method for UID named Orthogonal Decoupling\nContrastive Regularization (ODCR). Our method is grounded in the assumption\nthat an image consists of both haze-related features, which influence the\ndegree of haze, and haze-unrelated features, such as texture and semantic\ninformation. ODCR aims to ensure that the haze-related features of the dehazing\nresult closely resemble those of the clear image, while the haze-unrelated\nfeatures align with the input hazy image. To accomplish the motivation,\nOrthogonal MLPs optimized geometrically on the Stiefel manifold are proposed,\nwhich can project image features into an orthogonal space, thereby reducing the\nrelevance between different features. Furthermore, a task-driven Depth-wise\nFeature Classifier (DWFC) is proposed, which assigns weights to the orthogonal\nfeatures based on the contribution of each channel's feature in predicting\nwhether the feature source is hazy or clear in a self-supervised fashion.\nFinally, a Weighted PatchNCE (WPNCE) loss is introduced to achieve the pulling\nof haze-related features in the output image toward those of clear images,\nwhile bringing haze-unrelated features close to those of the hazy input.\nExtensive experiments demonstrate the superior performance of our ODCR method\non UID.\n", "date": "2024-04-27", "categories": []}
{"id": "2308.09375", "title": "Image Processing and Machine Learning for Hyperspectral Unmixing: An\n  Overview and the HySUPP Python Package", "abstract": "  Spectral pixels are often a mixture of the pure spectra of the materials,\ncalled endmembers, due to the low spatial resolution of hyperspectral sensors,\ndouble scattering, and intimate mixtures of materials in the scenes. Unmixing\nestimates the fractional abundances of the endmembers within the pixel.\nDepending on the prior knowledge of endmembers, linear unmixing can be divided\ninto three main groups: supervised, semi-supervised, and unsupervised (blind)\nlinear unmixing. Advances in Image processing and machine learning\nsubstantially affected unmixing. This paper provides an overview of advanced\nand conventional unmixing approaches. Additionally, we draw a critical\ncomparison between advanced and conventional techniques from the three\ncategories. We compare the performance of the unmixing techniques on three\nsimulated and two real datasets. The experimental results reveal the advantages\nof different unmixing categories for different unmixing scenarios. Moreover, we\nprovide an open-source Python-based package available at\nhttps://github.com/BehnoodRasti/HySUPP to reproduce the results.\n", "date": "2023-08-18", "categories": []}
{"id": "2404.17433", "title": "PromptCIR: Blind Compressed Image Restoration with Prompt Learning", "abstract": "  Blind Compressed Image Restoration (CIR) has garnered significant attention\ndue to its practical applications. It aims to mitigate compression artifacts\ncaused by unknown quality factors, particularly with JPEG codecs. Existing\nworks on blind CIR often seek assistance from a quality factor prediction\nnetwork to facilitate their network to restore compressed images. However, the\npredicted numerical quality factor lacks spatial information, preventing\nnetwork adaptability toward image contents. Recent studies in\nprompt-learning-based image restoration have showcased the potential of prompts\nto generalize across varied degradation types and degrees. This motivated us to\ndesign a prompt-learning-based compressed image restoration network, dubbed\nPromptCIR, which can effectively restore images from various compress levels.\nSpecifically, PromptCIR exploits prompts to encode compression information\nimplicitly, where prompts directly interact with soft weights generated from\nimage features, thus providing dynamic content-aware and distortion-aware\nguidance for the restoration process. The light-weight prompts enable our\nmethod to adapt to different compression levels, while introducing minimal\nparameter overhead. Overall, PromptCIR leverages the powerful transformer-based\nbackbone with the dynamic prompt module to proficiently handle blind CIR tasks,\nwinning first place in the NTIRE 2024 challenge of blind compressed image\nenhancement track. Extensive experiments have validated the effectiveness of\nour proposed PromptCIR. The code is available at\nhttps://github.com/lbc12345/PromptCIR-NTIRE24.\n", "date": "2024-04-26", "categories": []}
{"id": "2307.15388", "title": "An Empirical Study of Large-Scale Data-Driven Full Waveform Inversion", "abstract": "  This paper investigates the impact of big data on deep learning models to\nhelp solve the full waveform inversion (FWI) problem. While it is well known\nthat big data can boost the performance of deep learning models in many tasks,\nits effectiveness has not been validated for FWI. To address this gap, we\npresent an empirical study that investigates how deep learning models in FWI\nbehave when trained on OpenFWI, a collection of large-scale, multi-structural,\nsynthetic datasets published recently. In particular, we train and evaluate the\nFWI models on a combination of 10 2D subsets in OpenFWI that contain 470K pairs\nof seismic data and velocity maps in total. Our experiments demonstrate that\ntraining on the combined dataset yields an average improvement of 13.03% in\nMAE, 7.19% in MSE and 1.87% in SSIM compared to each split dataset, and an\naverage improvement of 28.60%, 21.55% and 8.22% in the leave-one-out\ngeneralization test. We further demonstrate that model capacity needs to scale\nin accordance with data size for optimal improvement, where our largest model\nyields an average improvement of 20.06%, 13.39% and 0.72% compared to the\nsmallest one.\n", "date": "2023-07-28", "categories": []}
{"id": "2310.16139", "title": "Pix2HDR -- A pixel-wise acquisition and deep learning-based synthesis\n  approach for high-speed HDR videos", "abstract": "  Accurately capturing dynamic scenes with wide-ranging motion and light\nintensity is crucial for many vision applications. However, acquiring\nhigh-speed high dynamic range (HDR) video is challenging because the camera's\nframe rate restricts its dynamic range. Existing methods sacrifice speed to\nacquire multi-exposure frames. Yet, misaligned motion in these frames can still\npose complications for HDR fusion algorithms, resulting in artifacts. Instead\nof frame-based exposures, we sample the videos using individual pixels at\nvarying exposures and phase offsets. Implemented on a monochrome pixel-wise\nprogrammable image sensor, our sampling pattern simultaneously captures fast\nmotion at a high dynamic range. We then transform pixel-wise outputs into an\nHDR video using end-to-end learned weights from deep neural networks, achieving\nhigh spatiotemporal resolution with minimized motion blurring. We demonstrate\naliasing-free HDR video acquisition at 1000 FPS, resolving fast motion under\nlow-light conditions and against bright backgrounds - both challenging\nconditions for conventional cameras. By combining the versatility of pixel-wise\nsampling patterns with the strength of deep neural networks at decoding complex\nscenes, our method greatly enhances the vision system's adaptability and\nperformance in dynamic conditions.\n", "date": "2023-10-24", "categories": []}
{"id": "2404.16223", "title": "Deep RAW Image Super-Resolution. A NTIRE 2024 Challenge Survey", "abstract": "  This paper reviews the NTIRE 2024 RAW Image Super-Resolution Challenge,\nhighlighting the proposed solutions and results. New methods for RAW\nSuper-Resolution could be essential in modern Image Signal Processing (ISP)\npipelines, however, this problem is not as explored as in the RGB domain. Th\ngoal of this challenge is to upscale RAW Bayer images by 2x, considering\nunknown degradations such as noise and blur. In the challenge, a total of 230\nparticipants registered, and 45 submitted results during thee challenge period.\nThe performance of the top-5 submissions is reviewed and provided here as a\ngauge for the current state-of-the-art in RAW Image Super-Resolution.\n", "date": "2024-04-24", "categories": []}
{"id": "2404.15231", "title": "Direct Zernike Coefficient Prediction from Point Spread Functions and\n  Extended Images using Deep Learning", "abstract": "  Optical imaging quality can be severely degraded by system and sample induced\naberrations. Existing adaptive optics systems typically rely on iterative\nsearch algorithm to correct for aberrations and improve images. This study\ndemonstrates the application of convolutional neural networks to characterise\nthe optical aberration by directly predicting the Zernike coefficients from two\nto three phase-diverse optical images. We evaluated our network on 600,000\nsimulated Point Spread Function (PSF) datasets randomly generated within the\nrange of -1 to 1 radians using the first 25 Zernike coefficients. The results\nshow that using only three phase-diverse images captured above, below and at\nthe focal plane with an amplitude of 1 achieves a low RMSE of 0.10 radians on\nthe simulated PSF dataset. Furthermore, this approach directly predicts Zernike\nmodes simulated extended 2D samples, while maintaining a comparable RMSE of\n0.15 radians. We demonstrate that this approach is effective using only a\nsingle prediction step, or can be iterated a small number of times. This simple\nand straightforward technique provides rapid and accurate method for predicting\nthe aberration correction using three or less phase-diverse images, paving the\nway for evaluation on real-world dataset.\n", "date": "2024-04-23", "categories": []}
{"id": "2404.15781", "title": "Real-Time Compressed Sensing for Joint Hyperspectral Image Transmission\n  and Restoration for CubeSat", "abstract": "  This paper addresses the challenges associated with hyperspectral image (HSI)\nreconstruction from miniaturized satellites, which often suffer from stripe\neffects and are computationally resource-limited. We propose a Real-Time\nCompressed Sensing (RTCS) network designed to be lightweight and require only\nrelatively few training samples for efficient and robust HSI reconstruction in\nthe presence of the stripe effect and under noisy transmission conditions. The\nRTCS network features a simplified architecture that reduces the required\ntraining samples and allows for easy implementation on integer-8-based\nencoders, facilitating rapid compressed sensing for stripe-like HSI, which\nexactly matches the moderate design of miniaturized satellites on push broom\nscanning mechanism. This contrasts optimization-based models that demand\nhigh-precision floating-point operations, making them difficult to deploy on\nedge devices. Our encoder employs an integer-8-compatible linear projection for\nstripe-like HSI data transmission, ensuring real-time compressed sensing.\nFurthermore, based on the novel two-streamed architecture, an efficient HSI\nrestoration decoder is proposed for the receiver side, allowing for edge-device\nreconstruction without needing a sophisticated central server. This is\nparticularly crucial as an increasing number of miniaturized satellites\nnecessitates significant computing resources on the ground station. Extensive\nexperiments validate the superior performance of our approach, offering new and\nvital capabilities for existing miniaturized satellite systems.\n", "date": "2024-04-24", "categories": []}
{"id": "2304.02162", "title": "Learning to Recover Spectral Reflectance from RGB Images", "abstract": "  This paper tackles spectral reflectance recovery (SRR) from RGB images. Since\ncapturing ground-truth spectral reflectance and camera spectral sensitivity are\nchallenging and costly, most existing approaches are trained on synthetic\nimages and utilize the same parameters for all unseen testing images, which are\nsuboptimal especially when the trained models are tested on real images because\nthey never exploit the internal information of the testing images. To address\nthis issue, we adopt a self-supervised meta-auxiliary learning (MAXL) strategy\nthat fine-tunes the well-trained network parameters with each testing image to\ncombine external with internal information. To the best of our knowledge, this\nis the first work that successfully adapts the MAXL strategy to this problem.\nInstead of relying on naive end-to-end training, we also propose a novel\narchitecture that integrates the physical relationship between the spectral\nreflectance and the corresponding RGB images into the network based on our\nmathematical analysis. Besides, since the spectral reflectance of a scene is\nindependent to its illumination while the corresponding RGB images are not, we\nrecover the spectral reflectance of a scene from its RGB images captured under\nmultiple illuminations to further reduce the unknown. Qualitative and\nquantitative evaluations demonstrate the effectiveness of our proposed network\nand of the MAXL. Our code and data are available at\nhttps://github.com/Dong-Huo/SRR-MAXL.\n", "date": "2023-04-04", "categories": []}
{"id": "2404.14674", "title": "HOIN: High-Order Implicit Neural Representations", "abstract": "  Implicit neural representations (INR) suffer from worsening spectral bias,\nwhich results in overly smooth solutions to the inverse problem. To deal with\nthis problem, we propose a universal framework for processing inverse problems\ncalled \\textbf{High-Order Implicit Neural Representations (HOIN)}. By refining\nthe traditional cascade structure to foster high-order interactions among\nfeatures, HOIN enhances the model's expressive power and mitigates spectral\nbias through its neural tangent kernel's (NTK) strong diagonal properties,\naccelerating and optimizing inverse problem resolution. By analyzing the\nmodel's expression space, high-order derivatives, and the NTK matrix, we\ntheoretically validate the feasibility of HOIN. HOIN realizes 1 to 3 dB\nimprovements in most inverse problems, establishing a new state-of-the-art\nrecovery quality and training efficiency, thus providing a new general paradigm\nfor INR and paving the way for it to solve the inverse problem.\n", "date": "2024-04-23", "categories": []}
{"id": "2404.15082", "title": "Harnessing Optical Imaging Limit through Atmospheric Scattering Media", "abstract": "  Recording and identifying faint objects through atmospheric scattering media\nby an optical system are fundamentally interesting and technologically\nimportant. In this work, we introduce a comprehensive model that incorporates\ncontributions from target characteristics, atmospheric effects, imaging system,\ndigital processing, and visual perception to assess the ultimate perceptible\nlimit of geometrical imaging, specifically the angular resolution at the\nboundary of visible distance. The model allows to reevaluate the effectiveness\nof conventional imaging recording, processing, and perception and to analyze\nthe limiting factors that constrain image recognition capabilities in\natmospheric media. The simulations were compared with the experimental results\nmeasured in a fog chamber and outdoor settings. The results reveal general good\nagreement between analysis and experimental, pointing out the way to harnessing\nthe physical limit for optical imaging in scattering media. An immediate\napplication of the study is the extension of the image range by an amount of\n1.2 times with noise reduction via multi-frame averaging, hence greatly\nenhancing the capability of optical imaging in the atmosphere.\n", "date": "2024-04-23", "categories": []}
{"id": "2404.15174", "title": "Fourier-enhanced Implicit Neural Fusion Network for Multispectral and\n  Hyperspectral Image Fusion", "abstract": "  Recently, implicit neural representations (INR) have made significant strides\nin various vision-related domains, providing a novel solution for Multispectral\nand Hyperspectral Image Fusion (MHIF) tasks. However, INR is prone to losing\nhigh-frequency information and is confined to the lack of global perceptual\ncapabilities. To address these issues, this paper introduces a Fourier-enhanced\nImplicit Neural Fusion Network (FeINFN) specifically designed for MHIF task,\ntargeting the following phenomena: The Fourier amplitudes of the HR-HSI latent\ncode and LR-HSI are remarkably similar; however, their phases exhibit different\npatterns. In FeINFN, we innovatively propose a spatial and frequency implicit\nfusion function (Spa-Fre IFF), helping INR capture high-frequency information\nand expanding the receptive field. Besides, a new decoder employing a complex\nGabor wavelet activation function, called Spatial-Frequency Interactive Decoder\n(SFID), is invented to enhance the interaction of INR features. Especially, we\nfurther theoretically prove that the Gabor wavelet activation possesses a\ntime-frequency tightness property that favors learning the optimal bandwidths\nin the decoder. Experiments on two benchmark MHIF datasets verify the\nstate-of-the-art (SOTA) performance of the proposed method, both visually and\nquantitatively. Also, ablation studies demonstrate the mentioned contributions.\nThe code will be available on Anonymous GitHub\n(https://anonymous.4open.science/r/FeINFN-15C9/) after possible acceptance.\n", "date": "2024-04-23", "categories": []}
{"id": "2304.05060", "title": "SPIRiT-Diffusion: Self-Consistency Driven Diffusion Model for\n  Accelerated MRI", "abstract": "  Diffusion models have emerged as a leading methodology for image generation\nand have proven successful in the realm of magnetic resonance imaging (MRI)\nreconstruction. However, existing reconstruction methods based on diffusion\nmodels are primarily formulated in the image domain, making the reconstruction\nquality susceptible to inaccuracies in coil sensitivity maps (CSMs). k-space\ninterpolation methods can effectively address this issue but conventional\ndiffusion models are not readily applicable in k-space interpolation. To\novercome this challenge, we introduce a novel approach called SPIRiT-Diffusion,\nwhich is a diffusion model for k-space interpolation inspired by the iterative\nself-consistent SPIRiT method. Specifically, we utilize the iterative solver of\nthe self-consistent term (i.e., k-space physical prior) in SPIRiT to formulate\na novel stochastic differential equation (SDE) governing the diffusion process.\nSubsequently, k-space data can be interpolated by executing the diffusion\nprocess. This innovative approach highlights the optimization model's role in\ndesigning the SDE in diffusion models, enabling the diffusion process to align\nclosely with the physics inherent in the optimization model, a concept referred\nto as model-driven diffusion. We evaluated the proposed SPIRiT-Diffusion method\nusing a 3D joint intracranial and carotid vessel wall imaging dataset. The\nresults convincingly demonstrate its superiority over image-domain\nreconstruction methods, achieving high reconstruction quality even at a\nsubstantial acceleration rate of 10.\n", "date": "2023-04-11", "categories": []}
{"id": "2404.13102", "title": "Single-sample image-fusion upsampling of fluorescence lifetime images", "abstract": "  Fluorescence lifetime imaging microscopy (FLIM) provides detailed information\nabout molecular interactions and biological processes. A major bottleneck for\nFLIM is image resolution at high acquisition speeds, due to the engineering and\nsignal-processing limitations of time-resolved imaging technology. Here we\npresent single-sample image-fusion upsampling (SiSIFUS), a data-fusion approach\nto computational FLIM super-resolution that combines measurements from a\nlow-resolution time-resolved detector (that measures photon arrival time) and a\nhigh-resolution camera (that measures intensity only). To solve this otherwise\nill-posed inverse retrieval problem, we introduce statistically informed priors\nthat encode local and global dependencies between the two single-sample\nmeasurements. This bypasses the risk of out-of-distribution hallucination as in\ntraditional data-driven approaches and delivers enhanced images compared for\nexample to standard bilinear interpolation. The general approach laid out by\nSiSIFUS can be applied to other image super-resolution problems where two\ndifferent datasets are available.\n", "date": "2024-04-19", "categories": []}
{"id": "2404.13159", "title": "Equivariant Imaging for Self-supervised Hyperspectral Image Inpainting", "abstract": "  Hyperspectral imaging (HSI) is a key technology for earth observation,\nsurveillance, medical imaging and diagnostics, astronomy and space exploration.\nThe conventional technology for HSI in remote sensing applications is based on\nthe push-broom scanning approach in which the camera records the spectral image\nof a stripe of the scene at a time, while the image is generated by the\naggregation of measurements through time. In real-world airborne and spaceborne\nHSI instruments, some empty stripes would appear at certain locations, because\nplatforms do not always maintain a constant programmed attitude, or have access\nto accurate digital elevation maps (DEM), and the travelling track is not\nnecessarily aligned with the hyperspectral cameras at all times. This makes the\nenhancement of the acquired HS images from incomplete or corrupted observations\nan essential task. We introduce a novel HSI inpainting algorithm here, called\nHyperspectral Equivariant Imaging (Hyper-EI). Hyper-EI is a self-supervised\nlearning-based method which does not require training on extensive datasets or\naccess to a pre-trained model. Experimental results show that the proposed\nmethod achieves state-of-the-art inpainting performance compared to the\nexisting methods.\n", "date": "2024-04-19", "categories": []}
{"id": "2404.14132", "title": "CRNet: A Detail-Preserving Network for Unified Image Restoration and\n  Enhancement Task", "abstract": "  In real-world scenarios, images captured often suffer from blurring, noise,\nand other forms of image degradation, and due to sensor limitations, people\nusually can only obtain low dynamic range images. To achieve high-quality\nimages, researchers have attempted various image restoration and enhancement\noperations on photographs, including denoising, deblurring, and high dynamic\nrange imaging. However, merely performing a single type of image enhancement\nstill cannot yield satisfactory images. In this paper, to deal with the\nchallenge above, we propose the Composite Refinement Network (CRNet) to address\nthis issue using multiple exposure images. By fully integrating\ninformation-rich multiple exposure inputs, CRNet can perform unified image\nrestoration and enhancement. To improve the quality of image details, CRNet\nexplicitly separates and strengthens high and low-frequency information through\npooling layers, using specially designed Multi-Branch Blocks for effective\nfusion of these frequencies. To increase the receptive field and fully\nintegrate input features, CRNet employs the High-Frequency Enhancement Module,\nwhich includes large kernel convolutions and an inverted bottleneck ConvFFN.\nOur model secured third place in the first track of the Bracketing Image\nRestoration and Enhancement Challenge, surpassing previous SOTA models in both\ntesting metrics and visual quality.\n", "date": "2024-04-22", "categories": []}
{"id": "2404.11884", "title": "Seeing Motion at Nighttime with an Event Camera", "abstract": "  We focus on a very challenging task: imaging at nighttime dynamic scenes.\nMost previous methods rely on the low-light enhancement of a conventional RGB\ncamera. However, they would inevitably face a dilemma between the long exposure\ntime of nighttime and the motion blur of dynamic scenes. Event cameras react to\ndynamic changes with higher temporal resolution (microsecond) and higher\ndynamic range (120dB), offering an alternative solution. In this work, we\npresent a novel nighttime dynamic imaging method with an event camera.\nSpecifically, we discover that the event at nighttime exhibits temporal\ntrailing characteristics and spatial non-stationary distribution. Consequently,\nwe propose a nighttime event reconstruction network (NER-Net) which mainly\nincludes a learnable event timestamps calibration module (LETC) to align the\ntemporal trailing events and a non-uniform illumination aware module (NIAM) to\nstabilize the spatiotemporal distribution of events. Moreover, we construct a\npaired real low-light event dataset (RLED) through a co-axial imaging system,\nincluding 64,200 spatially and temporally aligned image GTs and low-light\nevents. Extensive experiments demonstrate that the proposed method outperforms\nstate-of-the-art methods in terms of visual quality and generalization ability\non real-world nighttime datasets. The project are available at:\nhttps://github.com/Liu-haoyue/NER-Net.\n", "date": "2024-04-18", "categories": []}
{"id": "2404.12142", "title": "SDIP: Self-Reinforcement Deep Image Prior Framework for Image Processing", "abstract": "  Deep image prior (DIP) proposed in recent research has revealed the inherent\ntrait of convolutional neural networks (CNN) for capturing substantial\nlow-level image statistics priors. This framework efficiently addresses the\ninverse problems in image processing and has induced extensive applications in\nvarious domains. However, as the whole algorithm is initialized randomly, the\nDIP algorithm often lacks stability. Thus, this method still has space for\nfurther improvement. In this paper, we propose the self-reinforcement deep\nimage prior (SDIP) as an improved version of the original DIP. We observed that\nthe changes in the DIP networks' input and output are highly correlated during\neach iteration. SDIP efficiently utilizes this trait in a reinforcement\nlearning manner, where the current iteration's output is utilized by a steering\nalgorithm to update the network input for the next iteration, guiding the\nalgorithm toward improved results. Experimental results across multiple\napplications demonstrate that our proposed SDIP framework offers improvement\ncompared to the original DIP method and other state-of-the-art methods.\n", "date": "2024-04-17", "categories": []}
{"id": "2308.13072", "title": "Full-dose Whole-body PET Synthesis from Low-dose PET Using\n  High-efficiency Denoising Diffusion Probabilistic Model: PET Consistency\n  Model", "abstract": "  Objective: Positron Emission Tomography (PET) has been a commonly used\nimaging modality in broad clinical applications. One of the most important\ntradeoffs in PET imaging is between image quality and radiation dose: high\nimage quality comes with high radiation exposure. Improving image quality is\ndesirable for all clinical applications while minimizing radiation exposure is\nneeded to reduce risk to patients. Approach: We introduce PET Consistency Model\n(PET-CM), an efficient diffusion-based method for generating high-quality\nfull-dose PET images from low-dose PET images. It employs a two-step process,\nadding Gaussian noise to full-dose PET images in the forward diffusion, and\nthen denoising them using a PET Shifted-window Vision Transformer (PET-VIT)\nnetwork in the reverse diffusion. The PET-VIT network learns a consistency\nfunction that enables direct denoising of Gaussian noise into clean full-dose\nPET images. PET-CM achieves state-of-the-art image quality while requiring\nsignificantly less computation time than other methods. Results: In experiments\ncomparing eighth-dose to full-dose images, PET-CM demonstrated impressive\nperformance with NMAE of 1.278+/-0.122%, PSNR of 33.783+/-0.824dB, SSIM of\n0.964+/-0.009, NCC of 0.968+/-0.011, HRS of 4.543, and SUV Error of\n0.255+/-0.318%, with an average generation time of 62 seconds per patient. This\nis a significant improvement compared to the state-of-the-art diffusion-based\nmodel with PET-CM reaching this result 12x faster. Similarly, in the\nquarter-dose to full-dose image experiments, PET-CM delivered competitive\noutcomes, achieving an NMAE of 0.973+/-0.066%, PSNR of 36.172+/-0.801dB, SSIM\nof 0.984+/-0.004, NCC of 0.990+/-0.005, HRS of 4.428, and SUV Error of\n0.151+/-0.192% using the same generation process, which underlining its high\nquantitative and clinical precision in both denoising scenario.\n", "date": "2023-08-24", "categories": []}
{"id": "2305.11443", "title": "Equivariant Multi-Modality Image Fusion", "abstract": "  Multi-modality image fusion is a technique that combines information from\ndifferent sensors or modalities, enabling the fused image to retain\ncomplementary features from each modality, such as functional highlights and\ntexture details. However, effective training of such fusion models is\nchallenging due to the scarcity of ground truth fusion data. To tackle this\nissue, we propose the Equivariant Multi-Modality imAge fusion (EMMA) paradigm\nfor end-to-end self-supervised learning. Our approach is rooted in the prior\nknowledge that natural imaging responses are equivariant to certain\ntransformations. Consequently, we introduce a novel training paradigm that\nencompasses a fusion module, a pseudo-sensing module, and an equivariant fusion\nmodule. These components enable the net training to follow the principles of\nthe natural sensing-imaging process while satisfying the equivariant imaging\nprior. Extensive experiments confirm that EMMA yields high-quality fusion\nresults for infrared-visible and medical images, concurrently facilitating\ndownstream multi-modal segmentation and detection tasks. The code is available\nat https://github.com/Zhaozixiang1228/MMIF-EMMA.\n", "date": "2023-05-19", "categories": []}
{"id": "2307.08123", "title": "Solving Inverse Problems with Latent Diffusion Models via Hard Data\n  Consistency", "abstract": "  Diffusion models have recently emerged as powerful generative priors for\nsolving inverse problems. However, training diffusion models in the pixel space\nare both data-intensive and computationally demanding, which restricts their\napplicability as priors for high-dimensional real-world data such as medical\nimages. Latent diffusion models, which operate in a much lower-dimensional\nspace, offer a solution to these challenges. However, incorporating latent\ndiffusion models to solve inverse problems remains a challenging problem due to\nthe nonlinearity of the encoder and decoder. To address these issues, we\npropose \\textit{ReSample}, an algorithm that can solve general inverse problems\nwith pre-trained latent diffusion models. Our algorithm incorporates data\nconsistency by solving an optimization problem during the reverse sampling\nprocess, a concept that we term as hard data consistency. Upon solving this\noptimization problem, we propose a novel resampling scheme to map the\nmeasurement-consistent sample back onto the noisy data manifold and\ntheoretically demonstrate its benefits. Lastly, we apply our algorithm to solve\na wide range of linear and nonlinear inverse problems in both natural and\nmedical images, demonstrating that our approach outperforms existing\nstate-of-the-art approaches, including those based on pixel-space diffusion\nmodels.\n", "date": "2023-07-16", "categories": []}
{"id": "2404.10342", "title": "Referring Flexible Image Restoration", "abstract": "  In reality, images often exhibit multiple degradations, such as rain and fog\nat night (triple degradations). However, in many cases, individuals may not\nwant to remove all degradations, for instance, a blurry lens revealing a\nbeautiful snowy landscape (double degradations). In such scenarios, people may\nonly desire to deblur. These situations and requirements shed light on a new\nchallenge in image restoration, where a model must perceive and remove specific\ndegradation types specified by human commands in images with multiple\ndegradations. We term this task Referring Flexible Image Restoration (RFIR). To\naddress this, we first construct a large-scale synthetic dataset called RFIR,\ncomprising 153,423 samples with the degraded image, text prompt for specific\ndegradation removal and restored image. RFIR consists of five basic degradation\ntypes: blur, rain, haze, low light and snow while six main sub-categories are\nincluded for varying degrees of degradation removal. To tackle the challenge,\nwe propose a novel transformer-based multi-task model named TransRFIR, which\nsimultaneously perceives degradation types in the degraded image and removes\nspecific degradation upon text prompt. TransRFIR is based on two devised\nattention modules, Multi-Head Agent Self-Attention (MHASA) and Multi-Head Agent\nCross Attention (MHACA), where MHASA and MHACA introduce the agent token and\nreach the linear complexity, achieving lower computation cost than vanilla\nself-attention and cross-attention and obtaining competitive performances. Our\nTransRFIR achieves state-of-the-art performances compared with other\ncounterparts and is proven as an effective architecture for image restoration.\nWe release our project at https://github.com/GuanRunwei/FIR-CP.\n", "date": "2024-04-16", "categories": []}
{"id": "2404.10358", "title": "Improving Bracket Image Restoration and Enhancement with Flow-guided\n  Alignment and Enhanced Feature Aggregation", "abstract": "  In this paper, we address the Bracket Image Restoration and Enhancement\n(BracketIRE) task using a novel framework, which requires restoring a\nhigh-quality high dynamic range (HDR) image from a sequence of noisy, blurred,\nand low dynamic range (LDR) multi-exposure RAW inputs. To overcome this\nchallenge, we present the IREANet, which improves the multiple exposure\nalignment and aggregation with a Flow-guide Feature Alignment Module (FFAM) and\nan Enhanced Feature Aggregation Module (EFAM). Specifically, the proposed FFAM\nincorporates the inter-frame optical flow as guidance to facilitate the\ndeformable alignment and spatial attention modules for better feature\nalignment. The EFAM further employs the proposed Enhanced Residual Block (ERB)\nas a foundational component, wherein a unidirectional recurrent network\naggregates the aligned temporal features to better reconstruct the results. To\nimprove model generalization and performance, we additionally employ the Bayer\npreserving augmentation (BayerAug) strategy to augment the multi-exposure RAW\ninputs. Our experimental evaluations demonstrate that the proposed IREANet\nshows state-of-the-art performance compared with previous methods.\n", "date": "2024-04-16", "categories": []}
{"id": "2404.10689", "title": "Network architecture search of X-ray based scientific applications", "abstract": "  X-ray and electron diffraction-based microscopy use bragg peak detection and\nptychography to perform 3-D imaging at an atomic resolution. Typically, these\ntechniques are implemented using computationally complex tasks such as a\nPsuedo-Voigt function or solving a complex inverse problem. Recently, the use\nof deep neural networks has improved the existing state-of-the-art approaches.\nHowever, the design and development of the neural network models depends on\ntime and labor intensive tuning of the model by application experts. To that\nend, we propose a hyperparameter (HPS) and neural architecture search (NAS)\napproach to automate the design and optimization of the neural network models\nfor model size, energy consumption and throughput. We demonstrate the improved\nperformance of the auto-tuned models when compared to the manually tuned\nBraggNN and PtychoNN benchmark. We study and demonstrate the importance of the\nexploring the search space of tunable hyperparameters in enhancing the\nperformance of bragg peak detection and ptychographic reconstruction. Our NAS\nand HPS of (1) BraggNN achieves a 31.03\\% improvement in bragg peak detection\naccuracy with a 87.57\\% reduction in model size, and (2) PtychoNN achieves a\n16.77\\% improvement in model accuracy and a 12.82\\% reduction in model size\nwhen compared to the baseline PtychoNN model. When inferred on the Orin-AGX\nplatform, the optimized Braggnn and Ptychonn models demonstrate a 10.51\\% and\n9.47\\% reduction in inference latency and a 44.18\\% and 15.34\\% reduction in\nenergy consumption when compared to their respective baselines, when inferred\nin the Orin-AGX edge platform.\n", "date": "2024-04-16", "categories": []}
{"id": "2308.11949", "title": "High-quality Image Dehazing with Diffusion Model", "abstract": "  Image dehazing is quite challenging in dense-haze scenarios, where quite less\noriginal information remains in the hazy image. Though previous methods have\nmade marvelous progress, they still suffer from information loss in content and\ncolor in dense-haze scenarios. The recently emerged Denoising Diffusion\nProbabilistic Model (DDPM) exhibits strong generation ability, showing\npotential for solving this problem. However, DDPM fails to consider the physics\nproperty of dehazing task, limiting its information completion capacity. In\nthis work, we propose DehazeDDPM: A DDPM-based and physics-aware image dehazing\nframework that applies to complex hazy scenarios. Specifically, DehazeDDPM\nworks in two stages. The former stage physically models the dehazing task with\nthe Atmospheric Scattering Model (ASM), pulling the distribution closer to the\nclear data and endowing DehazeDDPM with fog-aware ability. The latter stage\nexploits the strong generation ability of DDPM to compensate for the\nhaze-induced huge information loss, by working in conjunction with the physical\nmodelling. Extensive experiments demonstrate that our method attains\nstate-of-the-art performance on both synthetic and real-world hazy datasets.\n", "date": "2023-08-23", "categories": []}
{"id": "2312.13313", "title": "ParamISP: Learned Forward and Inverse ISPs using Camera Parameters", "abstract": "  RAW images are rarely shared mainly due to its excessive data size compared\nto their sRGB counterparts obtained by camera ISPs. Learning the forward and\ninverse processes of camera ISPs has been recently demonstrated, enabling\nphysically-meaningful RAW-level image processing on input sRGB images. However,\nexisting learning-based ISP methods fail to handle the large variations in the\nISP processes with respect to camera parameters such as ISO and exposure time,\nand have limitations when used for various applications. In this paper, we\npropose ParamISP, a learning-based method for forward and inverse conversion\nbetween sRGB and RAW images, that adopts a novel neural-network module to\nutilize camera parameters, which is dubbed as ParamNet. Given the camera\nparameters provided in the EXIF data, ParamNet converts them into a feature\nvector to control the ISP networks. Extensive experiments demonstrate that\nParamISP achieve superior RAW and sRGB reconstruction results compared to\nprevious methods and it can be effectively used for a variety of applications\nsuch as deblurring dataset synthesis, raw deblurring, HDR reconstruction, and\ncamera-to-camera transfer.\n", "date": "2023-12-20", "categories": []}
{"id": "2312.16519", "title": "Image Restoration by Denoising Diffusion Models with Iteratively\n  Preconditioned Guidance", "abstract": "  Training deep neural networks has become a common approach for addressing\nimage restoration problems. An alternative for training a \"task-specific\"\nnetwork for each observation model is to use pretrained deep denoisers for\nimposing only the signal's prior within iterative algorithms, without\nadditional training. Recently, a sampling-based variant of this approach has\nbecome popular with the rise of diffusion/score-based generative models. Using\ndenoisers for general purpose restoration requires guiding the iterations to\nensure agreement of the signal with the observations. In low-noise settings,\nguidance that is based on back-projection (BP) has been shown to be a promising\nstrategy (used recently also under the names \"pseudoinverse\" or\n\"range/null-space\" guidance). However, the presence of noise in the\nobservations hinders the gains from this approach. In this paper, we propose a\nnovel guidance technique, based on preconditioning that allows traversing from\nBP-based guidance to least squares based guidance along the restoration scheme.\nThe proposed approach is robust to noise while still having much simpler\nimplementation than alternative methods (e.g., it does not require SVD or a\nlarge number of iterations). We use it within both an optimization scheme and a\nsampling-based scheme, and demonstrate its advantages over existing methods for\nimage deblurring and super-resolution.\n", "date": "2023-12-27", "categories": []}
{"id": "2404.09389", "title": "Masked and Shuffled Blind Spot Denoising for Real-World Images", "abstract": "  We introduce a novel approach to single image denoising based on the Blind\nSpot Denoising principle, which we call MAsked and SHuffled Blind Spot\nDenoising (MASH). We focus on the case of correlated noise, which often plagues\nreal images. MASH is the result of a careful analysis to determine the\nrelationships between the level of blindness (masking) of the input and the\n(unknown) noise correlation. Moreover, we introduce a shuffling technique to\nweaken the local correlation of noise, which in turn yields an additional\ndenoising performance improvement. We evaluate MASH via extensive experiments\non real-world noisy image datasets. We demonstrate on par or better results\ncompared to existing self-supervised denoising methods.\n", "date": "2024-04-15", "categories": []}
{"id": "2404.08350", "title": "Self-Supervised k-Space Regularization for Motion-Resolved Abdominal MRI\n  Using Neural Implicit k-Space Representation", "abstract": "  Neural implicit k-space representations have shown promising results for\ndynamic MRI at high temporal resolutions. Yet, their exclusive training in\nk-space limits the application of common image regularization methods to\nimprove the final reconstruction. In this work, we introduce the concept of\nparallel imaging-inspired self-consistency (PISCO), which we incorporate as\nnovel self-supervised k-space regularization enforcing a consistent\nneighborhood relationship. At no additional data cost, the proposed\nregularization significantly improves neural implicit k-space reconstructions\non simulated data. Abdominal in-vivo reconstructions using PISCO result in\nenhanced spatio-temporal image quality compared to state-of-the-art methods.\nCode is available at https://github.com/vjspi/PISCO-NIK.\n", "date": "2024-04-12", "categories": []}
{"id": "2404.07551", "title": "Event-Enhanced Snapshot Compressive Videography at 10K FPS", "abstract": "  Video snapshot compressive imaging (SCI) encodes the target dynamic scene\ncompactly into a snapshot and reconstructs its high-speed frame sequence\nafterward, greatly reducing the required data footprint and transmission\nbandwidth as well as enabling high-speed imaging with a low frame rate\nintensity camera. In implementation, high-speed dynamics are encoded via\ntemporally varying patterns, and only frames at corresponding temporal\nintervals can be reconstructed, while the dynamics occurring between\nconsecutive frames are lost. To unlock the potential of conventional snapshot\ncompressive videography, we propose a novel hybrid \"intensity+event\" imaging\nscheme by incorporating an event camera into a video SCI setup. Our proposed\nsystem consists of a dual-path optical setup to record the coded intensity\nmeasurement and intermediate event signals simultaneously, which is compact and\nphoton-efficient by collecting the half photons discarded in conventional video\nSCI. Correspondingly, we developed a dual-branch Transformer utilizing the\nreciprocal relationship between two data modes to decode dense video frames.\nExtensive experiments on both simulated and real-captured data demonstrate our\nsuperiority to state-of-the-art video SCI and video frame interpolation (VFI)\nmethods. Benefiting from the new hybrid design leveraging both intrinsic\nredundancy in videos and the unique feature of event cameras, we achieve\nhigh-quality videography at 0.1ms time intervals with a low-cost CMOS image\nsensor working at 24 FPS.\n", "date": "2024-04-11", "categories": []}
{"id": "2404.07985", "title": "WaveMo: Learning Wavefront Modulations to See Through Scattering", "abstract": "  Imaging through scattering media is a fundamental and pervasive challenge in\nfields ranging from medical diagnostics to astronomy. A promising strategy to\novercome this challenge is wavefront modulation, which induces measurement\ndiversity during image acquisition. Despite its importance, designing optimal\nwavefront modulations to image through scattering remains under-explored. This\npaper introduces a novel learning-based framework to address the gap. Our\napproach jointly optimizes wavefront modulations and a computationally\nlightweight feedforward \"proxy\" reconstruction network. This network is trained\nto recover scenes obscured by scattering, using measurements that are modified\nby these modulations. The learned modulations produced by our framework\ngeneralize effectively to unseen scattering scenarios and exhibit remarkable\nversatility. During deployment, the learned modulations can be decoupled from\nthe proxy network to augment other more computationally expensive restoration\nalgorithms. Through extensive experiments, we demonstrate our approach\nsignificantly advances the state of the art in imaging through scattering\nmedia. Our project webpage is at https://wavemo-2024.github.io/.\n", "date": "2024-04-11", "categories": []}
{"id": "2311.10568", "title": "Phase Guided Light Field for Spatial-Depth High Resolution 3D Imaging", "abstract": "  On 3D imaging, light field cameras typically are of single shot, and however,\nthey heavily suffer from low spatial resolution and depth accuracy. In this\npaper, by employing an optical projector to project a group of single\nhigh-frequency phase-shifted sinusoid patterns, we propose a phase guided light\nfield algorithm to significantly improve both the spatial and depth resolutions\nfor off-the-shelf light field cameras. First, for correcting the axial\naberrations caused by the main lens of our light field camera, we propose a\ndeformed cone model to calibrate our structured light field system. Second,\nover wrapped phases computed from patterned images, we propose a stereo\nmatching algorithm, i.e. phase guided sum of absolute difference, to robustly\nobtain the correspondence for each pair of neighbored two lenslets. Finally, by\nintroducing a virtual camera according to the basic geometrical optics of light\nfield imaging, we propose a reorganization strategy to reconstruct 3D point\nclouds with spatial-depth high resolution. Experimental results show that,\ncompared with the state-of-the-art active light field methods, the proposed\nreconstructs 3D point clouds with a spatial resolution of 1280$\\times$720 with\nfactors 10$\\times$ increased, while maintaining the same high depth resolution\nand needing merely a single group of high-frequency patterns.\n", "date": "2023-11-17", "categories": []}
{"id": "2404.01563", "title": "Two-Phase Multi-Dose-Level PET Image Reconstruction with Dose Level\n  Awareness", "abstract": "  To obtain high-quality positron emission tomography (PET) while minimizing\nradiation exposure, a range of methods have been designed to reconstruct\nstandard-dose PET (SPET) from corresponding low-dose PET (LPET) images.\nHowever, most current methods merely learn the mapping between\nsingle-dose-level LPET and SPET images, but omit the dose disparity of LPET\nimages in clinical scenarios. In this paper, to reconstruct high-quality SPET\nimages from multi-dose-level LPET images, we design a novel two-phase\nmulti-dose-level PET reconstruction algorithm with dose level awareness,\ncontaining a pre-training phase and a SPET prediction phase. Specifically, the\npre-training phase is devised to explore both fine-grained discriminative\nfeatures and effective semantic representation. The SPET prediction phase\nadopts a coarse prediction network utilizing pre-learned dose level prior to\ngenerate preliminary result, and a refinement network to precisely preserve the\ndetails. Experiments on MICCAI 2022 Ultra-low Dose PET Imaging Challenge\nDataset have demonstrated the superiority of our method.\n", "date": "2024-04-02", "categories": []}
{"id": "2404.04586", "title": "PIE: Physics-inspired Low-light Enhancement", "abstract": "  In this paper, we propose a physics-inspired contrastive learning paradigm\nfor low-light enhancement, called PIE. PIE primarily addresses three issues:\n(i) To resolve the problem of existing learning-based methods often training a\nLLE model with strict pixel-correspondence image pairs, we eliminate the need\nfor pixel-correspondence paired training data and instead train with unpaired\nimages. (ii) To address the disregard for negative samples and the inadequacy\nof their generation in existing methods, we incorporate physics-inspired\ncontrastive learning for LLE and design the Bag of Curves (BoC) method to\ngenerate more reasonable negative samples that closely adhere to the underlying\nphysical imaging principle. (iii) To overcome the reliance on semantic ground\ntruths in existing methods, we propose an unsupervised regional segmentation\nmodule, ensuring regional brightness consistency while eliminating the\ndependency on semantic ground truths. Overall, the proposed PIE can effectively\nlearn from unpaired positive/negative samples and smoothly realize non-semantic\nregional enhancement, which is clearly different from existing LLE efforts.\nBesides the novel architecture of PIE, we explore the gain of PIE on downstream\ntasks such as semantic segmentation and face detection. Training on readily\navailable open data and extensive experiments demonstrate that our method\nsurpasses the state-of-the-art LLE models over six independent cross-scenes\ndatasets. PIE runs fast with reasonable GFLOPs in test time, making it easy to\nuse on mobile devices.\n", "date": "2024-04-06", "categories": []}
{"id": "2403.10376", "title": "PASTA: Towards Flexible and Efficient HDR Imaging Via Progressively\n  Aggregated Spatio-Temporal Alignment", "abstract": "  Leveraging Transformer attention has led to great advancements in HDR\ndeghosting. However, the intricate nature of self-attention introduces\npractical challenges, as existing state-of-the-art methods often demand\nhigh-end GPUs or exhibit slow inference speeds, especially for high-resolution\nimages like 2K. Striking an optimal balance between performance and latency\nremains a critical concern. In response, this work presents PASTA, a novel\nProgressively Aggregated Spatio-Temporal Alignment framework for HDR\ndeghosting. Our approach achieves effectiveness and efficiency by harnessing\nhierarchical representation during feature distanglement. Through the\nutilization of diverse granularities within the hierarchical structure, our\nmethod substantially boosts computational speed and optimizes the HDR imaging\nworkflow. In addition, we explore within-scale feature modeling with local and\nglobal attention, gradually merging and refining them in a coarse-to-fine\nfashion. Experimental results showcase PASTA's superiority over current SOTA\nmethods in both visual quality and performance metrics, accompanied by a\nsubstantial 3-fold (x3) increase in inference speed.\n", "date": "2024-03-15", "categories": []}
{"id": "2209.03737", "title": "CGAN-ECT: Tomography Image Reconstruction from Electrical Capacitance\n  Measurements Using CGANs", "abstract": "  Due to the rapid growth of Electrical Capacitance Tomography (ECT)\napplications in several industrial fields, there is a crucial need for\ndeveloping high quality, yet fast, methodologies of image reconstruction from\nraw capacitance measurements. Deep learning, as an effective non-linear mapping\ntool for complicated functions, has been going viral in many fields including\nelectrical tomography. In this paper, we propose a Conditional Generative\nAdversarial Network (CGAN) model for reconstructing ECT images from capacitance\nmeasurements. The initial image of the CGAN model is constructed from the\ncapacitance measurement. To our knowledge, this is the first time to represent\nthe capacitance measurements in an image form. We have created a new massive\nECT dataset of 320K synthetic image measurements pairs for training, and\ntesting the proposed model. The feasibility and generalization ability of the\nproposed CGAN-ECT model are evaluated using testing dataset, contaminated data\nand flow patterns that are not exposed to the model during the training phase.\nThe evaluation results prove that the proposed CGAN-ECT model can efficiently\ncreate more accurate ECT images than traditional and other deep learning-based\nimage reconstruction algorithms. CGAN-ECT achieved an average image correlation\ncoefficient of more than 99.3% and an average relative image error about 0.07.\n", "date": "2022-09-07", "categories": []}
{"id": "2311.11825", "title": "Holistic Inverse Rendering of Complex Facade via Aerial 3D Scanning", "abstract": "  In this work, we use multi-view aerial images to reconstruct the geometry,\nlighting, and material of facades using neural signed distance fields (SDFs).\nWithout the requirement of complex equipment, our method only takes simple RGB\nimages captured by a drone as inputs to enable physically based and\nphotorealistic novel-view rendering, relighting, and editing. However, a\nreal-world facade usually has complex appearances ranging from diffuse rocks\nwith subtle details to large-area glass windows with specular reflections,\nmaking it hard to attend to everything. As a result, previous methods can\npreserve the geometry details but fail to reconstruct smooth glass windows or\nverse vise. In order to address this challenge, we introduce three spatial- and\nsemantic-adaptive optimization strategies, including a semantic regularization\napproach based on zero-shot segmentation techniques to improve material\nconsistency, a frequency-aware geometry regularization to balance surface\nsmoothness and details in different surfaces, and a visibility probe-based\nscheme to enable efficient modeling of the local lighting in large-scale\noutdoor environments. In addition, we capture a real-world facade aerial 3D\nscanning image set and corresponding point clouds for training and\nbenchmarking. The experiment demonstrates the superior quality of our method on\nfacade holistic inverse rendering, novel view synthesis, and scene editing\ncompared to state-of-the-art baselines.\n", "date": "2023-11-20", "categories": []}
{"id": "2401.04244", "title": "Spatio-Temporal Turbulence Mitigation: A Translational Perspective", "abstract": "  Recovering images distorted by atmospheric turbulence is a challenging\ninverse problem due to the stochastic nature of turbulence. Although numerous\nturbulence mitigation (TM) algorithms have been proposed, their efficiency and\ngeneralization to real-world dynamic scenarios remain severely limited.\nBuilding upon the intuitions of classical TM algorithms, we present the Deep\nAtmospheric TUrbulence Mitigation network (DATUM). DATUM aims to overcome major\nchallenges when transitioning from classical to deep learning approaches. By\ncarefully integrating the merits of classical multi-frame TM methods into a\ndeep network structure, we demonstrate that DATUM can efficiently perform\nlong-range temporal aggregation using a recurrent fashion, while deformable\nattention and temporal-channel attention seamlessly facilitate pixel\nregistration and lucky imaging. With additional supervision, tilt and blur\ndegradation can be jointly mitigated. These inductive biases empower DATUM to\nsignificantly outperform existing methods while delivering a tenfold increase\nin processing speed. A large-scale training dataset, ATSyn, is presented as a\nco-invention to enable generalization in real turbulence. Our code and datasets\nare available at https://xg416.github.io/DATUM.\n", "date": "2024-01-08", "categories": []}
{"id": "2403.02449", "title": "Optimizing Illuminant Estimation in Dual-Exposure HDR Imaging", "abstract": "  High dynamic range (HDR) imaging involves capturing a series of frames of the\nsame scene, each with different exposure settings, to broaden the dynamic range\nof light. This can be achieved through burst capturing or using staggered HDR\nsensors that capture long and short exposures simultaneously in the camera\nimage signal processor (ISP). Within camera ISP pipeline, illuminant estimation\nis a crucial step aiming to estimate the color of the global illuminant in the\nscene. This estimation is used in camera ISP white-balance module to remove\nundesirable color cast in the final image. Despite the multiple frames captured\nin the HDR pipeline, conventional illuminant estimation methods often rely only\non a single frame of the scene. In this paper, we explore leveraging\ninformation from frames captured with different exposure times. Specifically,\nwe introduce a simple feature extracted from dual-exposure images to guide\nilluminant estimators, referred to as the dual-exposure feature (DEF). To\nvalidate the efficiency of DEF, we employed two illuminant estimators using the\nproposed DEF: 1) a multilayer perceptron network (MLP), referred to as\nexposure-based MLP (EMLP), and 2) a modified version of the convolutional color\nconstancy (CCC) to integrate our DEF, that we call ECCC. Both EMLP and ECCC\nachieve promising results, in some cases surpassing prior methods that require\nhundreds of thousands or millions of parameters, with only a few hundred\nparameters for EMLP and a few thousand parameters for ECCC.\n", "date": "2024-03-04", "categories": []}
{"id": "2404.05001", "title": "Dual-Scale Transformer for Large-Scale Single-Pixel Imaging", "abstract": "  Single-pixel imaging (SPI) is a potential computational imaging technique\nwhich produces image by solving an illposed reconstruction problem from few\nmeasurements captured by a single-pixel detector. Deep learning has achieved\nimpressive success on SPI reconstruction. However, previous poor reconstruction\nperformance and impractical imaging model limit its real-world applications. In\nthis paper, we propose a deep unfolding network with hybrid-attention\nTransformer on Kronecker SPI model, dubbed HATNet, to improve the imaging\nquality of real SPI cameras. Specifically, we unfold the computation graph of\nthe iterative shrinkagethresholding algorithm (ISTA) into two alternative\nmodules: efficient tensor gradient descent and hybrid-attention multiscale\ndenoising. By virtue of Kronecker SPI, the gradient descent module can avoid\nhigh computational overheads rooted in previous gradient descent modules based\non vectorized SPI. The denoising module is an encoder-decoder architecture\npowered by dual-scale spatial attention for high- and low-frequency aggregation\nand channel attention for global information recalibration. Moreover, we build\na SPI prototype to verify the effectiveness of the proposed method. Extensive\nexperiments on synthetic and real data demonstrate that our method achieves the\nstate-of-the-art performance. The source code and pre-trained models are\navailable at https://github.com/Gang-Qu/HATNet-SPI.\n", "date": "2024-04-07", "categories": []}
{"id": "2404.05024", "title": "PathFinder: Attention-Driven Dynamic Non-Line-of-Sight Tracking with a\n  Mobile Robot", "abstract": "  The study of non-line-of-sight (NLOS) imaging is growing due to its many\npotential applications, including rescue operations and pedestrian detection by\nself-driving cars. However, implementing NLOS imaging on a moving camera\nremains an open area of research. Existing NLOS imaging methods rely on\ntime-resolved detectors and laser configurations that require precise optical\nalignment, making it difficult to deploy them in dynamic environments. This\nwork proposes a data-driven approach to NLOS imaging, PathFinder, that can be\nused with a standard RGB camera mounted on a small, power-constrained mobile\nrobot, such as an aerial drone. Our experimental pipeline is designed to\naccurately estimate the 2D trajectory of a person who moves in a\nManhattan-world environment while remaining hidden from the camera's\nfield-of-view. We introduce a novel approach to process a sequence of dynamic\nsuccessive frames in a line-of-sight (LOS) video using an attention-based\nneural network that performs inference in real-time. The method also includes a\npreprocessing selection metric that analyzes images from a moving camera which\ncontain multiple vertical planar surfaces, such as walls and building facades,\nand extracts planes that return maximum NLOS information. We validate the\napproach on in-the-wild scenes using a drone for video capture, thus\ndemonstrating low-cost NLOS imaging in dynamic capture environments.\n", "date": "2024-04-07", "categories": []}
{"id": "2404.05274", "title": "Deep Optics for Video Snapshot Compressive Imaging", "abstract": "  Video snapshot compressive imaging (SCI) aims to capture a sequence of video\nframes with only a single shot of a 2D detector, whose backbones rest in\noptical modulation patterns (also known as masks) and a computational\nreconstruction algorithm. Advanced deep learning algorithms and mature hardware\nare putting video SCI into practical applications. Yet, there are two clouds in\nthe sunshine of SCI: i) low dynamic range as a victim of high temporal\nmultiplexing, and ii) existing deep learning algorithms' degradation on real\nsystem. To address these challenges, this paper presents a deep optics\nframework to jointly optimize masks and a reconstruction network. Specifically,\nwe first propose a new type of structural mask to realize motion-aware and\nfull-dynamic-range measurement. Considering the motion awareness property in\nmeasurement domain, we develop an efficient network for video SCI\nreconstruction using Transformer to capture long-term temporal dependencies,\ndubbed Res2former. Moreover, sensor response is introduced into the forward\nmodel of video SCI to guarantee end-to-end model training close to real system.\nFinally, we implement the learned structural masks on a digital micro-mirror\ndevice. Experimental results on synthetic and real data validate the\neffectiveness of the proposed framework. We believe this is a milestone for\nreal-world video SCI. The source code and data are available at\nhttps://github.com/pwangcs/DeepOpticsSCI.\n", "date": "2024-04-08", "categories": []}
{"id": "2312.14239", "title": "PlatoNeRF: 3D Reconstruction in Plato's Cave via Single-View Two-Bounce\n  Lidar", "abstract": "  3D reconstruction from a single-view is challenging because of the ambiguity\nfrom monocular cues and lack of information about occluded regions. Neural\nradiance fields (NeRF), while popular for view synthesis and 3D reconstruction,\nare typically reliant on multi-view images. Existing methods for single-view 3D\nreconstruction with NeRF rely on either data priors to hallucinate views of\noccluded regions, which may not be physically accurate, or shadows observed by\nRGB cameras, which are difficult to detect in ambient light and low albedo\nbackgrounds. We propose using time-of-flight data captured by a single-photon\navalanche diode to overcome these limitations. Our method models two-bounce\noptical paths with NeRF, using lidar transient data for supervision. By\nleveraging the advantages of both NeRF and two-bounce light measured by lidar,\nwe demonstrate that we can reconstruct visible and occluded geometry without\ndata priors or reliance on controlled ambient lighting or scene albedo. In\naddition, we demonstrate improved generalization under practical constraints on\nsensor spatial- and temporal-resolution. We believe our method is a promising\ndirection as single-photon lidars become ubiquitous on consumer devices, such\nas phones, tablets, and headsets.\n", "date": "2023-12-21", "categories": []}
{"id": "2402.12891", "title": "Mind the Exit Pupil Gap: Revisiting the Intrinsics of a Standard\n  Plenoptic Camera", "abstract": "  Among the common applications of plenoptic cameras are depth reconstruction\nand post-shot refocusing. These require a calibration relating the camera-side\nlight field to that of the scene. Numerous methods with this goal have been\ndeveloped based on thin lens models for the plenoptic camera's main lens and\nmicrolenses. Our work addresses the often-overlooked role of the main lens exit\npupil in these models and specifically in the decoding process of standard\nplenoptic camera (SPC) images. We formally deduce the connection between the\nrefocusing distance and the resampling parameter for the decoded light field\nand provide an analysis of the errors that arise when the exit pupil is not\nconsidered. In addition, previous work is revisited with respect to the exit\npupil's role and all theoretical results are validated through a\nray-tracing-based simulation. With the public release of the evaluated SPC\ndesigns alongside our simulation and experimental data we aim to contribute to\na more accurate and nuanced understanding of plenoptic camera optics.\n", "date": "2024-02-20", "categories": []}
{"id": "2404.04211", "title": "Robust Gaussian Splatting", "abstract": "  In this paper, we address common error sources for 3D Gaussian Splatting\n(3DGS) including blur, imperfect camera poses, and color inconsistencies, with\nthe goal of improving its robustness for practical applications like\nreconstructions from handheld phone captures. Our main contribution involves\nmodeling motion blur as a Gaussian distribution over camera poses, allowing us\nto address both camera pose refinement and motion blur correction in a unified\nway. Additionally, we propose mechanisms for defocus blur compensation and for\naddressing color in-consistencies caused by ambient light, shadows, or due to\ncamera-related factors like varying white balancing settings. Our proposed\nsolutions integrate in a seamless way with the 3DGS formulation while\nmaintaining its benefits in terms of training efficiency and rendering speed.\nWe experimentally validate our contributions on relevant benchmark datasets\nincluding Scannet++ and Deblur-NeRF, obtaining state-of-the-art results and\nthus consistent improvements over relevant baselines.\n", "date": "2024-04-05", "categories": []}
{"id": "2306.10482", "title": "Weighted structure tensor total variation for image denoising", "abstract": "  For image denoising problems, the structure tensor total variation\n(STV)-based models show good performances when compared with other competing\nregularization approaches. However, the STV regularizer does not couple the\nlocal information of the image and may not maintain the image details.\nTherefore, we employ the anisotropic weighted matrix introduced in the\nanisotropic total variation (ATV) model to improve the STV model. By applying\nthe weighted matrix to the discrete gradient of the patch-based Jacobian\noperator in STV, our proposed weighted STV (WSTV) model can effectively capture\nlocal information from images and maintain their details during the denoising\nprocess. The optimization problem in the model is solved by a fast first-order\ngradient projection algorithm with a complexity result of $O(1 / i^2)$. For\nimages with different Gaussian noise levels, the experimental results\ndemonstrate that the WSTV model can effectively improve the quality of restored\nimages compared to other TV and STV-based models.\n", "date": "2023-06-18", "categories": []}
{"id": "2401.13627", "title": "Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic\n  Image Restoration In the Wild", "abstract": "  We introduce SUPIR (Scaling-UP Image Restoration), a groundbreaking image\nrestoration method that harnesses generative prior and the power of model\nscaling up. Leveraging multi-modal techniques and advanced generative prior,\nSUPIR marks a significant advance in intelligent and realistic image\nrestoration. As a pivotal catalyst within SUPIR, model scaling dramatically\nenhances its capabilities and demonstrates new potential for image restoration.\nWe collect a dataset comprising 20 million high-resolution, high-quality images\nfor model training, each enriched with descriptive text annotations. SUPIR\nprovides the capability to restore images guided by textual prompts, broadening\nits application scope and potential. Moreover, we introduce negative-quality\nprompts to further improve perceptual quality. We also develop a\nrestoration-guided sampling method to suppress the fidelity issue encountered\nin generative-based restoration. Experiments demonstrate SUPIR's exceptional\nrestoration effects and its novel capacity to manipulate restoration through\ntextual prompts.\n", "date": "2024-01-24", "categories": []}
{"id": "2404.02387", "title": "An inversion problem for optical spectrum data via physics-guided\n  machine learning", "abstract": "  We propose the regularized recurrent inference machine (rRIM), a novel\nmachine-learning approach to solve the challenging problem of deriving the\npairing glue function from measured optical spectra. The rRIM incorporates\nphysical principles into both training and inference and affords noise\nrobustness, flexibility with out-of-distribution data, and reduced data\nrequirements. It effectively obtains reliable pairing glue functions from\nexperimental optical spectra and yields promising solutions for similar inverse\nproblems of the Fredholm integral equation of the first kind.\n", "date": "2024-04-03", "categories": []}
{"id": "2404.02460", "title": "TSNet:A Two-stage Network for Image Dehazing with Multi-scale Fusion and\n  Adaptive Learning", "abstract": "  Image dehazing has been a popular topic of research for a long time. Previous\ndeep learning-based image dehazing methods have failed to achieve satisfactory\ndehazing effects on both synthetic datasets and real-world datasets, exhibiting\npoor generalization. Moreover, single-stage networks often result in many\nregions with artifacts and color distortion in output images. To address these\nissues, this paper proposes a two-stage image dehazing network called TSNet,\nmainly consisting of the multi-scale fusion module (MSFM) and the adaptive\nlearning module (ALM). Specifically, MSFM and ALM enhance the generalization of\nTSNet. The MSFM can obtain large receptive fields at multiple scales and\nintegrate features at different frequencies to reduce the differences between\ninputs and learning objectives. The ALM can actively learn of regions of\ninterest in images and restore texture details more effectively. Additionally,\nTSNet is designed as a two-stage network, where the first-stage network\nperforms image dehazing, and the second-stage network is employed to improve\nissues such as artifacts and color distortion present in the results of the\nfirst-stage network. We also change the learning objective from ground truth\nimages to opposite fog maps, which improves the learning efficiency of TSNet.\nExtensive experiments demonstrate that TSNet exhibits superior dehazing\nperformance on both synthetic and real-world datasets compared to previous\nstate-of-the-art methods.\n", "date": "2024-04-03", "categories": []}
{"id": "2404.02731", "title": "Event Camera Demosaicing via Swin Transformer and Pixel-focus Loss", "abstract": "  Recent research has highlighted improvements in high-quality imaging guided\nby event cameras, with most of these efforts concentrating on the RGB domain.\nHowever, these advancements frequently neglect the unique challenges introduced\nby the inherent flaws in the sensor design of event cameras in the RAW domain.\nSpecifically, this sensor design results in the partial loss of pixel values,\nposing new challenges for RAW domain processes like demosaicing. The challenge\nintensifies as most research in the RAW domain is based on the premise that\neach pixel contains a value, making the straightforward adaptation of these\nmethods to event camera demosaicing problematic. To end this, we present a\nSwin-Transformer-based backbone and a pixel-focus loss function for demosaicing\nwith missing pixel values in RAW domain processing. Our core motivation is to\nrefine a general and widely applicable foundational model from the RGB domain\nfor RAW domain processing, thereby broadening the model's applicability within\nthe entire imaging process. Our method harnesses multi-scale processing and\nspace-to-depth techniques to ensure efficiency and reduce computing complexity.\nWe also proposed the Pixel-focus Loss function for network fine-tuning to\nimprove network convergence based on our discovery of a long-tailed\ndistribution in training loss. Our method has undergone validation on the MIPI\nDemosaic Challenge dataset, with subsequent analytical experimentation\nconfirming its efficacy. All code and trained models are released here:\nhttps://github.com/yunfanLu/ev-demosaic\n", "date": "2024-04-03", "categories": []}
{"id": "2308.00622", "title": "NeRT: Implicit Neural Representations for General Unsupervised\n  Turbulence Mitigation", "abstract": "  The atmospheric and water turbulence mitigation problems have emerged as\nchallenging inverse problems in computer vision and optics communities over the\nyears. However, current methods either rely heavily on the quality of the\ntraining dataset or fail to generalize over various scenarios, such as static\nscenes, dynamic scenes, and text reconstructions. We propose a general implicit\nneural representation for unsupervised atmospheric and water turbulence\nmitigation (NeRT). NeRT leverages the implicit neural representations and the\nphysically correct tilt-then-blur turbulence model to reconstruct the clean,\nundistorted image, given only dozens of distorted input images. Moreover, we\nshow that NeRT outperforms the state-of-the-art through various qualitative and\nquantitative evaluations of atmospheric and water turbulence datasets.\nFurthermore, we demonstrate the ability of NeRT to eliminate uncontrolled\nturbulence from real-world environments. Lastly, we incorporate NeRT into\ncontinuously captured video sequences and demonstrate $48 \\times$ speedup.\n", "date": "2023-08-01", "categories": []}
{"id": "2404.01612", "title": "Spin-UP: Spin Light for Natural Light Uncalibrated Photometric Stereo", "abstract": "  Natural Light Uncalibrated Photometric Stereo (NaUPS) relieves the strict\nenvironment and light assumptions in classical Uncalibrated Photometric Stereo\n(UPS) methods. However, due to the intrinsic ill-posedness and high-dimensional\nambiguities, addressing NaUPS is still an open question. Existing works impose\nstrong assumptions on the environment lights and objects' material, restricting\nthe effectiveness in more general scenarios. Alternatively, some methods\nleverage supervised learning with intricate models while lacking\ninterpretability, resulting in a biased estimation. In this work, we proposed\nSpin Light Uncalibrated Photometric Stereo (Spin-UP), an unsupervised method to\ntackle NaUPS in various environment lights and objects. The proposed method\nuses a novel setup that captures the object's images on a rotatable platform,\nwhich mitigates NaUPS's ill-posedness by reducing unknowns and provides\nreliable priors to alleviate NaUPS's ambiguities. Leveraging neural inverse\nrendering and the proposed training strategies, Spin-UP recovers surface\nnormals, environment light, and isotropic reflectance under complex natural\nlight with low computational cost. Experiments have shown that Spin-UP\noutperforms other supervised / unsupervised NaUPS methods and achieves\nstate-of-the-art performance on synthetic and real-world datasets. Codes and\ndata are available at https://github.com/LMozart/CVPR2024-SpinUP.\n", "date": "2024-04-02", "categories": []}
{"id": "2307.16120", "title": "Deep Unrolling Networks with Recurrent Momentum Acceleration for\n  Nonlinear Inverse Problems", "abstract": "  Combining the strengths of model-based iterative algorithms and data-driven\ndeep learning solutions, deep unrolling networks (DuNets) have become a popular\ntool to solve inverse imaging problems. While DuNets have been successfully\napplied to many linear inverse problems, nonlinear problems tend to impair the\nperformance of the method. Inspired by momentum acceleration techniques that\nare often used in optimization algorithms, we propose a recurrent momentum\nacceleration (RMA) framework that uses a long short-term memory recurrent\nneural network (LSTM-RNN) to simulate the momentum acceleration process. The\nRMA module leverages the ability of the LSTM-RNN to learn and retain knowledge\nfrom the previous gradients. We apply RMA to two popular DuNets -- the learned\nproximal gradient descent (LPGD) and the learned primal-dual (LPD) methods,\nresulting in LPGD-RMA and LPD-RMA respectively. We provide experimental results\non two nonlinear inverse problems: a nonlinear deconvolution problem, and an\nelectrical impedance tomography problem with limited boundary measurements. In\nthe first experiment we have observed that the improvement due to RMA largely\nincreases with respect to the nonlinearity of the problem. The results of the\nsecond example further demonstrate that the RMA schemes can significantly\nimprove the performance of DuNets in strongly ill-posed problems.\n", "date": "2023-07-30", "categories": []}
{"id": "2312.04553", "title": "SPIDeRS: Structured Polarization for Invisible Depth and Reflectance\n  Sensing", "abstract": "  Can we capture shape and reflectance in stealth? Such capability would be\nvaluable for many application domains in vision, xR, robotics, and HCI. We\nintroduce structured polarization for invisible depth and reflectance sensing\n(SPIDeRS), the first depth and reflectance sensing method using patterns of\npolarized light. The key idea is to modulate the angle of linear polarization\n(AoLP) of projected light at each pixel. The use of polarization makes it\ninvisible and lets us recover not only depth but also directly surface normals\nand even reflectance. We implement SPIDeRS with a liquid crystal spatial light\nmodulator (SLM) and a polarimetric camera. We derive a novel method for\nrobustly extracting the projected structured polarization pattern from the\npolarimetric object appearance. We evaluate the effectiveness of SPIDeRS by\napplying it to a number of real-world objects. The results show that our method\nsuccessfully reconstructs object shapes of various materials and is robust to\ndiffuse reflection and ambient light. We also demonstrate relighting using\nrecovered surface normals and reflectance. We believe SPIDeRS opens a new\navenue of polarization use in visual sensing.\n", "date": "2023-12-07", "categories": []}
{"id": "2402.18102", "title": "Passive Snapshot Coded Aperture Dual-Pixel RGB-D Imaging", "abstract": "  Passive, compact, single-shot 3D sensing is useful in many application areas\nsuch as microscopy, medical imaging, surgical navigation, and autonomous\ndriving where form factor, time, and power constraints can exist. Obtaining\nRGB-D scene information over a short imaging distance, in an ultra-compact form\nfactor, and in a passive, snapshot manner is challenging. Dual-pixel (DP)\nsensors are a potential solution to achieve the same. DP sensors collect light\nrays from two different halves of the lens in two interleaved pixel arrays,\nthus capturing two slightly different views of the scene, like a stereo camera\nsystem. However, imaging with a DP sensor implies that the defocus blur size is\ndirectly proportional to the disparity seen between the views. This creates a\ntrade-off between disparity estimation vs. deblurring accuracy. To improve this\ntrade-off effect, we propose CADS (Coded Aperture Dual-Pixel Sensing), in which\nwe use a coded aperture in the imaging lens along with a DP sensor. In our\napproach, we jointly learn an optimal coded pattern and the reconstruction\nalgorithm in an end-to-end optimization setting. Our resulting CADS imaging\nsystem demonstrates improvement of >1.5dB PSNR in all-in-focus (AIF) estimates\nand 5-6% in depth estimation quality over naive DP sensing for a wide range of\naperture settings. Furthermore, we build the proposed CADS prototypes for DSLR\nphotography settings and in an endoscope and a dermoscope form factor. Our\nnovel coded dual-pixel sensing approach demonstrates accurate RGB-D\nreconstruction results in simulations and real-world experiments in a passive,\nsnapshot, and compact manner.\n", "date": "2024-02-28", "categories": []}
{"id": "2403.11270", "title": "Bilateral Propagation Network for Depth Completion", "abstract": "  Depth completion aims to derive a dense depth map from sparse depth\nmeasurements with a synchronized color image. Current state-of-the-art (SOTA)\nmethods are predominantly propagation-based, which work as an iterative\nrefinement on the initial estimated dense depth. However, the initial depth\nestimations mostly result from direct applications of convolutional layers on\nthe sparse depth map. In this paper, we present a Bilateral Propagation Network\n(BP-Net), that propagates depth at the earliest stage to avoid directly\nconvolving on sparse data. Specifically, our approach propagates the target\ndepth from nearby depth measurements via a non-linear model, whose coefficients\nare generated through a multi-layer perceptron conditioned on both\n\\emph{radiometric difference} and \\emph{spatial distance}. By integrating\nbilateral propagation with multi-modal fusion and depth refinement in a\nmulti-scale framework, our BP-Net demonstrates outstanding performance on both\nindoor and outdoor scenes. It achieves SOTA on the NYUv2 dataset and ranks 1st\non the KITTI depth completion benchmark at the time of submission. Experimental\nresults not only show the effectiveness of bilateral propagation but also\nemphasize the significance of early-stage propagation in contrast to the\nrefinement stage. Our code and trained models will be available on the project\npage.\n", "date": "2024-03-17", "categories": []}
{"id": "2404.00313", "title": "Harmonizing Light and Darkness: A Symphony of Prior-guided Data\n  Synthesis and Adaptive Focus for Nighttime Flare Removal", "abstract": "  Intense light sources often produce flares in captured images at night, which\ndeteriorates the visual quality and negatively affects downstream applications.\nIn order to train an effective flare removal network, a reliable dataset is\nessential. The mainstream flare removal datasets are semi-synthetic to reduce\nhuman labour, but these datasets do not cover typical scenarios involving\nmultiple scattering flares. To tackle this issue, we synthesize a prior-guided\ndataset named Flare7K*, which contains multi-flare images where the brightness\nof flares adheres to the laws of illumination. Besides, flares tend to occupy\nlocalized regions of the image but existing networks perform flare removal on\nthe entire image and sometimes modify clean areas incorrectly. Therefore, we\npropose a plug-and-play Adaptive Focus Module (AFM) that can adaptively mask\nthe clean background areas and assist models in focusing on the regions\nseverely affected by flares. Extensive experiments demonstrate that our data\nsynthesis method can better simulate real-world scenes and several models\nequipped with AFM achieve state-of-the-art performance on the real-world test\ndataset.\n", "date": "2024-03-30", "categories": []}
{"id": "2404.00545", "title": "Unified, Verifiable Neural Simulators for Electromagnetic Wave Inverse\n  Problems", "abstract": "  Simulators based on neural networks offer a path to orders-of-magnitude\nfaster electromagnetic wave simulations. Existing models, however, only address\nnarrowly tailored classes of problems and only scale to systems of a few dozen\ndegrees of freedom (DoFs). Here, we demonstrate a single, unified model capable\nof addressing scattering simulations with thousands of DoFs, of any wavelength,\nany illumination wavefront, and freeform materials, within broad configurable\nbounds. Based on an attentional multi-conditioning strategy, our method also\nallows non-recurrent supervision on and prediction of intermediate physical\nstates, which provides improved generalization with no additional\ndata-generation cost. Using this O(1)-time intermediate prediction capability,\nwe propose and prove a rigorous, efficiently computable upper bound on\nprediction error, allowing accuracy guarantees at inference time for all\npredictions. After training solely on randomized systems, we demonstrate the\nunified model across a suite of challenging multi-disciplinary inverse\nproblems, finding strong efficacy and speed improvements up to 96% for problems\nin optical tomography, beam shaping through volumetric random media, and\nfreeform photonic inverse design, with no problem-specific training. Our\nfindings demonstrate a path to universal, verifiably accurate neural surrogates\nfor existing scattering simulators, and our conditioning and training methods\nare directly applicable to any PDE admitting a time-domain iterative solver.\n", "date": "2024-03-31", "categories": []}
{"id": "2404.00807", "title": "GAMA-IR: Global Additive Multidimensional Averaging for Fast Image\n  Restoration", "abstract": "  Deep learning-based methods have shown remarkable success for various image\nrestoration tasks such as denoising and deblurring. The current\nstate-of-the-art networks are relatively deep and utilize (variants of) self\nattention mechanisms. Those networks are significantly slower than shallow\nconvolutional networks, which however perform worse. In this paper, we\nintroduce an image restoration network that is both fast and yields excellent\nimage quality. The network is designed to minimize the latency and memory\nconsumption when executed on a standard GPU, while maintaining state-of-the-art\nperformance. The network is a simple shallow network with an efficient block\nthat implements global additive multidimensional averaging operations. This\nblock can capture global information and enable a large receptive field even\nwhen used in shallow networks with minimal computational overhead. Through\nextensive experiments and evaluations on diverse tasks, we demonstrate that our\nnetwork achieves comparable or even superior results to existing\nstate-of-the-art image restoration networks with less latency. For instance, we\nexceed the state-of-the-art result on real-world SIDD denoising by 0.11dB,\nwhile being 2 to 10 times faster.\n", "date": "2024-03-31", "categories": []}
{"id": "2403.20018", "title": "SCINeRF: Neural Radiance Fields from a Snapshot Compressive Image", "abstract": "  In this paper, we explore the potential of Snapshot Compressive Imaging (SCI)\ntechnique for recovering the underlying 3D scene representation from a single\ntemporal compressed image. SCI is a cost-effective method that enables the\nrecording of high-dimensional data, such as hyperspectral or temporal\ninformation, into a single image using low-cost 2D imaging sensors. To achieve\nthis, a series of specially designed 2D masks are usually employed, which not\nonly reduces storage requirements but also offers potential privacy protection.\nInspired by this, to take one step further, our approach builds upon the\npowerful 3D scene representation capabilities of neural radiance fields (NeRF).\nSpecifically, we formulate the physical imaging process of SCI as part of the\ntraining of NeRF, allowing us to exploit its impressive performance in\ncapturing complex scene structures. To assess the effectiveness of our method,\nwe conduct extensive evaluations using both synthetic data and real data\ncaptured by our SCI system. Extensive experimental results demonstrate that our\nproposed approach surpasses the state-of-the-art methods in terms of image\nreconstruction and novel view image synthesis. Moreover, our method also\nexhibits the ability to restore high frame-rate multi-view consistent images by\nleveraging SCI and the rendering capabilities of NeRF. The code is available at\nhttps://github.com/WU-CVGL/SCINeRF.\n", "date": "2024-03-29", "categories": []}
{"id": "2311.16473", "title": "GS-IR: 3D Gaussian Splatting for Inverse Rendering", "abstract": "  We propose GS-IR, a novel inverse rendering approach based on 3D Gaussian\nSplatting (GS) that leverages forward mapping volume rendering to achieve\nphotorealistic novel view synthesis and relighting results. Unlike previous\nworks that use implicit neural representations and volume rendering (e.g.\nNeRF), which suffer from low expressive power and high computational\ncomplexity, we extend GS, a top-performance representation for novel view\nsynthesis, to estimate scene geometry, surface material, and environment\nillumination from multi-view images captured under unknown lighting conditions.\nThere are two main problems when introducing GS to inverse rendering: 1) GS\ndoes not support producing plausible normal natively; 2) forward mapping (e.g.\nrasterization and splatting) cannot trace the occlusion like backward mapping\n(e.g. ray tracing). To address these challenges, our GS-IR proposes an\nefficient optimization scheme that incorporates a depth-derivation-based\nregularization for normal estimation and a baking-based occlusion to model\nindirect lighting. The flexible and expressive GS representation allows us to\nachieve fast and compact geometry reconstruction, photorealistic novel view\nsynthesis, and effective physically-based rendering. We demonstrate the\nsuperiority of our method over baseline methods through qualitative and\nquantitative evaluations on various challenging scenes.\n", "date": "2023-11-26", "categories": []}
{"id": "2401.03707", "title": "FMA-Net: Flow-Guided Dynamic Filtering and Iterative Feature Refinement\n  with Multi-Attention for Joint Video Super-Resolution and Deblurring", "abstract": "  We present a joint learning scheme of video super-resolution and deblurring,\ncalled VSRDB, to restore clean high-resolution (HR) videos from blurry\nlow-resolution (LR) ones. This joint restoration problem has drawn much less\nattention compared to single restoration problems. In this paper, we propose a\nnovel flow-guided dynamic filtering (FGDF) and iterative feature refinement\nwith multi-attention (FRMA), which constitutes our VSRDB framework, denoted as\nFMA-Net. Specifically, our proposed FGDF enables precise estimation of both\nspatio-temporally-variant degradation and restoration kernels that are aware of\nmotion trajectories through sophisticated motion representation learning.\nCompared to conventional dynamic filtering, the FGDF enables the FMA-Net to\neffectively handle large motions into the VSRDB. Additionally, the stacked FRMA\nblocks trained with our novel temporal anchor (TA) loss, which temporally\nanchors and sharpens features, refine features in a course-to-fine manner\nthrough iterative updates. Extensive experiments demonstrate the superiority of\nthe proposed FMA-Net over state-of-the-art methods in terms of both\nquantitative and qualitative quality. Codes and pre-trained models are\navailable at: https://kaist-viclab.github.io/fmanet-site\n", "date": "2024-01-08", "categories": []}
{"id": "2403.19066", "title": "Generative Quanta Color Imaging", "abstract": "  The astonishing development of single-photon cameras has created an\nunprecedented opportunity for scientific and industrial imaging. However, the\nhigh data throughput generated by these 1-bit sensors creates a significant\nbottleneck for low-power applications. In this paper, we explore the\npossibility of generating a color image from a single binary frame of a\nsingle-photon camera. We evidently find this problem being particularly\ndifficult to standard colorization approaches due to the substantial degree of\nexposure variation. The core innovation of our paper is an exposure synthesis\nmodel framed under a neural ordinary differential equation (Neural ODE) that\nallows us to generate a continuum of exposures from a single observation. This\ninnovation ensures consistent exposure in binary images that colorizers take\non, resulting in notably enhanced colorization. We demonstrate applications of\nthe method in single-image and burst colorization and show superior generative\nperformance over baselines. Project website can be found at\nhttps://vishal-s-p.github.io/projects/2023/generative_quanta_color.html.\n", "date": "2024-03-28", "categories": []}
{"id": "2403.18548", "title": "A Semi-supervised Nighttime Dehazing Baseline with Spatial-Frequency\n  Aware and Realistic Brightness Constraint", "abstract": "  Existing research based on deep learning has extensively explored the problem\nof daytime image dehazing. However, few studies have considered the\ncharacteristics of nighttime hazy scenes. There are two distinctions between\nnighttime and daytime haze. First, there may be multiple active colored light\nsources with lower illumination intensity in nighttime scenes, which may cause\nhaze, glow and noise with localized, coupled and frequency inconsistent\ncharacteristics. Second, due to the domain discrepancy between simulated and\nreal-world data, unrealistic brightness may occur when applying a dehazing\nmodel trained on simulated data to real-world data. To address the above two\nissues, we propose a semi-supervised model for real-world nighttime dehazing.\nFirst, the spatial attention and frequency spectrum filtering are implemented\nas a spatial-frequency domain information interaction module to handle the\nfirst issue. Second, a pseudo-label-based retraining strategy and a local\nwindow-based brightness loss for semi-supervised training process is designed\nto suppress haze and glow while achieving realistic brightness. Experiments on\npublic benchmarks validate the effectiveness of the proposed method and its\nsuperiority over state-of-the-art methods. The source code and Supplementary\nMaterials are placed in the https://github.com/Xiaofeng-life/SFSNiD.\n", "date": "2024-03-27", "categories": []}
{"id": "2206.03709", "title": "Hypernetwork-based Personalized Federated Learning for\n  Multi-Institutional CT Imaging", "abstract": "  Computed tomography (CT) is of great importance in clinical practice due to\nits powerful ability to provide patients' anatomical information without any\ninvasive inspection, but its potential radiation risk is raising people's\nconcerns. Deep learning-based methods are considered promising in CT\nreconstruction, but these network models are usually trained with the measured\ndata obtained from specific scanning protocol and need to centralizedly collect\nlarge amounts of data, which will lead to serious data domain shift, and\nprivacy concerns. To relieve these problems, in this paper, we propose a\nhypernetwork-based federated learning method for personalized CT imaging,\ndubbed as HyperFed. The basic assumption of HyperFed is that the optimization\nproblem for each institution can be divided into two parts: the local data\nadaption problem and the global CT imaging problem, which are implemented by an\ninstitution-specific hypernetwork and a global-sharing imaging network,\nrespectively. The purpose of global-sharing imaging network is to learn stable\nand effective common features from different institutions. The\ninstitution-specific hypernetwork is carefully designed to obtain\nhyperparameters to condition the global-sharing imaging network for\npersonalized local CT reconstruction. Experiments show that HyperFed achieves\ncompetitive performance in CT reconstruction compared with several other\nstate-of-the-art methods. It is believed as a promising direction to improve CT\nimaging quality and achieve personalized demands of different institutions or\nscanners without privacy data sharing. The codes will be released at\nhttps://github.com/Zi-YuanYang/HyperFed.\n", "date": "2022-06-08", "categories": []}
{"id": "2311.10959", "title": "Structure-Aware Sparse-View X-ray 3D Reconstruction", "abstract": "  X-ray, known for its ability to reveal internal structures of objects, is\nexpected to provide richer information for 3D reconstruction than visible\nlight. Yet, existing neural radiance fields (NeRF) algorithms overlook this\nimportant nature of X-ray, leading to their limitations in capturing structural\ncontents of imaged objects. In this paper, we propose a framework,\nStructure-Aware X-ray Neural Radiodensity Fields (SAX-NeRF), for sparse-view\nX-ray 3D reconstruction. Firstly, we design a Line Segment-based Transformer\n(Lineformer) as the backbone of SAX-NeRF. Linefomer captures internal\nstructures of objects in 3D space by modeling the dependencies within each line\nsegment of an X-ray. Secondly, we present a Masked Local-Global (MLG) ray\nsampling strategy to extract contextual and geometric information in 2D\nprojection. Plus, we collect a larger-scale dataset X3D covering wider X-ray\napplications. Experiments on X3D show that SAX-NeRF surpasses previous\nNeRF-based methods by 12.56 and 2.49 dB on novel view synthesis and CT\nreconstruction. Code, models, and data are released at\nhttps://github.com/caiyuanhao1998/SAX-NeRF\n", "date": "2023-11-18", "categories": []}
{"id": "2403.16974", "title": "Self-STORM: Deep Unrolled Self-Supervised Learning for Super-Resolution\n  Microscopy", "abstract": "  The use of fluorescent molecules to create long sequences of low-density,\ndiffraction-limited images enables highly-precise molecule localization.\nHowever, this methodology requires lengthy imaging times, which limits the\nability to view dynamic interactions of live cells on short time scales. Many\ntechniques have been developed to reduce the number of frames needed for\nlocalization, from classic iterative optimization to deep neural networks.\nParticularly, deep algorithm unrolling utilizes both the structure of iterative\nsparse recovery algorithms and the performance gains of supervised deep\nlearning. However, the robustness of this approach is highly dependant on\nhaving sufficient training data. In this paper we introduce deep unrolled\nself-supervised learning, which alleviates the need for such data by training a\nsequence-specific, model-based autoencoder that learns only from given\nmeasurements. Our proposed method exceeds the performance of its supervised\ncounterparts, thus allowing for robust, dynamic imaging well below the\ndiffraction limit without any labeled training samples. Furthermore, the\nsuggested model-based autoencoder scheme can be utilized to enhance\ngeneralization in any sparse recovery framework, without the need for external\ntraining data.\n", "date": "2024-03-25", "categories": []}
{"id": "2403.14863", "title": "Distribution-informed and wavelength-flexible data-driven photoacoustic\n  oximetry", "abstract": "  Significance: Photoacoustic imaging (PAI) promises to measure\nspatially-resolved blood oxygen saturation, but suffers from a lack of accurate\nand robust spectral unmixing methods to deliver on this promise. Accurate blood\noxygenation estimation could have important clinical applications, from cancer\ndetection to quantifying inflammation.\n  Aim: This study addresses the inflexibility of existing data-driven methods\nfor estimating blood oxygenation in PAI by introducing a recurrent neural\nnetwork architecture.\n  Approach: We created 25 simulated training dataset variations to assess\nneural network performance. We used a long short-term memory network to\nimplement a wavelength-flexible network architecture and proposed the\nJensen-Shannon divergence to predict the most suitable training dataset.\n  Results: The network architecture can handle arbitrary input wavelengths and\noutperforms linear unmixing and the previously proposed learned spectral\ndecolouring method. Small changes in the training data significantly affect the\naccuracy of our method, but we find that the Jensen-Shannon divergence\ncorrelates with the estimation error and is thus suitable for predicting the\nmost appropriate training datasets for any given application.\n  Conclusions: A flexible data-driven network architecture combined with the\nJensen-Shannon Divergence to predict the best training data set provides a\npromising direction that might enable robust data-driven photoacoustic oximetry\nfor clinical use cases.\n", "date": "2024-03-21", "categories": []}
{"id": "2401.03379", "title": "Towards Effective Multiple-in-One Image Restoration: A Sequential and\n  Prompt Learning Strategy", "abstract": "  While single task image restoration (IR) has achieved significant successes,\nit remains a challenging issue to train a single model which can tackle\nmultiple IR tasks. In this work, we investigate in-depth the multiple-in-one\n(MiO) IR problem, which comprises seven popular IR tasks. We point out that MiO\nIR faces two pivotal challenges: the optimization of diverse objectives and the\nadaptation to multiple tasks. To tackle these challenges, we present two simple\nyet effective strategies. The first strategy, referred to as sequential\nlearning, attempts to address how to optimize the diverse objectives, which\nguides the network to incrementally learn individual IR tasks in a sequential\nmanner rather than mixing them together. The second strategy, i.e., prompt\nlearning, attempts to address how to adapt to the different IR tasks, which\nassists the network to understand the specific task and improves the\ngeneralization ability. By evaluating on 19 test sets, we demonstrate that the\nsequential and prompt learning strategies can significantly enhance the MiO\nperformance of commonly used CNN and Transformer backbones. Our experiments\nalso reveal that the two strategies can supplement each other to learn better\ndegradation representations and enhance the model robustness. It is expected\nthat our proposed MiO IR formulation and strategies could facilitate the\nresearch on how to train IR models with higher generalization capabilities.\n", "date": "2024-01-07", "categories": []}
{"id": "2403.12970", "title": "Hybrid deep learning and physics-based neural network for programmable\n  illumination computational microscopy", "abstract": "  Relying on either deep models or physical models are two mainstream\napproaches for solving inverse sample reconstruction problems in programmable\nillumination computational microscopy. Solutions based on physical models\npossess strong generalization capabilities while struggling with global\noptimization of inverse problems due to a lack of insufficient physical\nconstraints. In contrast, deep learning methods have strong problem-solving\nabilities, but their generalization ability is often questioned because of the\nunclear physical principles. Besides, conventional deep models are difficult to\napply to some specific scenes because of the difficulty in acquiring\nhigh-quality training data and their limited capacity to generalize across\ndifferent scenarios. In this paper, to combine the advantages of deep models\nand physical models together, we propose a hybrid framework consisting of three\nsub-neural networks (two deep learning networks and one physics-based network).\nWe first obtain a result with rich semantic information through a light deep\nlearning neural network and then use it as the initial value of the physical\nnetwork to make its output comply with physical process constraints. These two\nresults are then used as the input of a fusion deep learning neural work which\nutilizes the paired features between the reconstruction results of two\ndifferent models to further enhance imaging quality. The final result\nintegrates the advantages of both deep models and physical models and can\nquickly solve the computational reconstruction inverse problem in programmable\nillumination computational microscopy and achieve better results. We verified\nthe feasibility and effectiveness of the proposed hybrid framework with\ntheoretical analysis and actual experiments on resolution targets and\nbiological samples.\n", "date": "2024-01-17", "categories": []}
{"id": "2403.12054", "title": "Haze Removal via Regional Saturation-Value Translation and Soft\n  Segmentation", "abstract": "  This paper proposes a single image dehazing prior, called Regional\nSaturation-Value Translation (RSVT), to tackle the color distortion problems\ncaused by conventional dehazing approaches in bright regions. The RSVT prior is\ndeveloped based on two key observations regarding the relationship between hazy\nand haze-free points in the HSV color space. First, the hue component shows\nmarginal variation between corresponding hazy and haze-free points,\nconsolidating a hypothesis that the pixel value variability induced by haze\nprimarily occurs in the saturation and value spaces. Second, in the 2D\nsaturation-value coordinate system, most lines passing through hazy-clean point\npairs are likely to intersect near the atmospheric light coordinates.\nAccordingly, haze removal for the bright regions can be performed by properly\ntranslating saturation-value coordinates. In addition, an effective soft\nsegmentation method based on a morphological min-max channel is introduced. By\ncombining the soft segmentation mask with the RSVT prior, a comprehensive\nsingle image dehazing framework is devised. Experimental results on various\nsynthetic and realistic hazy image datasets demonstrate that the proposed\nscheme successfully addresses color distortion issues and restores visually\nappealing images. The code of this work is available at\nhttps://github.com/tranleanh/rsvt.\n", "date": "2024-01-07", "categories": []}
{"id": "2403.12056", "title": "Enhancing Digital Hologram Reconstruction Using Reverse-Attention Loss\n  for Untrained Physics-Driven Deep Learning Models with Uncertain Distance", "abstract": "  Untrained Physics-based Deep Learning (DL) methods for digital holography\nhave gained significant attention due to their benefits, such as not requiring\nan annotated training dataset, and providing interpretability since utilizing\nthe governing laws of hologram formation. However, they are sensitive to the\nhard-to-obtain precise object distance from the imaging plane, posing the\n$\\textit{Autofocusing}$ challenge. Conventional solutions involve\nreconstructing image stacks for different potential distances and applying\nfocus metrics to select the best results, which apparently is computationally\ninefficient. In contrast, recently developed DL-based methods treat it as a\nsupervised task, which again needs annotated data and lacks generalizability.\nTo address this issue, we propose $\\textit{reverse-attention loss}$, a weighted\nsum of losses for all possible candidates with learnable weights. This is a\npioneering approach to addressing the Autofocusing challenge in untrained\ndeep-learning methods. Both theoretical analysis and experiments demonstrate\nits superiority in efficiency and accuracy. Interestingly, our method presents\na significant reconstruction performance over rival methods (i.e. alternating\ndescent-like optimization, non-weighted loss integration, and random distance\nassignment) and even is almost equal to that achieved with a precisely known\nobject distance. For example, the difference is less than 1dB in PSNR and 0.002\nin SSIM for the target sample in our experiment.\n", "date": "2024-01-11", "categories": []}
{"id": "2403.12770", "title": "Multispectral Image Restoration by Generalized Opponent Transformation\n  Total Variation", "abstract": "  Multispectral images (MSI) contain light information in different wavelengths\nof objects, which convey spectral-spatial information and help improve the\nperformance of various image processing tasks. Numerous techniques have been\ncreated to extend the application of total variation regularization in\nrestoring multispectral images, for example, based on channel coupling and\nadaptive total variation regularization. The primary contribution of this paper\nis to propose and develop a new multispectral total variation regularization in\na generalized opponent transformation domain instead of the original\nmultispectral image domain. Here opponent transformations for multispectral\nimages are generalized from a well-known opponent transformation for color\nimages. We will explore the properties of generalized opponent transformation\ntotal variation (GOTTV) regularization and the corresponding optimization\nformula for multispectral image restoration. To evaluate the effectiveness of\nthe new GOTTV method, we provide numerical examples that showcase its superior\nperformance compared to existing multispectral image total variation methods,\nusing criteria such as MPSNR and MSSIM.\n", "date": "2024-03-19", "categories": []}
{"id": "2403.12933", "title": "Zero-Reference Low-Light Enhancement via Physical Quadruple Priors", "abstract": "  Understanding illumination and reducing the need for supervision pose a\nsignificant challenge in low-light enhancement. Current approaches are highly\nsensitive to data usage during training and illumination-specific\nhyper-parameters, limiting their ability to handle unseen scenarios. In this\npaper, we propose a new zero-reference low-light enhancement framework\ntrainable solely with normal light images. To accomplish this, we devise an\nillumination-invariant prior inspired by the theory of physical light transfer.\nThis prior serves as the bridge between normal and low-light images. Then, we\ndevelop a prior-to-image framework trained without low-light data. During\ntesting, this framework is able to restore our illumination-invariant prior\nback to images, automatically achieving low-light enhancement. Within this\nframework, we leverage a pretrained generative diffusion model for model\nability, introduce a bypass decoder to handle detail distortion, as well as\noffer a lightweight version for practicality. Extensive experiments demonstrate\nour framework's superiority in various scenarios as well as good\ninterpretability, robustness, and efficiency. Code is available on our project\nhomepage: http://daooshee.github.io/QuadPrior-Website/\n", "date": "2024-03-19", "categories": []}
{"id": "2210.06891", "title": "Experimental Design for Multi-Channel Imaging via Task-Driven Feature\n  Selection", "abstract": "  This paper presents a data-driven, task-specific paradigm for experimental\ndesign, to shorten acquisition time, reduce costs, and accelerate the\ndeployment of imaging devices. Current approaches in experimental design focus\non model-parameter estimation and require specification of a particular model,\nwhereas in imaging, other tasks may drive the design. Furthermore, such\napproaches often lead to intractable optimization problems in real-world\nimaging applications. Here we present a new paradigm for experimental design\nthat simultaneously optimizes the design (set of image channels) and trains a\nmachine-learning model to execute a user-specified image-analysis task. The\napproach obtains data densely-sampled over the measurement space (many image\nchannels) for a small number of acquisitions, then identifies a subset of\nchannels of prespecified size that best supports the task. We propose a method:\nTADRED for TAsk-DRiven Experimental Design in imaging, to identify the most\ninformative channel-subset whilst simultaneously training a network to execute\nthe task given the subset. Experiments demonstrate the potential of TADRED in\ndiverse imaging applications: several clinically-relevant tasks in magnetic\nresonance imaging; and remote sensing and physiological applications of\nhyperspectral imaging. Results show substantial improvement over classical\nexperimental design, two recent application-specific methods within the new\nparadigm, and state-of-the-art approaches in supervised feature selection. We\nanticipate further applications of our approach. Code is available:\nhttps://github.com/sbb-gh/experimental-design-multichannel\n", "date": "2022-10-13", "categories": []}
{"id": "2311.12528", "title": "Inverse Problems with Learned Forward Operators", "abstract": "  Solving inverse problems requires the knowledge of the forward operator, but\naccurate models can be computationally expensive and hence cheaper variants\nthat do not compromise the reconstruction quality are desired. This chapter\nreviews reconstruction methods in inverse problems with learned forward\noperators that follow two different paradigms. The first one is completely\nagnostic to the forward operator and learns its restriction to the subspace\nspanned by the training data. The framework of regularisation by projection is\nthen used to find a reconstruction. The second one uses a simplified model of\nthe physics of the measurement process and only relies on the training data to\nlearn a model correction. We present the theory of these two approaches and\ncompare them numerically. A common theme emerges: both methods require, or at\nleast benefit from, training data not only for the forward operator, but also\nfor its adjoint.\n", "date": "2023-11-21", "categories": []}
{"id": "2312.12833", "title": "Learning Exhaustive Correlation for Spectral Super-Resolution: Where\n  Spatial-Spectral Attention Meets Linear Dependence", "abstract": "  Spectral super-resolution that aims to recover hyperspectral image (HSI) from\neasily obtainable RGB image has drawn increasing interest in the field of\ncomputational photography. The crucial aspect of spectral super-resolution lies\nin exploiting the correlation within HSIs. However, two types of bottlenecks in\nexisting Transformers limit performance improvement and practical applications.\nFirst, existing Transformers often separately emphasize either spatial-wise or\nspectral-wise correlation, disrupting the 3D features of HSI and hindering the\nexploitation of unified spatial-spectral correlation. Second, existing\nself-attention mechanism always establishes full-rank correlation matrix by\nlearning the correlation between pairs of tokens, leading to its inability to\ndescribe linear dependence widely existing in HSI among multiple tokens. To\naddress these issues, we propose a novel Exhaustive Correlation Transformer\n(ECT) for spectral super-resolution. First, we propose a Spectral-wise\nDiscontinuous 3D (SD3D) splitting strategy, which models unified\nspatial-spectral correlation by integrating spatial-wise continuous splitting\nstrategy and spectral-wise discontinuous splitting strategy. Second, we propose\na Dynamic Low-Rank Mapping (DLRM) model, which captures linear dependence among\nmultiple tokens through a dynamically calculated low-rank dependence map. By\nintegrating unified spatial-spectral attention and linear dependence, our ECT\ncan model exhaustive correlation within HSI. The experimental results on both\nsimulated and real data indicate that our method achieves state-of-the-art\nperformance. Codes and pretrained models will be available later.\n", "date": "2023-12-20", "categories": []}
{"id": "2403.11649", "title": "Gridless 2D Recovery of Lines using the Sliding Frank-Wolfe Algorithm", "abstract": "  We present a new approach leveraging the Sliding Frank--Wolfe algorithm to\naddress the challenge of line recovery in degraded images. Building upon\nadvances in conditional gradient methods for sparse inverse problems with\ndifferentiable measurement models, we propose two distinct models tailored for\nline detection tasks within the realm of blurred line deconvolution and ridge\ndetection of linear chirps in spectrogram images.\n", "date": "2024-03-18", "categories": []}
{"id": "2403.11961", "title": "Enhanced Event-Based Video Reconstruction with Motion Compensation", "abstract": "  Deep neural networks for event-based video reconstruction often suffer from a\nlack of interpretability and have high memory demands. A lightweight network\ncalled CISTA-LSTC has recently been introduced showing that high-quality\nreconstruction can be achieved through the systematic design of its\narchitecture. However, its modelling assumption that input signals and output\nreconstructed frame share the same sparse representation neglects the\ndisplacement caused by motion. To address this, we propose warping the input\nintensity frames and sparse codes to enhance reconstruction quality. A\nCISTA-Flow network is constructed by integrating a flow network with CISTA-LSTC\nfor motion compensation. The system relies solely on events, in which predicted\nflow aids in reconstruction and then reconstructed frames are used to\nfacilitate flow estimation. We also introduce an iterative training framework\nfor this combined system. Results demonstrate that our approach achieves\nstate-of-the-art reconstruction accuracy and simultaneously provides reliable\ndense flow estimation. Furthermore, our model exhibits flexibility in that it\ncan integrate different flow networks, suggesting its potential for further\nperformance enhancement.\n", "date": "2024-03-18", "categories": []}
{"id": "2303.07033", "title": "SelfPromer: Self-Prompt Dehazing Transformers with Depth-Consistency", "abstract": "  This work presents an effective depth-consistency self-prompt Transformer for\nimage dehazing. It is motivated by an observation that the estimated depths of\nan image with haze residuals and its clear counterpart vary. Enforcing the\ndepth consistency of dehazed images with clear ones, therefore, is essential\nfor dehazing. For this purpose, we develop a prompt based on the features of\ndepth differences between the hazy input images and corresponding clear\ncounterparts that can guide dehazing models for better restoration.\nSpecifically, we first apply deep features extracted from the input images to\nthe depth difference features for generating the prompt that contains the haze\nresidual information in the input. Then we propose a prompt embedding module\nthat is designed to perceive the haze residuals, by linearly adding the prompt\nto the deep features. Further, we develop an effective prompt attention module\nto pay more attention to haze residuals for better removal. By incorporating\nthe prompt, prompt embedding, and prompt attention into an encoder-decoder\nnetwork based on VQGAN, we can achieve better perception quality. As the depths\nof clear images are not available at inference, and the dehazed images with\none-time feed-forward execution may still contain a portion of haze residuals,\nwe propose a new continuous self-prompt inference that can iteratively correct\nthe dehazing model towards better haze-free image generation. Extensive\nexperiments show that our method performs favorably against the\nstate-of-the-art approaches on both synthetic and real-world datasets in terms\nof perception metrics including NIQE, PI, and PIQE.\n", "date": "2023-03-13", "categories": []}
{"id": "2403.08247", "title": "A Dual-domain Regularization Method for Ring Artifact Removal of X-ray\n  CT", "abstract": "  Ring artifacts in computed tomography images, arising from the undesirable\nresponses of detector units, significantly degrade image quality and diagnostic\nreliability. To address this challenge, we propose a dual-domain regularization\nmodel to effectively remove ring artifacts, while maintaining the integrity of\nthe original CT image. The proposed model corrects the vertical stripe\nartifacts on the sinogram by innovatively updating the response inconsistency\ncompensation coefficients of detector units, which is achieved by employing the\ngroup sparse constraint and the projection-view direction sparse constraint on\nthe stripe artifacts. Simultaneously, we apply the sparse constraint on the\nreconstructed image to further rectified ring artifacts in the image domain.\nThe key advantage of the proposed method lies in considering the relationship\nbetween the response inconsistency compensation coefficients of the detector\nunits and the projection views, which enables a more accurate correction of the\nresponse of the detector units. An alternating minimization method is designed\nto solve the model. Comparative experiments on real photon counting detector\ndata demonstrate that the proposed method not only surpasses existing methods\nin removing ring artifacts but also excels in preserving structural details and\nimage fidelity.\n", "date": "2024-03-13", "categories": []}
{"id": "2403.10374", "title": "Overcoming Distribution Shifts in Plug-and-Play Methods with Test-Time\n  Training", "abstract": "  Plug-and-Play Priors (PnP) is a well-known class of methods for solving\ninverse problems in computational imaging. PnP methods combine physical forward\nmodels with learned prior models specified as image denoisers. A common issue\nwith the learned models is that of a performance drop when there is a\ndistribution shift between the training and testing data. Test-time training\n(TTT) was recently proposed as a general strategy for improving the performance\nof learned models when training and testing data come from different\ndistributions. In this paper, we propose PnP-TTT as a new method for overcoming\ndistribution shifts in PnP. PnP-TTT uses deep equilibrium learning (DEQ) for\noptimizing a self-supervised loss at the fixed points of PnP iterations.\nPnP-TTT can be directly applied on a single test sample to improve the\ngeneralization of PnP. We show through simulations that given a sufficient\nnumber of measurements, PnP-TTT enables the use of image priors trained on\nnatural images for image reconstruction in magnetic resonance imaging (MRI).\n", "date": "2024-03-15", "categories": []}
{"id": "2211.11257", "title": "Computational Imaging for Machine Perception: Transferring Semantic\n  Segmentation beyond Aberrations", "abstract": "  Semantic scene understanding with Minimalist Optical Systems (MOS) in mobile\nand wearable applications remains a challenge due to the corrupted imaging\nquality induced by optical aberrations. However, previous works only focus on\nimproving the subjective imaging quality through the Computational Imaging (CI)\ntechnique, ignoring the feasibility of advancing semantic segmentation. In this\npaper, we pioneer the investigation of Semantic Segmentation under Optical\nAberrations (SSOA) with MOS. To benchmark SSOA, we construct Virtual Prototype\nLens (VPL) groups through optical simulation, generating Cityscapes-ab and\nKITTI-360-ab datasets under different behaviors and levels of aberrations. We\nlook into SSOA via an unsupervised domain adaptation perspective to address the\nscarcity of labeled aberration data in real-world scenarios. Further, we\npropose Computational Imaging Assisted Domain Adaptation (CIADA) to leverage\nprior knowledge of CI for robust performance in SSOA. Based on our benchmark,\nwe conduct experiments on the robustness of classical segmenters against\naberrations. In addition, extensive evaluations of possible solutions to SSOA\nreveal that CIADA achieves superior performance under all aberration\ndistributions, bridging the gap between computational imaging and downstream\napplications for MOS. The project page is at\nhttps://github.com/zju-jiangqi/CIADA.\n", "date": "2022-11-21", "categories": []}
{"id": "2312.16024", "title": "Plug-and-Play Regularization on Magnitude with Deep Priors for 3D\n  Near-Field MIMO Imaging", "abstract": "  Near-field radar imaging systems are used in a wide range of applications\nsuch as concealed weapon detection and medical diagnosis. In this paper, we\nconsider the problem of reconstructing the three-dimensional (3D)\ncomplex-valued reflectivity distribution of the near-field scene by enforcing\nregularization on its magnitude. We solve this inverse problem by using the\nalternating direction method of multipliers (ADMM) framework. For this, we\nprovide a general expression for the proximal mapping associated with such\nregularization functionals. This equivalently corresponds to the solution of a\ncomplex-valued denoising problem which involves regularization on the\nmagnitude. By utilizing this expression, we develop a novel and efficient\nplug-and-play (PnP) reconstruction method that consists of simple update steps.\nDue to the success of data-adaptive deep priors in imaging, we also train a 3D\ndeep denoiser to exploit within the developed PnP framework. The effectiveness\nof the developed approach is demonstrated for multiple-input multiple-output\n(MIMO) imaging under various compressive and noisy observation scenarios using\nboth simulated and experimental data. The performance is also compared with the\ncommonly used direct inversion and sparsity-based reconstruction approaches.\nThe results demonstrate that the developed technique not only provides\nstate-of-the-art performance for 3D real-world targets, but also enables fast\ncomputation. Our approach provides a unified general framework to effectively\nhandle arbitrary regularization on the magnitude of a complex-valued unknown\nand is equally applicable to other radar image formation problems (including\nSAR).\n", "date": "2023-12-26", "categories": []}
{"id": "2403.09096", "title": "Deep unfolding Network for Hyperspectral Image Super-Resolution with\n  Automatic Exposure Correction", "abstract": "  In recent years, the fusion of high spatial resolution multispectral image\n(HR-MSI) and low spatial resolution hyperspectral image (LR-HSI) has been\nrecognized as an effective method for HSI super-resolution (HSI-SR). However,\nboth HSI and MSI may be acquired under extreme conditions such as night or\npoorly illuminating scenarios, which may cause different exposure levels,\nthereby seriously downgrading the yielded HSISR. In contrast to most existing\nmethods based on respective low-light enhancements (LLIE) of MSI and HSI\nfollowed by their fusion, a deep Unfolding HSI Super-Resolution with Automatic\nExposure Correction (UHSR-AEC) is proposed, that can effectively generate a\nhigh-quality fused HSI-SR (in texture and features) even under very imbalanced\nexposures, thanks to the correlation between LLIE and HSI-SR taken into\naccount. Extensive experiments are provided to demonstrate the state-of-the-art\noverall performance of the proposed UHSR-AEC, including comparison with some\nbenchmark peer methods.\n", "date": "2024-03-14", "categories": []}
{"id": "2403.09392", "title": "Event-based Asynchronous HDR Imaging by Temporal Incident Light\n  Modulation", "abstract": "  Dynamic Range (DR) is a pivotal characteristic of imaging systems. Current\nframe-based cameras struggle to achieve high dynamic range imaging due to the\nconflict between globally uniform exposure and spatially variant scene\nillumination. In this paper, we propose AsynHDR, a Pixel-Asynchronous HDR\nimaging system, based on key insights into the challenges in HDR imaging and\nthe unique event-generating mechanism of Dynamic Vision Sensors (DVS). Our\nproposed AsynHDR system integrates the DVS with a set of LCD panels. The LCD\npanels modulate the irradiance incident upon the DVS by altering their\ntransparency, thereby triggering the pixel-independent event streams. The HDR\nimage is subsequently decoded from the event streams through our\ntemporal-weighted algorithm. Experiments under standard test platform and\nseveral challenging scenes have verified the feasibility of the system in HDR\nimaging task.\n", "date": "2024-03-14", "categories": []}
{"id": "2403.08749", "title": "Clinically Feasible Diffusion Reconstruction for Highly-Accelerated\n  Cardiac Cine MRI", "abstract": "  The currently limited quality of accelerated cardiac cine reconstruction may\npotentially be improved by the emerging diffusion models, but the clinically\nunacceptable long processing time poses a challenge. We aim to develop a\nclinically feasible diffusion-model-based reconstruction pipeline to improve\nthe image quality of cine MRI. A multi-in multi-out diffusion enhancement model\ntogether with fast inference strategies were developed to be used in\nconjunction with a reconstruction model. The diffusion reconstruction reduced\nspatial and temporal blurring in prospectively undersampled clinical data, as\nvalidated by experts inspection. The 1.5s per video processing time enabled the\napproach to be applied in clinical scenarios.\n", "date": "2024-03-13", "categories": []}
{"id": "2306.13325", "title": "Differentiable Display Photometric Stereo", "abstract": "  Photometric stereo leverages variations in illumination conditions to\nreconstruct surface normals. Display photometric stereo, which employs a\nconventional monitor as an illumination source, has the potential to overcome\nlimitations often encountered in bulky and difficult-to-use conventional\nsetups. In this paper, we present differentiable display photometric stereo\n(DDPS), addressing an often overlooked challenge in display photometric stereo:\nthe design of display patterns. Departing from using heuristic display\npatterns, DDPS learns the display patterns that yield accurate normal\nreconstruction for a target system in an end-to-end manner. To this end, we\npropose a differentiable framework that couples basis-illumination image\nformation with analytic photometric-stereo reconstruction. The differentiable\nframework facilitates the effective learning of display patterns via\nauto-differentiation. Also, for training supervision, we propose to use 3D\nprinting for creating a real-world training dataset, enabling accurate\nreconstruction on the target real-world setup. Finally, we exploit that\nconventional LCD monitors emit polarized light, which allows for the optical\nseparation of diffuse and specular reflections when combined with a\npolarization camera, leading to accurate normal reconstruction. Extensive\nevaluation of DDPS shows improved normal-reconstruction accuracy compared to\nheuristic patterns and demonstrates compelling properties such as robustness to\npattern initialization, calibration errors, and simplifications in image\nformation and reconstruction.\n", "date": "2023-06-23", "categories": []}
{"id": "2308.00759", "title": "Decomposition Ascribed Synergistic Learning for Unified Image\n  Restoration", "abstract": "  Learning to restore multiple image degradations within a single model is\nquite beneficial for real-world applications. Nevertheless, existing works\ntypically concentrate on regarding each degradation independently, while their\nrelationship has been less exploited to ensure the synergistic learning. To\nthis end, we revisit the diverse degradations through the lens of singular\nvalue decomposition, with the observation that the decomposed singular vectors\nand singular values naturally undertake the different types of degradation\ninformation, dividing various restoration tasks into two groups, \\ie, singular\nvector dominated and singular value dominated. The above analysis renders a\nmore unified perspective to ascribe the diverse degradations, compared to\nprevious task-level independent learning. The dedicated optimization of\ndegraded singular vectors and singular values inherently utilizes the potential\nrelationship among diverse restoration tasks, attributing to the Decomposition\nAscribed Synergistic Learning (DASL). Specifically, DASL comprises two\neffective operators, namely, Singular VEctor Operator (SVEO) and Singular VAlue\nOperator (SVAO), to favor the decomposed optimization, which can be lightly\nintegrated into existing image restoration backbone. Moreover, the congruous\ndecomposition loss has been devised for auxiliary. Extensive experiments on\nblended five image restoration tasks demonstrate the effectiveness of our\nmethod.\n", "date": "2023-08-01", "categories": []}
{"id": "2403.07026", "title": "Whiteness-based bilevel learning of regularization parameters in imaging", "abstract": "  We consider an unsupervised bilevel optimization strategy for learning\nregularization parameters in the context of imaging inverse problems in the\npresence of additive white Gaussian noise. Compared to supervised and\nsemi-supervised metrics relying either on the prior knowledge of reference data\nand/or on some (partial) knowledge on the noise statistics, the proposed\napproach optimizes the whiteness of the residual between the observed data and\nthe observation model with no need of ground-truth data.We validate the\napproach on standard Total Variation-regularized image deconvolution problems\nwhich show that the proposed quality metric provides estimates close to the\nmean-square error oracle and to discrepancy-based principles.\n", "date": "2024-03-10", "categories": []}
{"id": "2403.07390", "title": "Learning Correction Errors via Frequency-Self Attention for Blind Image\n  Super-Resolution", "abstract": "  Previous approaches for blind image super-resolution (SR) have relied on\ndegradation estimation to restore high-resolution (HR) images from their\nlow-resolution (LR) counterparts. However, accurate degradation estimation\nposes significant challenges. The SR model's incompatibility with degradation\nestimation methods, particularly the Correction Filter, may significantly\nimpair performance as a result of correction errors. In this paper, we\nintroduce a novel blind SR approach that focuses on Learning Correction Errors\n(LCE). Our method employs a lightweight Corrector to obtain a corrected\nlow-resolution (CLR) image. Subsequently, within an SR network, we jointly\noptimize SR performance by utilizing both the original LR image and the\nfrequency learning of the CLR image. Additionally, we propose a new\nFrequency-Self Attention block (FSAB) that enhances the global information\nutilization ability of Transformer. This block integrates both self-attention\nand frequency spatial attention mechanisms. Extensive ablation and comparison\nexperiments conducted across various settings demonstrate the superiority of\nour method in terms of visual quality and accuracy. Our approach effectively\naddresses the challenges associated with degradation estimation and correction\nerrors, paving the way for more accurate blind image SR.\n", "date": "2024-03-12", "categories": []}
{"id": "2307.12070", "title": "Fast and Stable Diffusion Inverse Solver with History Gradient Update", "abstract": "  Diffusion models have recently been recognised as efficient inverse problem\nsolvers due to their ability to produce high-quality reconstruction results\nwithout relying on pairwise data training. Existing diffusion-based solvers\nutilize Gradient Descent strategy to get a optimal sample solution. However,\nthese solvers only calculate the current gradient and have not utilized any\nhistory information of sampling process, thus resulting in unstable\noptimization progresses and suboptimal solutions. To address this issue, we\npropose to utilize the history information of the diffusion-based inverse\nsolvers. In this paper, we first prove that, in previous work, using the\ngradient descent method to optimize the data fidelity term is convergent.\nBuilding on this, we introduce the incorporation of historical gradients into\nthis optimization process, termed History Gradient Update (HGU). We also\nprovide theoretical evidence that HGU ensures the convergence of the entire\nalgorithm. It's worth noting that HGU is applicable to both pixel-based and\nlatent-based diffusion model solvers. Experimental results demonstrate that,\ncompared to previous sampling algorithms, sampling algorithms with HGU achieves\nstate-of-the-art results in medical image reconstruction, surpassing even\nsupervised learning methods. Additionally, it achieves competitive results on\nnatural images.\n", "date": "2023-07-22", "categories": []}
{"id": "2311.11638", "title": "Reti-Diff: Illumination Degradation Image Restoration with Retinex-based\n  Latent Diffusion Model", "abstract": "  Illumination degradation image restoration (IDIR) techniques aim to improve\nthe visibility of degraded images and mitigate the adverse effects of\ndeteriorated illumination. Among these algorithms, diffusion model (DM)-based\nmethods have shown promising performance but are often burdened by heavy\ncomputational demands and pixel misalignment issues when predicting the\nimage-level distribution. To tackle these problems, we propose to leverage DM\nwithin a compact latent space to generate concise guidance priors and introduce\na novel solution called Reti-Diff for the IDIR task. Reti-Diff comprises two\nkey components: the Retinex-based latent DM (RLDM) and the Retinex-guided\ntransformer (RGformer). To ensure detailed reconstruction and illumination\ncorrection, RLDM is empowered to acquire Retinex knowledge and extract\nreflectance and illumination priors. These priors are subsequently utilized by\nRGformer to guide the decomposition of image features into their respective\nreflectance and illumination components. Following this, RGformer further\nenhances and consolidates the decomposed features, resulting in the production\nof refined images with consistent content and robustness to handle complex\ndegradation scenarios. Extensive experiments show that Reti-Diff outperforms\nexisting methods on three IDIR tasks, as well as downstream applications. Code\nwill be available at \\url{https://github.com/ChunmingHe/Reti-Diff}.\n", "date": "2023-11-20", "categories": []}
{"id": "2403.05660", "title": "Decoupling Degradations with Recurrent Network for Video Restoration in\n  Under-Display Camera", "abstract": "  Under-display camera (UDC) systems are the foundation of full-screen display\ndevices in which the lens mounts under the display. The pixel array of\nlight-emitting diodes used for display diffracts and attenuates incident light,\ncausing various degradations as the light intensity changes. Unlike general\nvideo restoration which recovers video by treating different degradation\nfactors equally, video restoration for UDC systems is more challenging that\nconcerns removing diverse degradation over time while preserving temporal\nconsistency. In this paper, we introduce a novel video restoration network,\ncalled D$^2$RNet, specifically designed for UDC systems. It employs a set of\nDecoupling Attention Modules (DAM) that effectively separate the various video\ndegradation factors. More specifically, a soft mask generation function is\nproposed to formulate each frame into flare and haze based on the diffraction\narising from incident light of different intensities, followed by the proposed\nflare and haze removal components that leverage long- and short-term feature\nlearning to handle the respective degradations. Such a design offers an\ntargeted and effective solution to eliminating various types of degradation in\nUDC systems. We further extend our design into multi-scale to overcome the\nscale-changing of degradation that often occur in long-range videos. To\ndemonstrate the superiority of D$^2$RNet, we propose a large-scale UDC video\nbenchmark by gathering HDR videos and generating realistically degraded videos\nusing the point spread function measured by a commercial UDC system. Extensive\nquantitative and qualitative evaluations demonstrate the superiority of\nD$^2$RNet compared to other state-of-the-art video restoration and UDC image\nrestoration methods. Code is available at\nhttps://github.com/ChengxuLiu/DDRNet.git\n", "date": "2024-03-08", "categories": []}
{"id": "2403.05886", "title": "Generalizing to Out-of-Sample Degradations via Model Reprogramming", "abstract": "  Existing image restoration models are typically designed for specific tasks\nand struggle to generalize to out-of-sample degradations not encountered during\ntraining. While zero-shot methods can address this limitation by fine-tuning\nmodel parameters on testing samples, their effectiveness relies on predefined\nnatural priors and physical models of specific degradations. Nevertheless,\ndetermining out-of-sample degradations faced in real-world scenarios is always\nimpractical. As a result, it is more desirable to train restoration models with\ninherent generalization ability. To this end, this work introduces the\nOut-of-Sample Restoration (OSR) task, which aims to develop restoration models\ncapable of handling out-of-sample degradations. An intuitive solution involves\npre-translating out-of-sample degradations to known degradations of restoration\nmodels. However, directly translating them in the image space could lead to\ncomplex image translation issues. To address this issue, we propose a model\nreprogramming framework, which translates out-of-sample degradations by quantum\nmechanic and wave functions. Specifically, input images are decoupled as wave\nfunctions of amplitude and phase terms. The translation of out-of-sample\ndegradation is performed by adapting the phase term. Meanwhile, the image\ncontent is maintained and enhanced in the amplitude term. By taking these two\nterms as inputs, restoration models are able to handle out-of-sample\ndegradations without fine-tuning. Through extensive experiments across multiple\nevaluation cases, we demonstrate the effectiveness and flexibility of our\nproposed framework. Our codes are available at\n\\href{https://github.com/ddghjikle/Out-of-sample-restoration}{Github}.\n", "date": "2024-03-09", "categories": []}
{"id": "2403.06275", "title": "UNICORN: Ultrasound Nakagami Imaging via Score Matching and Adaptation", "abstract": "  Nakagami imaging holds promise for visualizing and quantifying tissue\nscattering in ultrasound waves, with potential applications in tumor diagnosis\nand fat fraction estimation which are challenging to discern by conventional\nultrasound B-mode images. Existing methods struggle with optimal window size\nselection and suffer from estimator instability, leading to degraded resolution\nimages. To address this, here we propose a novel method called UNICORN\n(Ultrasound Nakagami Imaging via Score Matching and Adaptation), that offers an\naccurate, closed-form estimator for Nakagami parameter estimation in terms of\nthe score function of ultrasonic envelope. Extensive experiments using\nsimulation and real ultrasound RF data demonstrate UNICORN's superiority over\nconventional approaches in accuracy and resolution quality.\n", "date": "2024-03-10", "categories": []}
{"id": "2403.04228", "title": "Single-Image HDR Reconstruction Assisted Ghost Suppression and Detail\n  Preservation Network for Multi-Exposure HDR Imaging", "abstract": "  The reconstruction of high dynamic range (HDR) images from multi-exposure low\ndynamic range (LDR) images in dynamic scenes presents significant challenges,\nespecially in preserving and restoring information in oversaturated regions and\navoiding ghosting artifacts. While current methods often struggle to address\nthese challenges, our work aims to bridge this gap by developing a\nmulti-exposure HDR image reconstruction network for dynamic scenes,\ncomplemented by single-frame HDR image reconstruction. This network, comprising\nsingle-frame HDR reconstruction with enhanced stop image (SHDR-ESI) and\nSHDR-ESI-assisted multi-exposure HDR reconstruction (SHDRA-MHDR), effectively\nleverages the ghost-free characteristic of single-frame HDR reconstruction and\nthe detail-enhancing capability of ESI in oversaturated areas. Specifically,\nSHDR-ESI innovatively integrates single-frame HDR reconstruction with the\nutilization of ESI. This integration not only optimizes the single image HDR\nreconstruction process but also effectively guides the synthesis of\nmulti-exposure HDR images in SHDR-AMHDR. In this method, the single-frame HDR\nreconstruction is specifically applied to reduce potential ghosting effects in\nmultiexposure HDR synthesis, while the use of ESI images assists in enhancing\nthe detail information in the HDR synthesis process. Technically, SHDR-ESI\nincorporates a detail enhancement mechanism, which includes a\nself-representation module and a mutual-representation module, designed to\naggregate crucial information from both reference image and ESI. To fully\nleverage the complementary information from non-reference images, a feature\ninteraction fusion module is integrated within SHDRA-MHDR. Additionally, a\nghost suppression module, guided by the ghost-free results of SHDR-ESI, is\nemployed to suppress the ghosting artifacts.\n", "date": "2024-03-07", "categories": []}
{"id": "2301.00406", "title": "Curvature regularization for Non-line-of-sight Imaging from\n  Under-sampled Data", "abstract": "  Non-line-of-sight (NLOS) imaging aims to reconstruct the three-dimensional\nhidden scenes from the data measured in the line-of-sight, which uses photon\ntime-of-flight information encoded in light after multiple diffuse reflections.\nThe under-sampled scanning data can facilitate fast imaging. However, the\nresulting reconstruction problem becomes a serious ill-posed inverse problem,\nthe solution of which is highly possibility to be degraded due to noises and\ndistortions. In this paper, we propose novel NLOS reconstruction models based\non curvature regularization, i.e., the object-domain curvature regularization\nmodel and the dual (signal and object)-domain curvature regularization model.\nIn what follows, we develop efficient optimization algorithms relying on the\nalternating direction method of multipliers (ADMM) with the backtracking\nstepsize rule, for which all solvers can be implemented on GPUs. We evaluate\nthe proposed algorithms on both synthetic and real datasets, which achieve\nstate-of-the-art performance, especially in the compressed sensing setting.\nBased on GPU computing, our algorithm is the most effective among iterative\nmethods, balancing reconstruction quality and computational time. All our codes\nand data are available at https://github.com/Duanlab123/CurvNLOS.\n", "date": "2023-01-01", "categories": []}
{"id": "2403.03447", "title": "HDRFlow: Real-Time HDR Video Reconstruction with Large Motions", "abstract": "  Reconstructing High Dynamic Range (HDR) video from image sequences captured\nwith alternating exposures is challenging, especially in the presence of large\ncamera or object motion. Existing methods typically align low dynamic range\nsequences using optical flow or attention mechanism for deghosting. However,\nthey often struggle to handle large complex motions and are computationally\nexpensive. To address these challenges, we propose a robust and efficient flow\nestimator tailored for real-time HDR video reconstruction, named HDRFlow.\nHDRFlow has three novel designs: an HDR-domain alignment loss (HALoss), an\nefficient flow network with a multi-size large kernel (MLK), and a new HDR flow\ntraining scheme. The HALoss supervises our flow network to learn an\nHDR-oriented flow for accurate alignment in saturated and dark regions. The MLK\ncan effectively model large motions at a negligible cost. In addition, we\nincorporate synthetic data, Sintel, into our training dataset, utilizing both\nits provided forward flow and backward flow generated by us to supervise our\nflow network, enhancing our performance in large motion regions. Extensive\nexperiments demonstrate that our HDRFlow outperforms previous methods on\nstandard benchmarks. To the best of our knowledge, HDRFlow is the first\nreal-time HDR video reconstruction method for video sequences captured with\nalternating exposures, capable of processing 720p resolution inputs at 25ms.\n", "date": "2024-03-06", "categories": []}
{"id": "2403.03896", "title": "DART: Implicit Doppler Tomography for Radar Novel View Synthesis", "abstract": "  Simulation is an invaluable tool for radio-frequency system designers that\nenables rapid prototyping of various algorithms for imaging, target detection,\nclassification, and tracking. However, simulating realistic radar scans is a\nchallenging task that requires an accurate model of the scene, radio frequency\nmaterial properties, and a corresponding radar synthesis function. Rather than\nspecifying these models explicitly, we propose DART - Doppler Aided Radar\nTomography, a Neural Radiance Field-inspired method which uses radar-specific\nphysics to create a reflectance and transmittance-based rendering pipeline for\nrange-Doppler images. We then evaluate DART by constructing a custom data\ncollection platform and collecting a novel radar dataset together with accurate\nposition and instantaneous velocity measurements from lidar-based localization.\nIn comparison to state-of-the-art baselines, DART synthesizes superior radar\nrange-Doppler images from novel views across all datasets and additionally can\nbe used to generate high quality tomographic images.\n", "date": "2024-03-06", "categories": []}
{"id": "2311.00434", "title": "Event-based Background-Oriented Schlieren", "abstract": "  Schlieren imaging is an optical technique to observe the flow of transparent\nmedia, such as air or water, without any particle seeding. However,\nconventional frame-based techniques require both high spatial and temporal\nresolution cameras, which impose bright illumination and expensive computation\nlimitations. Event cameras offer potential advantages (high dynamic range, high\ntemporal resolution, and data efficiency) to overcome such limitations due to\ntheir bio-inspired sensing principle. This paper presents a novel technique for\nperceiving air convection using events and frames by providing the first\ntheoretical analysis that connects event data and schlieren. We formulate the\nproblem as a variational optimization one combining the linearized event\ngeneration model with a physically-motivated parameterization that estimates\nthe temporal derivative of the air density. The experiments with accurately\naligned frame- and event camera data reveal that the proposed method enables\nevent cameras to obtain on par results with existing frame-based optical flow\ntechniques. Moreover, the proposed method works under dark conditions where\nframe-based schlieren fails, and also enables slow-motion analysis by\nleveraging the event camera's advantages. Our work pioneers and opens a new\nstack of event camera applications, as we publish the source code as well as\nthe first schlieren dataset with high-quality frame and event data.\nhttps://github.com/tub-rip/event_based_bos\n", "date": "2023-11-01", "categories": []}
{"id": "2403.00897", "title": "VisRec: A Semi-Supervised Approach to Radio Interferometric Data\n  Reconstruction", "abstract": "  Radio telescopes produce visibility data about celestial objects, but these\ndata are sparse and noisy. As a result, images created on raw visibility data\nare of low quality. Recent studies have used deep learning models to\nreconstruct visibility data to get cleaner images. However, these methods rely\non a substantial amount of labeled training data, which requires significant\nlabeling effort from radio astronomers. Addressing this challenge, we propose\nVisRec, a model-agnostic semi-supervised learning approach to the\nreconstruction of visibility data. Specifically, VisRec consists of both a\nsupervised learning module and an unsupervised learning module. In the\nsupervised learning module, we introduce a set of data augmentation functions\nto produce diverse training examples. In comparison, the unsupervised learning\nmodule in VisRec augments unlabeled data and uses reconstructions from\nnon-augmented visibility data as pseudo-labels for training. This hybrid\napproach allows VisRec to effectively leverage both labeled and unlabeled data.\nThis way, VisRec performs well even when labeled data is scarce. Our evaluation\nresults show that VisRec outperforms all baseline methods in reconstruction\nquality, robustness against common observation perturbation, and\ngeneralizability to different telescope configurations.\n", "date": "2024-03-01", "categories": []}
{"id": "2403.01692", "title": "PI-AstroDeconv: A Physics-Informed Unsupervised Learning Method for\n  Astronomical Image Deconvolution", "abstract": "  In the imaging process of an astronomical telescope, the deconvolution of its\nbeam or Point Spread Function (PSF) is a crucial task. However, deconvolution\npresents a classical and challenging inverse computation problem. In scenarios\nwhere the beam or PSF is complex or inaccurately measured, such as in\ninterferometric arrays and certain radio telescopes, the resultant blurry\nimages are often challenging to interpret visually or analyze using traditional\nphysical detection methods. We argue that traditional methods frequently lack\nspecific prior knowledge, thereby leading to suboptimal performance. To address\nthis issue and achieve image deconvolution and reconstruction, we propose an\nunsupervised network architecture that incorporates prior physical information.\nThe network adopts an encoder-decoder structure while leveraging the\ntelescope's PSF as prior knowledge. During network training, we introduced\naccelerated Fast Fourier Transform (FFT) convolution to enable efficient\nprocessing of high-resolution input images and PSFs. We explored various\nclassic regression networks, including autoencoder (AE) and U-Net, and\nconducted a comprehensive performance evaluation through comparative analysis.\n", "date": "2024-03-04", "categories": []}
{"id": "2403.00402", "title": "Spatio-temporal reconstruction of substance dynamics using compressed\n  sensing in multi-spectral magnetic resonance spectroscopic imaging", "abstract": "  The objective of our study is to observe dynamics of multiple substances in\nvivo with high temporal resolution from multi-spectral magnetic resonance\nspectroscopic imaging (MRSI) data. The multi-spectral MRSI can effectively\nseparate spectral peaks of multiple substances and is useful to measure spatial\ndistributions of substances. However it is difficult to measure time-varying\nsubstance distributions directly by ordinary full sampling because the\nmeasurement requires a significantly long time. In this study, we propose a\nnovel method to reconstruct the spatio-temporal distributions of substances\nfrom randomly undersampled multi-spectral MRSI data on the basis of compressed\nsensing (CS) and the partially separable function model with base spectra of\nsubstances. In our method, we have employed spatio-temporal sparsity and\ntemporal smoothness of the substance distributions as prior knowledge to\nperform CS. The effectiveness of our method has been evaluated using phantom\ndata sets of glass tubes filled with glucose or lactate solution in increasing\namounts over time and animal data sets of a tumor-bearing mouse to observe the\nmetabolic dynamics involved in the Warburg effect in vivo. The reconstructed\nresults are consistent with the expected behaviors, showing that our method can\nreconstruct the spatio-temporal distribution of substances with a temporal\nresolution of four seconds which is extremely short time scale compared with\nthat of full sampling. Since this method utilizes only prior knowledge\nnaturally assumed for the spatio-temporal distributions of substances and is\nindependent of the number of the spectral and spatial dimensions or the\nacquisition sequence of MRSI, it is expected to contribute to revealing the\nunderlying substance dynamics in MRSI data already acquired or to be acquired\nin the future.\n", "date": "2024-03-01", "categories": []}
{"id": "2403.00426", "title": "Deep Learning Computed Tomography based on the Defrise and Clack\n  Algorithm", "abstract": "  This study presents a novel approach for reconstructing cone beam computed\ntomography (CBCT) for specific orbits using known operator learning. Unlike\ntraditional methods, this technique employs a filtered backprojection type\n(FBP-type) algorithm, which integrates a unique, adaptive filtering process.\nThis process involves a series of operations, including weightings,\ndifferentiations, the 2D Radon transform, and backprojection. The filter is\ndesigned for a specific orbit geometry and is obtained using a data-driven\napproach based on deep learning. The approach efficiently learns and optimizes\nthe orbit-related component of the filter. The method has demonstrated its\nability through experimentation by successfully learning parameters from\ncircular orbit projection data. Subsequently, the optimized parameters are used\nto reconstruct images, resulting in outcomes that closely resemble the\nanalytical solution. This demonstrates the potential of the method to learn\nappropriate parameters from any specific orbit projection data and achieve\nreconstruction. The algorithm has demonstrated improvement, particularly in\nenhancing reconstruction speed and reducing memory usage for handling specific\norbit reconstruction.\n", "date": "2024-03-01", "categories": []}
{"id": "2308.08567", "title": "CMISR: Circular Medical Image Super-Resolution", "abstract": "  Classical methods of medical image super-resolution (MISR) utilize open-loop\narchitecture with implicit under-resolution (UR) unit and explicit\nsuper-resolution (SR) unit. The UR unit can always be given, assumed, or\nestimated, while the SR unit is elaborately designed according to various SR\nalgorithms. The closed-loop feedback mechanism is widely employed in current\nMISR approaches and can efficiently improve their performance. The feedback\nmechanism may be divided into two categories: local feedback and global\nfeedback. Therefore, this paper proposes a global feedback-based closed-cycle\nframework, circular MISR (CMISR), with unambiguous UR and advanced SR elements.\nMathematical model and closed-loop equation of CMISR are built. Mathematical\nproof with Taylor-series approximation indicates that CMISR has zero recovery\nerror in steady-state. In addition, CMISR holds plug-and-play characteristic\nthat fuses model-based and learning-based approaches and can be established on\nany existing MISR algorithms. Five CMISR algorithms are respectively proposed\nbased on the state-of-the-art open-loop MISR algorithms. Experimental results\nwith three scale factors and on three open medical image datasets show that\nCMISR is superior to MISR in reconstruction performance and is particularly\nsuited to medical images with strong edges or intense contrast.\n", "date": "2023-08-15", "categories": []}
{"id": "2402.19111", "title": "Deep Network for Image Compressed Sensing Coding Using Local Structural\n  Sampling", "abstract": "  Existing image compressed sensing (CS) coding frameworks usually solve an\ninverse problem based on measurement coding and optimization-based image\nreconstruction, which still exist the following two challenges: 1) The widely\nused random sampling matrix, such as the Gaussian Random Matrix (GRM), usually\nleads to low measurement coding efficiency. 2) The optimization-based\nreconstruction methods generally maintain a much higher computational\ncomplexity. In this paper, we propose a new CNN based image CS coding framework\nusing local structural sampling (dubbed CSCNet) that includes three functional\nmodules: local structural sampling, measurement coding and Laplacian pyramid\nreconstruction. In the proposed framework, instead of GRM, a new local\nstructural sampling matrix is first developed, which is able to enhance the\ncorrelation between the measurements through a local perceptual sampling\nstrategy. Besides, the designed local structural sampling matrix can be jointly\noptimized with the other functional modules during training process. After\nsampling, the measurements with high correlations are produced, which are then\ncoded into final bitstreams by the third-party image codec. At last, a\nLaplacian pyramid reconstruction network is proposed to efficiently recover the\ntarget image from the measurement domain to the image domain. Extensive\nexperimental results demonstrate that the proposed scheme outperforms the\nexisting state-of-the-art CS coding methods, while maintaining fast\ncomputational speed.\n", "date": "2024-02-29", "categories": []}
{"id": "2310.01018", "title": "Controlling Vision-Language Models for Multi-Task Image Restoration", "abstract": "  Vision-language models such as CLIP have shown great impact on diverse\ndownstream tasks for zero-shot or label-free predictions. However, when it\ncomes to low-level vision such as image restoration their performance\ndeteriorates dramatically due to corrupted inputs. In this paper, we present a\ndegradation-aware vision-language model (DA-CLIP) to better transfer pretrained\nvision-language models to low-level vision tasks as a multi-task framework for\nimage restoration. More specifically, DA-CLIP trains an additional controller\nthat adapts the fixed CLIP image encoder to predict high-quality feature\nembeddings. By integrating the embedding into an image restoration network via\ncross-attention, we are able to pilot the model to learn a high-fidelity image\nreconstruction. The controller itself will also output a degradation feature\nthat matches the real corruptions of the input, yielding a natural classifier\nfor different degradation types. In addition, we construct a mixed degradation\ndataset with synthetic captions for DA-CLIP training. Our approach advances\nstate-of-the-art performance on both \\emph{degradation-specific} and\n\\emph{unified} image restoration tasks, showing a promising direction of\nprompting image restoration with large-scale pretrained vision-language models.\nOur code is available at https://github.com/Algolzw/daclip-uir.\n", "date": "2023-10-02", "categories": []}
{"id": "2310.01840", "title": "Self-Supervised High Dynamic Range Imaging with Multi-Exposure Images in\n  Dynamic Scenes", "abstract": "  Merging multi-exposure images is a common approach for obtaining high dynamic\nrange (HDR) images, with the primary challenge being the avoidance of ghosting\nartifacts in dynamic scenes. Recent methods have proposed using deep neural\nnetworks for deghosting. However, the methods typically rely on sufficient data\nwith HDR ground-truths, which are difficult and costly to collect. In this\nwork, to eliminate the need for labeled data, we propose SelfHDR, a\nself-supervised HDR reconstruction method that only requires dynamic\nmulti-exposure images during training. Specifically, SelfHDR learns a\nreconstruction network under the supervision of two complementary components,\nwhich can be constructed from multi-exposure images and focus on HDR color as\nwell as structure, respectively. The color component is estimated from aligned\nmulti-exposure images, while the structure one is generated through a\nstructure-focused network that is supervised by the color component and an\ninput reference (\\eg, medium-exposure) image. During testing, the learned\nreconstruction network is directly deployed to predict an HDR image.\nExperiments on real-world images demonstrate our SelfHDR achieves superior\nresults against the state-of-the-art self-supervised methods, and comparable\nperformance to supervised ones. Codes are available at\nhttps://github.com/cszhilu1998/SelfHDR\n", "date": "2023-10-03", "categories": []}
{"id": "2402.17960", "title": "Rapid hyperspectral photothermal mid-infrared spectroscopic imaging from\n  sparse data for gynecologic cancer tissue subtyping", "abstract": "  Ovarian cancer detection has traditionally relied on a multi-step process\nthat includes biopsy, tissue staining, and morphological analysis by\nexperienced pathologists. While widely practiced, this conventional approach\nsuffers from several drawbacks: it is qualitative, time-intensive, and heavily\ndependent on the quality of staining. Mid-infrared (MIR) hyperspectral\nphotothermal imaging is a label-free, biochemically quantitative technology\nthat, when combined with machine learning algorithms, can eliminate the need\nfor staining and provide quantitative results comparable to traditional\nhistology. However, this technology is slow. This work presents a novel\napproach to MIR photothermal imaging that enhances its speed by an order of\nmagnitude. Our method significantly accelerates data collection by capturing a\ncombination of high-resolution and interleaved, lower-resolution infrared band\nimages and applying computational techniques for data interpolation. We\neffectively minimize data collection requirements by leveraging sparse data\nacquisition and employing curvelet-based reconstruction algorithms. This method\nenables the reconstruction of high-quality, high-resolution images from\nundersampled datasets and achieving a 10X improvement in data acquisition time.\nWe assessed the performance of our sparse imaging methodology using a variety\nof quantitative metrics, including mean squared error (MSE), structural\nsimilarity index (SSIM), and tissue subtype classification accuracies,\nemploying both random forest and convolutional neural network (CNN) models,\naccompanied by ROC curves. Our statistically robust analysis, based on data\nfrom 100 ovarian cancer patient samples and over 65 million data points,\ndemonstrates the method's capability to produce superior image quality and\naccurately distinguish between different gynecological tissue types with\nsegmentation accuracy exceeding 95%.\n", "date": "2024-02-28", "categories": []}
{"id": "2402.18175", "title": "Self-Supervised Spatially Variant PSF Estimation for Aberration-Aware\n  Depth-from-Defocus", "abstract": "  In this paper, we address the task of aberration-aware depth-from-defocus\n(DfD), which takes account of spatially variant point spread functions (PSFs)\nof a real camera. To effectively obtain the spatially variant PSFs of a real\ncamera without requiring any ground-truth PSFs, we propose a novel\nself-supervised learning method that leverages the pair of real sharp and\nblurred images, which can be easily captured by changing the aperture setting\nof the camera. In our PSF estimation, we assume rotationally symmetric PSFs and\nintroduce the polar coordinate system to more accurately learn the PSF\nestimation network. We also handle the focus breathing phenomenon that occurs\nin real DfD situations. Experimental results on synthetic and real data\ndemonstrate the effectiveness of our method regarding both the PSF estimation\nand the depth estimation.\n", "date": "2024-02-28", "categories": []}
{"id": "2312.12439", "title": "Single-pixel 3D imaging based on fusion temporal data of single photon\n  detector and millimeter-wave radar", "abstract": "  Recently, there has been increased attention towards 3D imaging using\nsingle-pixel single-photon detection (also known as temporal data) due to its\npotential advantages in terms of cost and power efficiency. However, to\neliminate the symmetry blur in the reconstructed images, a fixed background is\nrequired. This paper proposes a fusion-data-based 3D imaging method that\nutilizes a single-pixel single-photon detector and a millimeter-wave radar to\ncapture temporal histograms of a scene from multiple perspectives.\nSubsequently, the 3D information can be reconstructed from the one-dimensional\nfusion temporal data by using Artificial Neural Network (ANN). Both the\nsimulation and experimental results demonstrate that our fusion method\neffectively eliminates symmetry blur and improves the quality of the\nreconstructed images.\n", "date": "2023-10-20", "categories": []}
{"id": "2304.14491", "title": "Enhancing Electrical Impedance Tomography reconstruction using Learned\n  Half-Quadratic Splitting Networks with Anderson Acceleration", "abstract": "  Electrical Impedance Tomography (EIT) is widely applied in medical diagnosis,\nindustrial inspection, and environmental monitoring. Combining the physical\nprinciples of the imaging system with the advantages of data-driven deep\nlearning networks, physics-embedded deep unrolling networks have recently\nemerged as a promising solution in computational imaging. However, the inherent\nnonlinear and ill-posed properties of EIT image reconstruction still present\nchallenges to existing methods in terms of accuracy and stability. To tackle\nthis challenge, we propose the learned half-quadratic splitting (HQSNet)\nalgorithm for incorporating physics into learning-based EIT imaging. We then\napply Anderson acceleration (AA) to the HQSNet algorithm, denoted as AA-HQSNet,\nwhich can be interpreted as AA applied to the Gauss-Newton step and the learned\nproximal gradient descent step of the HQSNet, respectively. AA is a widely-used\ntechnique for accelerating the convergence of fixed-point iterative algorithms\nand has gained significant interest in numerical optimization and machine\nlearning. However, the technique has received little attention in the inverse\nproblems community thus far. Employing AA enhances the convergence rate\ncompared to the standard HQSNet while simultaneously avoiding artifacts in the\nreconstructions. Lastly, we conduct rigorous numerical and visual experiments\nto show that the AA module strengthens the HQSNet, leading to robust, accurate,\nand considerably superior reconstructions compared to state-of-the-art methods.\nOur Anderson acceleration scheme to enhance HQSNet is generic and can be\napplied to improve the performance of various physics-embedded deep learning\nmethods.\n", "date": "2023-04-16", "categories": []}
{"id": "2310.02687", "title": "USB-NeRF: Unrolling Shutter Bundle Adjusted Neural Radiance Fields", "abstract": "  Neural Radiance Fields (NeRF) has received much attention recently due to its\nimpressive capability to represent 3D scene and synthesize novel view images.\nExisting works usually assume that the input images are captured by a global\nshutter camera. Thus, rolling shutter (RS) images cannot be trivially applied\nto an off-the-shelf NeRF algorithm for novel view synthesis. Rolling shutter\neffect would also affect the accuracy of the camera pose estimation (e.g. via\nCOLMAP), which further prevents the success of NeRF algorithm with RS images.\nIn this paper, we propose Unrolling Shutter Bundle Adjusted Neural Radiance\nFields (USB-NeRF). USB-NeRF is able to correct rolling shutter distortions and\nrecover accurate camera motion trajectory simultaneously under the framework of\nNeRF, by modeling the physical image formation process of a RS camera.\nExperimental results demonstrate that USB-NeRF achieves better performance\ncompared to prior works, in terms of RS effect removal, novel view image\nsynthesis as well as camera motion estimation. Furthermore, our algorithm can\nalso be used to recover high-fidelity high frame-rate global shutter video from\na sequence of RS images.\n", "date": "2023-10-04", "categories": []}
{"id": "2402.15750", "title": "Design, Implementation and Analysis of a Compressed Sensing\n  Photoacoustic Projection Imaging System", "abstract": "  Significance: Compressed sensing (CS) uses special measurement designs\ncombined with powerful mathematical algorithms to reduce the amount of data to\nbe collected while maintaining image quality. This is relevant to almost any\nimaging modality, and in this paper we focus on CS in photoacoustic projection\nimaging (PAPI) with integrating line detectors (ILDs).\n  Aim: Our previous research involved rather general CS measurements, where\neach ILD can contribute to any measurement. In the real world, however, the\ndesign of CS measurements is subject to practical constraints. In this\nresearch, we aim at a CS-PAPI system where each measurement involves only a\nsubset of ILDs, and which can be implemented in a cost-effective manner.\n  Approach: We extend the existing PAPI with a self-developed CS unit. The\nsystem provides structured CS matrices for which the existing recovery theory\ncannot be applied directly. A random search strategy is applied to select the\nCS measurement matrix within this class for which we obtain exact sparse\nrecovery.\n  Results: We implement a CS PAPI system for a compression factor of $4:3$,\nwhere specific measurements are made on separate groups of 16 ILDs. We\nalgorithmically design optimal CS measurements that have proven sparse CS\ncapabilities. Numerical experiments are used to support our results.\n  Conclusions: CS with proven sparse recovery capabilities can be integrated\ninto PAPI, and numerical results support this setup. Future work will focus on\napplying it to experimental data and utilizing data-driven approaches to\nenhance the compression factor and generalize the signal class.\n", "date": "2024-02-24", "categories": []}
{"id": "2402.13816", "title": "A unified framework of non-local parametric methods for image denoising", "abstract": "  We propose a unified view of non-local methods for single-image denoising,\nfor which BM3D is the most popular representative, that operate by gathering\nnoisy patches together according to their similarities in order to process them\ncollaboratively. Our general estimation framework is based on the minimization\nof the quadratic risk, which is approximated in two steps, and adapts to photon\nand electronic noises. Relying on unbiased risk estimation (URE) for the first\nstep and on ``internal adaptation'', a concept borrowed from deep learning\ntheory, for the second, we show that our approach enables to reinterpret and\nreconcile previous state-of-the-art non-local methods. Within this framework,\nwe propose a novel denoiser called NL-Ridge that exploits linear combinations\nof patches. While conceptually simpler, we show that NL-Ridge can outperform\nwell-established state-of-the-art single-image denoisers.\n", "date": "2024-02-21", "categories": []}
{"id": "2402.10204", "title": "Radio-astronomical Image Reconstruction with Conditional Denoising\n  Diffusion Model", "abstract": "  Reconstructing sky models from dirty radio images for accurate source\nlocalization and flux estimation is crucial for studying galaxy evolution at\nhigh redshift, especially in deep fields using instruments like the Atacama\nLarge Millimetre Array (ALMA). With new projects like the Square Kilometre\nArray (SKA), there's a growing need for better source extraction methods.\nCurrent techniques, such as CLEAN and PyBDSF, often fail to detect faint\nsources, highlighting the need for more accurate methods. This study proposes\nusing stochastic neural networks to rebuild sky models directly from dirty\nimages. This method can pinpoint radio sources and measure their fluxes with\nrelated uncertainties, marking a potential improvement in radio source\ncharacterization. We tested this approach on 10164 images simulated with the\nCASA tool simalma, based on ALMA's Cycle 5.3 antenna setup. We applied\nconditional Denoising Diffusion Probabilistic Models (DDPMs) for sky models\nreconstruction, then used Photutils to determine source coordinates and fluxes,\nassessing the model's performance across different water vapor levels. Our\nmethod showed excellence in source localization, achieving more than 90%\ncompleteness at a signal-to-noise ratio (SNR) as low as 2. It also surpassed\nPyBDSF in flux estimation, accurately identifying fluxes for 96% of sources in\nthe test set, a significant improvement over CLEAN+ PyBDSF's 57%. Conditional\nDDPMs is a powerful tool for image-to-image translation, yielding accurate and\nrobust characterisation of radio sources, and outperforming existing\nmethodologies. While this study underscores its significant potential for\napplications in radio astronomy, we also acknowledge certain limitations that\naccompany its usage, suggesting directions for further refinement and research.\n", "date": "2024-02-15", "categories": []}
{"id": "2303.05754", "title": "Decomposed Diffusion Sampler for Accelerating Large-Scale Inverse\n  Problems", "abstract": "  Krylov subspace, which is generated by multiplying a given vector by the\nmatrix of a linear transformation and its successive powers, has been\nextensively studied in classical optimization literature to design algorithms\nthat converge quickly for large linear inverse problems. For example, the\nconjugate gradient method (CG), one of the most popular Krylov subspace\nmethods, is based on the idea of minimizing the residual error in the Krylov\nsubspace. However, with the recent advancement of high-performance diffusion\nsolvers for inverse problems, it is not clear how classical wisdom can be\nsynergistically combined with modern diffusion models. In this study, we\npropose a novel and efficient diffusion sampling strategy that synergistically\ncombines the diffusion sampling and Krylov subspace methods. Specifically, we\nprove that if the tangent space at a denoised sample by Tweedie's formula forms\na Krylov subspace, then the CG initialized with the denoised data ensures the\ndata consistency update to remain in the tangent space. This negates the need\nto compute the manifold-constrained gradient (MCG), leading to a more efficient\ndiffusion sampling method. Our method is applicable regardless of the\nparametrization and setting (i.e., VE, VP). Notably, we achieve\nstate-of-the-art reconstruction quality on challenging real-world medical\ninverse imaging problems, including multi-coil MRI reconstruction and 3D CT\nreconstruction. Moreover, our proposed method achieves more than 80 times\nfaster inference time than the previous state-of-the-art method. Code is\navailable at https://github.com/HJ-harry/DDS\n", "date": "2023-03-10", "categories": []}
{"id": "2402.12292", "title": "Regularization by denoising: Bayesian model and Langevin-within-split\n  Gibbs sampling", "abstract": "  This paper introduces a Bayesian framework for image inversion by deriving a\nprobabilistic counterpart to the regularization-by-denoising (RED) paradigm. It\nadditionally implements a Monte Carlo algorithm specifically tailored for\nsampling from the resulting posterior distribution, based on an asymptotically\nexact data augmentation (AXDA). The proposed algorithm is an approximate\ninstance of split Gibbs sampling (SGS) which embeds one Langevin Monte Carlo\nstep. The proposed method is applied to common imaging tasks such as\ndeblurring, inpainting and super-resolution, demonstrating its efficacy through\nextensive numerical experiments. These contributions advance Bayesian inference\nin imaging by leveraging data-driven regularization strategies within a\nprobabilistic framework.\n", "date": "2024-02-19", "categories": []}
{"id": "2402.04139", "title": "U-shaped Vision Mamba for Single Image Dehazing", "abstract": "  Currently, Transformer is the most popular architecture for image dehazing,\nbut due to its large computational complexity, its ability to handle long-range\ndependency is limited on resource-constrained devices. To tackle this\nchallenge, we introduce the U-shaped Vision Mamba (UVM-Net), an efficient\nsingle-image dehazing network. Inspired by the State Space Sequence Models\n(SSMs), a new deep sequence model known for its power to handle long sequences,\nwe design a Bi-SSM block that integrates the local feature extraction ability\nof the convolutional layer with the ability of the SSM to capture long-range\ndependencies. Extensive experimental results demonstrate the effectiveness of\nour method. Our method provides a more highly efficient idea of long-range\ndependency modeling for image dehazing as well as other image restoration\ntasks. The URL of the code is \\url{https://github.com/zzr-idam/UVM-Net}. Our\nmethod takes only \\textbf{0.009} seconds to infer a $325 \\times 325$ resolution\nimage (100FPS) without I/O handling time.\n", "date": "2024-02-06", "categories": []}
{"id": "2402.10223", "title": "Integer Optimization of CT Trajectories using a Discrete Data\n  Completeness Formulation", "abstract": "  X-ray computed tomography (CT) plays a key role in digitizing\nthree-dimensional structures for a wide range of medical and industrial\napplications. Traditional CT systems often rely on standard circular and\nhelical scan trajectories, which may not be optimal for challenging scenarios\ninvolving large objects, complex structures, or resource constraints. In\nresponse to these challenges, we are exploring the potential of twin robotic CT\nsystems, which offer the flexibility to acquire projections from arbitrary\nviews around the object of interest. Ensuring complete and mathematically sound\nreconstructions becomes critical in such systems. In this work, we present an\ninteger programming-based CT trajectory optimization method. Utilizing discrete\ndata completeness conditions, we formulate an optimization problem to select an\noptimized set of projections. This approach enforces data completeness and\nconsiders absorption-based metrics for reliability evaluation. We compare our\nmethod with an equidistant circular CT trajectory and a greedy approach. While\ngreedy already performs well in some cases, we provide a way to improve\ngreedy-based projection selection using an integer optimization approach. Our\napproach improves CT trajectories and quantifies the optimality of the solution\nin terms of an optimality gap.\n", "date": "2024-01-29", "categories": []}
{"id": "2402.10520", "title": "Real-Time Model-Based Quantitative Ultrasound and Radar", "abstract": "  Ultrasound and radar signals are highly beneficial for medical imaging as\nthey are non-invasive and non-ionizing. Traditional imaging techniques have\nlimitations in terms of contrast and physical interpretation. Quantitative\nmedical imaging can display various physical properties such as speed of sound,\ndensity, conductivity, and relative permittivity. This makes it useful for a\nwider range of applications, including improving cancer detection, diagnosing\nfatty liver, and fast stroke imaging. However, current quantitative imaging\ntechniques that estimate physical properties from received signals, such as\nFull Waveform Inversion, are time-consuming and tend to converge to local\nminima, making them unsuitable for medical imaging. To address these\nchallenges, we propose a neural network based on the physical model of wave\npropagation, which defines the relationship between the received signals and\nphysical properties. Our network can reconstruct multiple physical properties\nin less than one second for complex and realistic scenarios, using data from\nonly eight elements. We demonstrate the effectiveness of our approach for both\nradar and ultrasound signals.\n", "date": "2024-02-16", "categories": []}
{"id": "2402.10831", "title": "GAN-driven Electromagnetic Imaging of 2-D Dielectric Scatterers", "abstract": "  Inverse scattering problems are inherently challenging, given the fact they\nare ill-posed and nonlinear. This paper presents a powerful deep learning-based\napproach that relies on generative adversarial networks to accurately and\nefficiently reconstruct randomly-shaped two-dimensional dielectric objects from\namplitudes of multi-frequency scattered electric fields. An adversarial\nautoencoder (AAE) is trained to learn to generate the scatterer's geometry from\na lower-dimensional latent representation constrained to adhere to the Gaussian\ndistribution. A cohesive inverse neural network (INN) framework is set up\ncomprising a sequence of appropriately designed dense layers, the\nalready-trained generator as well as a separately trained forward neural\nnetwork. The images reconstructed at the output of the inverse network are\nvalidated through comparison with outputs from the forward neural network,\naddressing the non-uniqueness challenge inherent to electromagnetic (EM)\nimaging problems. The trained INN demonstrates an enhanced robustness,\nevidenced by a mean binary cross-entropy (BCE) loss of $0.13$ and a structure\nsimilarity index (SSI) of $0.90$. The study not only demonstrates a significant\nreduction in computational load, but also marks a substantial improvement over\ntraditional objective-function-based methods. It contributes both to the fields\nof machine learning and EM imaging by offering a real-time quantitative imaging\napproach. The results obtained with the simulated data, for both training and\ntesting, yield promising results and may open new avenues for radio-frequency\ninverse imaging.\n", "date": "2024-02-16", "categories": []}
{"id": "2209.09247", "title": "Weak-signal extraction enabled by deep-neural-network denoising of\n  diffraction data", "abstract": "  Removal or cancellation of noise has wide-spread applications for imaging and\nacoustics. In every-day-life applications, denoising may even include\ngenerative aspects, which are unfaithful to the ground truth. For scientific\nuse, however, denoising must reproduce the ground truth accurately. Here, we\nshow how data can be denoised via a deep convolutional neural network such that\nweak signals appear with quantitative accuracy. In particular, we study X-ray\ndiffraction on crystalline materials. We demonstrate that weak signals stemming\nfrom charge ordering, insignificant in the noisy data, become visible and\naccurate in the denoised data. This success is enabled by supervised training\nof a deep neural network with pairs of measured low- and high-noise data. We\ndemonstrate that using artificial noise does not yield such quantitatively\naccurate results. Our approach thus illustrates a practical strategy for noise\nfiltering that can be applied to challenging acquisition problems.\n", "date": "2022-09-19", "categories": []}
{"id": "2303.10611", "title": "Rethinking Dual-Domain Undersampled MRI reconstruction: domain-specific\n  design from the perspective of the receptive field", "abstract": "  Undersampled MRI reconstruction is crucial for accelerating clinical\nscanning. Dual-domain reconstruction network is performant among SoTA deep\nlearning methods. In this paper, we rethink dual-domain model design from the\nperspective of the receptive field, which is needed for image recovery and\nK-space interpolation problems. Further, we introduce domain-specific modules\nfor dual-domain reconstruction, namely k-space global initialization and\nimage-domain parallel local detail enhancement. We evaluate our modules by\ntranslating a SoTA method DuDoRNet under different conventions of MRI\nreconstruction including image-domain, dual-domain, and reference-guided\nreconstruction on the public IXI dataset. Our model DuDoRNet+ achieves\nsignificant improvements over competing deep learning methods.\n", "date": "2023-03-19", "categories": []}
{"id": "2402.09561", "title": "Patch-based adaptive temporal filter and residual evaluation", "abstract": "  In coherent imaging systems, speckle is a signal-dependent noise that\nvisually strongly degrades images' appearance. A huge amount of SAR data has\nbeen acquired from different sensors with different wavelengths, resolutions,\nincidences and polarizations. We extend the nonlocal filtering strategy to the\ntemporal domain and propose a patch-based adaptive temporal filter (PATF) to\ntake advantage of well-registered multi-temporal SAR images. A patch-based\ngeneralised likelihood ratio test is processed to suppress the changed object\neffects on the multitemporal denoising results. Then, the similarities are\ntransformed into corresponding weights with an exponential function. The\ndenoised value is calculated with a temporal weighted average. Spatial adaptive\ndenoising methods can improve the patch-based weighted temporal average image\nwhen the time series is limited. The spatial adaptive denoising step is\noptional when the time series is large enough. Without reference image, we\npropose using a patch-based auto-covariance residual evaluation method to\nexamine the ratio image between the noisy and denoised images and look for\npossible remaining structural contents. It can process automatically and does\nnot rely on a supervised selection of homogeneous regions. It also provides a\nglobal score for the whole image. Numerous results demonstrate the\neffectiveness of the proposed time series denoising method and the usefulness\nof the residual evaluation method.\n", "date": "2024-02-14", "categories": []}
{"id": "2311.17515", "title": "Fusion of Single and Integral Multispectral Aerial Images", "abstract": "  An adequate fusion of the most significant salient information from multiple\ninput channels is essential for many aerial imaging tasks. While multispectral\nrecordings reveal features in various spectral ranges, synthetic aperture\nsensing makes occluded features visible. We present a first and hybrid (model-\nand learning-based) architecture for fusing the most significant features from\nconventional aerial images with the ones from integral aerial images that are\nthe result of synthetic aperture sensing for removing occlusion. It combines\nthe environment's spatial references with features of unoccluded targets that\nwould normally be hidden by dense vegetation. Our method outperforms\nstate-of-the-art two-channel and multi-channel fusion approaches visually and\nquantitatively in common metrics, such as mutual information, visual\ninformation fidelity, and peak signal-to-noise ratio. The proposed model does\nnot require manually tuned parameters, can be extended to an arbitrary number\nand arbitrary combinations of spectral channels, and is reconfigurable for\naddressing different use cases. We demonstrate examples for search and rescue,\nwildfire detection, and wildlife observation.\n", "date": "2023-11-29", "categories": []}
{"id": "2402.07900", "title": "Wavefront Randomization Improves Deconvolution", "abstract": "  The performance of an imaging system is limited by optical aberrations, which\ncause blurriness in the resulting image. Digital correction techniques, such as\ndeconvolution, have limited ability to correct the blur, since some spatial\nfrequencies in the scene are not measured adequately (i.e., 'zeros' of the\nsystem transfer function). We prove that the addition of a random mask to an\nimaging system removes its dependence on aberrations, reducing the likelihood\nof zeros in the transfer function and consequently decreasing the sensitivity\nto noise during deconvolution. In simulation, we show that this strategy\nimproves image quality over a range of aberration types, aberration strengths,\nand signal-to-noise ratios.\n", "date": "2024-02-12", "categories": []}
{"id": "2402.08235", "title": "Color Image Denoising Using The Green Channel Prior", "abstract": "  Noise removal in the standard RGB (sRGB) space remains a challenging task, in\nthat the noise statistics of real-world images can be different in R, G and B\nchannels. In fact, the green channel usually has twice the sampling rate in raw\ndata and a higher signal-to-noise ratio than red/blue ones. However, the green\nchannel prior (GCP) is often understated or ignored in color image denoising\nsince many existing approaches mainly focus on modeling the relationship among\nimage patches. In this paper, we propose a simple and effective one step\nGCP-based image denoising (GCP-ID) method, which aims to exploit the GCP for\ndenoising in the sRGB space by integrating it into the classic nonlocal\ntransform domain denoising framework. Briefly, we first take advantage of the\ngreen channel to guide the search of similar patches, which improves the patch\nsearch quality and encourages sparsity in the transform domain. Then we\nreformulate RGB patches into RGGB arrays to explicitly characterize the density\nof green samples. The block circulant representation is utilized to capture the\ncross-channel correlation and the channel redundancy. Experiments on both\nsynthetic and real-world datasets demonstrate the competitive performance of\nthe proposed GCP-ID method for the color image and video denoising tasks. The\ncode is available at github.com/ZhaomingKong/GCP-ID.\n", "date": "2024-02-13", "categories": []}
{"id": "2309.13385", "title": "Cine cardiac MRI reconstruction using a convolutional recurrent network\n  with refinement", "abstract": "  Cine Magnetic Resonance Imaging (MRI) allows for understanding of the heart's\nfunction and condition in a non-invasive manner. Undersampling of the $k$-space\nis employed to reduce the scan duration, thus increasing patient comfort and\nreducing the risk of motion artefacts, at the cost of reduced image quality. In\nthis challenge paper, we investigate the use of a convolutional recurrent\nneural network (CRNN) architecture to exploit temporal correlations in\nsupervised cine cardiac MRI reconstruction. This is combined with a\nsingle-image super-resolution refinement module to improve single coil\nreconstruction by 4.4\\% in structural similarity and 3.9\\% in normalised mean\nsquare error compared to a plain CRNN implementation. We deploy a high-pass\nfilter to our $\\ell_1$ loss to allow greater emphasis on high-frequency details\nwhich are missing in the original data. The proposed model demonstrates\nconsiderable enhancements compared to the baseline case and holds promising\npotential for further improving cardiac MRI reconstruction.\n", "date": "2023-09-23", "categories": []}
{"id": "2402.07298", "title": "Supervised Reconstruction for Silhouette Tomography", "abstract": "  In this paper, we introduce silhouette tomography, a novel formulation of\nX-ray computed tomography that relies only on the geometry of the imaging\nsystem. We formulate silhouette tomography mathematically and provide a simple\nmethod for obtaining a particular solution to the problem, assuming that any\nsolution exists. We then propose a supervised reconstruction approach that uses\na deep neural network to solve the silhouette tomography problem. We present\nexperimental results on a synthetic dataset that demonstrate the effectiveness\nof the proposed method.\n", "date": "2024-02-11", "categories": []}
{"id": "2402.06191", "title": "The Berkeley Single Cell Computational Microscopy (BSCCM) Dataset", "abstract": "  Computational microscopy, in which hardware and algorithms of an imaging\nsystem are jointly designed, shows promise for making imaging systems that cost\nless, perform more robustly, and collect new types of information. Often, the\nperformance of computational imaging systems, especially those that incorporate\nmachine learning, is sample-dependent. Thus, standardized datasets are an\nessential tool for comparing the performance of different approaches. Here, we\nintroduce the Berkeley Single Cell Computational Microscopy (BSCCM) dataset,\nwhich contains over ~12,000,000 images of 400,000 of individual white blood\ncells. The dataset contains images captured with multiple illumination patterns\non an LED array microscope and fluorescent measurements of the abundance of\nsurface proteins that mark different cell types. We hope this dataset will\nprovide a valuable resource for the development and testing of new algorithms\nin computational microscopy and computer vision with practical biomedical\napplications.\n", "date": "2024-02-09", "categories": []}
{"id": "2402.04097", "title": "Analysis of Deep Image Prior and Exploiting Self-Guidance for Image\n  Reconstruction", "abstract": "  The ability of deep image prior (DIP) to recover high-quality images from\nincomplete or corrupted measurements has made it popular in inverse problems in\nimage restoration and medical imaging including magnetic resonance imaging\n(MRI). However, conventional DIP suffers from severe overfitting and spectral\nbias effects. In this work, we first provide an analysis of how DIP recovers\ninformation from undersampled imaging measurements by analyzing the training\ndynamics of the underlying networks in the kernel regime for different\narchitectures. This study sheds light on important underlying properties for\nDIP-based recovery. Current research suggests that incorporating a reference\nimage as network input can enhance DIP's performance in image reconstruction\ncompared to using random inputs. However, obtaining suitable reference images\nrequires supervision, and raises practical difficulties. In an attempt to\novercome this obstacle, we further introduce a self-driven reconstruction\nprocess that concurrently optimizes both the network weights and the input\nwhile eliminating the need for training data. Our method incorporates a novel\ndenoiser regularization term which enables robust and stable joint estimation\nof both the network input and reconstructed image. We demonstrate that our\nself-guided method surpasses both the original DIP and modern supervised\nmethods in terms of MR image reconstruction performance and outperforms\nprevious DIP-based schemes for image inpainting.\n", "date": "2024-02-06", "categories": []}
{"id": "2304.04027", "title": "NeBLa: Neural Beer-Lambert for 3D Reconstruction of Oral Structures from\n  Panoramic Radiographs", "abstract": "  Panoramic radiography (Panoramic X-ray, PX) is a widely used imaging modality\nfor dental examination. However, PX only provides a flattened 2D image, lacking\nin a 3D view of the oral structure. In this paper, we propose NeBLa (Neural\nBeer-Lambert) to estimate 3D oral structures from real-world PX. NeBLa tackles\nfull 3D reconstruction for varying subjects (patients) where each\nreconstruction is based only on a single panoramic image. We create an\nintermediate representation called simulated PX (SimPX) from 3D Cone-beam\ncomputed tomography (CBCT) data based on the Beer-Lambert law of X-ray\nrendering and rotational principles of PX imaging. SimPX aims at not only\ntruthfully simulating PX, but also facilitates the reverting process back to 3D\ndata. We propose a novel neural model based on ray tracing which exploits both\nglobal and local input features to convert SimPX to 3D output. At inference, a\nreal PX image is translated to a SimPX-style image with semantic\nregularization, and the translated image is processed by generation module to\nproduce high-quality outputs. Experiments show that NeBLa outperforms prior\nstate-of-the-art in reconstruction tasks both quantitatively and qualitatively.\nUnlike prior methods, NeBLa does not require any prior information such as the\nshape of dental arches, nor the matched PX-CBCT dataset for training, which is\ndifficult to obtain in clinical practice. Our code is available at\nhttps://github.com/sihwa-park/nebla.\n", "date": "2023-04-08", "categories": []}
{"id": "2402.02411", "title": "Physics-Inspired Degradation Models for Hyperspectral Image Fusion", "abstract": "  The fusion of a low-spatial-resolution hyperspectral image (LR-HSI) with a\nhigh-spatial-resolution multispectral image (HR-MSI) has garnered increasing\nresearch interest. However, most fusion methods solely focus on the fusion\nalgorithm itself and overlook the degradation models, which results in\nunsatisfactory performance in practical scenarios. To fill this gap, we propose\nphysics-inspired degradation models (PIDM) to model the degradation of LR-HSI\nand HR-MSI, which comprises a spatial degradation network (SpaDN) and a\nspectral degradation network (SpeDN). SpaDN and SpeDN are designed based on two\ninsights. First, we employ spatial warping and spectral modulation operations\nto simulate lens aberrations, thereby introducing non-uniformity into the\nspatial and spectral degradation processes. Second, we utilize asymmetric\ndownsampling and parallel downsampling operations to separately reduce the\nspatial and spectral resolutions of the images, thus ensuring the matching of\nspatial and spectral degradation processes with specific physical\ncharacteristics. Once SpaDN and SpeDN are established, we adopt a\nself-supervised training strategy to optimize the network parameters and\nprovide a plug-and-play solution for fusion methods. Comprehensive experiments\ndemonstrate that our proposed PIDM can boost the fusion performance of existing\nfusion methods in practical scenarios.\n", "date": "2024-02-04", "categories": []}
{"id": "2402.03398", "title": "Deep Nonlinear Hyperspectral Unmixing Using Multi-task Learning", "abstract": "  Nonlinear hyperspectral unmixing has recently received considerable\nattention, as linear mixture models do not lead to an acceptable resolution in\nsome problems. In fact, most nonlinear unmixing methods are designed by\nassuming specific assumptions on the nonlinearity model which subsequently\nlimits the unmixing performance. In this paper, we propose an unsupervised\nnonlinear unmixing approach based on deep learning by incorporating a general\nnonlinear model with no special assumptions. This model consists of two\nbranches. In the first branch, endmembers are learned by reconstructing the\nrows of hyperspectral images using some hidden layers, and in the second\nbranch, abundance values are learned based on the columns of respective images.\nThen, using multi-task learning, we introduce an auxiliary task to enforce the\ntwo branches to work together. This technique can be considered as a\nregularizer mitigating overfitting, which improves the performance of the total\nnetwork. Extensive experiments on synthetic and real data verify the\neffectiveness of the proposed method compared to some state-of-the-art\nhyperspectral unmixing methods.\n", "date": "2024-02-05", "categories": []}
{"id": "2402.03761", "title": "Deep Learning-Based Correction and Unmixing of Hyperspectral Images for\n  Brain Tumor Surgery", "abstract": "  Hyperspectral Imaging (HSI) for fluorescence-guided brain tumor resection\nenables visualization of differences between tissues that are not\ndistinguishable to humans. This augmentation can maximize brain tumor\nresection, improving patient outcomes. However, much of the processing in HSI\nuses simplified linear methods that are unable to capture the non-linear,\nwavelength-dependent phenomena that must be modeled for accurate recovery of\nfluorophore abundances. We therefore propose two deep learning models for\ncorrection and unmixing, which can account for the nonlinear effects and\nproduce more accurate estimates of abundances. Both models use an\nautoencoder-like architecture to process the captured spectra. One is trained\nwith protoporphyrin IX (PpIX) concentration labels. The other undergoes\nsemi-supervised training, first learning hyperspectral unmixing self-supervised\nand then learning to correct fluorescence emission spectra for heterogeneous\noptical and geometric properties using a reference white-light reflectance\nspectrum in a few-shot manner. The models were evaluated against phantom and\npig brain data with known PpIX concentration; the supervised model achieved\nPearson correlation coefficients (R values) between the known and computed PpIX\nconcentrations of 0.997 and 0.990, respectively, whereas the classical approach\nachieved only 0.93 and 0.82. The semi-supervised approach's R values were 0.98\nand 0.91, respectively. On human data, the semi-supervised model gives\nqualitatively more realistic results than the classical method, better removing\nbright spots of specular reflectance and reducing the variance in PpIX\nabundance over biopsies that should be relatively homogeneous. These results\nshow promise for using deep learning to improve HSI in fluorescence-guided\nneurosurgery.\n", "date": "2024-02-06", "categories": []}
{"id": "2309.07921", "title": "OpenIllumination: A Multi-Illumination Dataset for Inverse Rendering\n  Evaluation on Real Objects", "abstract": "  We introduce OpenIllumination, a real-world dataset containing over 108K\nimages of 64 objects with diverse materials, captured under 72 camera views and\na large number of different illuminations. For each image in the dataset, we\nprovide accurate camera parameters, illumination ground truth, and foreground\nsegmentation masks. Our dataset enables the quantitative evaluation of most\ninverse rendering and material decomposition methods for real objects. We\nexamine several state-of-the-art inverse rendering methods on our dataset and\ncompare their performances. The dataset and code can be found on the project\npage: https://oppo-us-research.github.io/OpenIllumination.\n", "date": "2023-09-14", "categories": []}
{"id": "2310.11050", "title": "$k$-$t$ CLAIR: Self-Consistency Guided Multi-Prior Learning for Dynamic\n  Parallel MR Image Reconstruction", "abstract": "  Cardiac magnetic resonance imaging (CMR) has been widely used in clinical\npractice for the medical diagnosis of cardiac diseases. However, the long\nacquisition time hinders its development in real-time applications. Here, we\npropose a novel self-consistency guided multi-prior learning framework named\n$k$-$t$ CLAIR to exploit spatiotemporal correlations from highly undersampled\ndata for accelerated dynamic parallel MRI reconstruction. The $k$-$t$ CLAIR\nprogressively reconstructs faithful images by leveraging multiple complementary\npriors learned in the $x$-$t$, $x$-$f$, and $k$-$t$ domains in an iterative\nfashion, as dynamic MRI exhibits high spatiotemporal redundancy. Additionally,\n$k$-$t$ CLAIR incorporates calibration information for prior learning,\nresulting in a more consistent reconstruction. Experimental results on cardiac\ncine and T1W/T2W images demonstrate that $k$-$t$ CLAIR achieves high-quality\ndynamic MR reconstruction in terms of both quantitative and qualitative\nperformance.\n", "date": "2023-10-17", "categories": []}
{"id": "2304.14604", "title": "Deep Neural-network Prior for Orbit Recovery from Method of Moments", "abstract": "  Orbit recovery problems are a class of problems that often arise in practice\nand various forms. In these problems, we aim to estimate an unknown function\nafter being distorted by a group action and observed via a known operator.\nTypically, the observations are contaminated with a non-trivial level of noise.\nTwo particular orbit recovery problems of interest in this paper are\nmultireference alignment and single-particle cryo-EM modelling. In order to\nsuppress the noise, we suggest using the method of moments approach for both\nproblems while introducing deep neural network priors. In particular, our\nneural networks should output the signals and the distribution of group\nelements, with moments being the input. In the multireference alignment case,\nwe demonstrate the advantage of using the NN to accelerate the convergence for\nthe reconstruction of signals from the moments. Finally, we use our method to\nreconstruct simulated and biological volumes in the cryo-EM setting.\n", "date": "2023-04-28", "categories": []}
{"id": "2303.12861", "title": "Parallel Diffusion Model-based Sparse-view Cone-beam Breast CT", "abstract": "  Breast cancer is the most prevalent cancer among women worldwide, and early\ndetection is crucial for reducing its mortality rate and improving quality of\nlife. Dedicated breast computed tomography (CT) scanners offer better image\nquality than mammography and tomosynthesis in general but at higher radiation\ndose. To enable breast CT for cancer screening, the challenge is to minimize\nthe radiation dose without compromising image quality, according to the ALARA\nprinciple (as low as reasonably achievable). Over the past years, deep learning\nhas shown remarkable successes in various tasks, including low-dose CT\nespecially few-view CT. Currently, the diffusion model presents the state of\nthe art for CT reconstruction. To develop the first diffusion model-based\nbreast CT reconstruction method, here we report innovations to address the\nlarge memory requirement for breast cone-beam CT reconstruction and high\ncomputational cost of the diffusion model. Specifically, in this study we\ntransform the cutting-edge Denoising Diffusion Probabilistic Model (DDPM) into\na parallel framework for sub-volume-based sparse-view breast CT image\nreconstruction in projection and image domains. This novel approach involves\nthe concurrent training of two distinct DDPM models dedicated to processing\nprojection and image data synergistically in the dual domains. Our experimental\nfindings reveal that this method delivers competitive reconstruction\nperformance at half to one-third of the standard radiation doses. This\nadvancement demonstrates an exciting potential of diffusion-type models for\nvolumetric breast reconstruction at high-resolution with much-reduced radiation\ndose and as such hopefully redefines breast cancer screening and diagnosis.\n", "date": "2023-03-22", "categories": []}
{"id": "2401.14762", "title": "A Comparative Study of Compressive Sensing Algorithms for Hyperspectral\n  Imaging Reconstruction", "abstract": "  Hyperspectral Imaging comprises excessive data consequently leading to\nsignificant challenges for data processing, storage and transmission.\nCompressive Sensing has been used in the field of Hyperspectral Imaging as a\ntechnique to compress the large amount of data. This work addresses the\nrecovery of hyperspectral images 2.5x compressed. A comparative study in terms\nof the accuracy and the performance of the convex FISTA/ADMM in addition to the\ngreedy gOMP/BIHT/CoSaMP recovery algorithms is presented. The results indicate\nthat the algorithms recover successfully the compressed data, yet the gOMP\nalgorithm achieves superior accuracy and faster recovery in comparison to the\nother algorithms at the expense of high dependence on unknown sparsity level of\nthe data to recover.\n", "date": "2024-01-26", "categories": []}
{"id": "2309.06023", "title": "Learning from History: Task-agnostic Model Contrastive Learning for\n  Image Restoration", "abstract": "  Contrastive learning has emerged as a prevailing paradigm for high-level\nvision tasks, which, by introducing properly negative samples, has also been\nexploited for low-level vision tasks to achieve a compact optimization space to\naccount for their ill-posed nature. However, existing methods rely on manually\npredefined and task-oriented negatives, which often exhibit pronounced\ntask-specific biases. To address this challenge, our paper introduces an\ninnovative method termed 'learning from history', which dynamically generates\nnegative samples from the target model itself. Our approach, named Model\nContrastive Learning for Image Restoration (MCLIR), rejuvenates latency models\nas negative models, making it compatible with diverse image restoration tasks.\nWe propose the Self-Prior guided Negative loss (SPN) to enable it. This\napproach significantly enhances existing models when retrained with the\nproposed model contrastive paradigm. The results show significant improvements\nin image restoration across various tasks and architectures. For example,\nmodels retrained with SPN outperform the original FFANet and DehazeFormer by\n3.41 dB and 0.57 dB on the RESIDE indoor dataset for image dehazing. Similarly,\nthey achieve notable improvements of 0.47 dB on SPA-Data over IDT for image\nderaining and 0.12 dB on Manga109 for a 4x scale super-resolution over\nlightweight SwinIR, respectively. Code and retrained models are available at\nhttps://github.com/Aitical/MCLIR.\n", "date": "2023-09-12", "categories": []}
{"id": "2401.13221", "title": "Unified-Width Adaptive Dynamic Network for All-In-One Image Restoration", "abstract": "  In contrast to traditional image restoration methods, all-in-one image\nrestoration techniques are gaining increased attention for their ability to\nrestore images affected by diverse and unknown corruption types and levels.\nHowever, contemporary all-in-one image restoration methods omit task-wise\ndifficulties and employ the same networks to reconstruct images afflicted by\ndiverse degradations. This practice leads to an underestimation of the task\ncorrelations and suboptimal allocation of computational resources. To elucidate\ntask-wise complexities, we introduce a novel concept positing that intricate\nimage degradation can be represented in terms of elementary degradation.\nBuilding upon this foundation, we propose an innovative approach, termed the\nUnified-Width Adaptive Dynamic Network (U-WADN), consisting of two pivotal\ncomponents: a Width Adaptive Backbone (WAB) and a Width Selector (WS). The WAB\nincorporates several nested sub-networks with varying widths, which facilitates\nthe selection of the most apt computations tailored to each task, thereby\nstriking a balance between accuracy and computational efficiency during\nruntime. For different inputs, the WS automatically selects the most\nappropriate sub-network width, taking into account both task-specific and\nsample-specific complexities. Extensive experiments across a variety of image\nrestoration tasks demonstrate that the proposed U-WADN achieves better\nperformance while simultaneously reducing up to 32.3\\% of FLOPs and providing\napproximately 15.7\\% real-time acceleration. The code has been made available\nat \\url{https://github.com/xuyimin0926/U-WADN}.\n", "date": "2024-01-24", "categories": []}
{"id": "2208.05481", "title": "High-Frequency Space Diffusion Models for Accelerated MRI", "abstract": "  Diffusion models with continuous stochastic differential equations (SDEs)\nhave shown superior performances in image generation. It can serve as a deep\ngenerative prior to solving the inverse problem in magnetic resonance (MR)\nreconstruction. However, low-frequency regions of $k$-space data are typically\nfully sampled in fast MR imaging, while existing diffusion models are performed\nthroughout the entire image or $k$-space, inevitably introducing uncertainty in\nthe reconstruction of low-frequency regions. Additionally, existing diffusion\nmodels often demand substantial iterations to converge, resulting in\ntime-consuming reconstructions. To address these challenges, we propose a novel\nSDE tailored specifically for MR reconstruction with the diffusion process in\nhigh-frequency space (referred to as HFS-SDE). This approach ensures\ndeterminism in the fully sampled low-frequency regions and accelerates the\nsampling procedure of reverse diffusion. Experiments conducted on the publicly\navailable fastMRI dataset demonstrate that the proposed HFS-SDE method\noutperforms traditional parallel imaging methods, supervised deep learning, and\nexisting diffusion models in terms of reconstruction accuracy and stability.\nThe fast convergence properties are also confirmed through theoretical and\nexperimental validation. Our code and weights are available at\nhttps://github.com/Aboriginer/HFS-SDE.\n", "date": "2022-08-10", "categories": []}
{"id": "2305.10124", "title": "Principal Uncertainty Quantification with Spatial Correlation for Image\n  Restoration Problems", "abstract": "  Uncertainty quantification for inverse problems in imaging has drawn much\nattention lately. Existing approaches towards this task define uncertainty\nregions based on probable values per pixel, while ignoring spatial correlations\nwithin the image, resulting in an exaggerated volume of uncertainty. In this\npaper, we propose PUQ (Principal Uncertainty Quantification) -- a novel\ndefinition and corresponding analysis of uncertainty regions that takes into\naccount spatial relationships within the image, thus providing reduced volume\nregions. Using recent advancements in generative models, we derive uncertainty\nintervals around principal components of the empirical posterior distribution,\nforming an ambiguity region that guarantees the inclusion of true unseen values\nwith a user-defined confidence probability. To improve computational efficiency\nand interpretability, we also guarantee the recovery of true unseen values\nusing only a few principal directions, resulting in more informative\nuncertainty regions. Our approach is verified through experiments on image\ncolorization, super-resolution, and inpainting; its effectiveness is shown\nthrough comparison to baseline methods, demonstrating significantly tighter\nuncertainty regions.\n", "date": "2023-05-17", "categories": []}
{"id": "2308.01738", "title": "Enhancing Visibility in Nighttime Haze Images Using Guided APSF and\n  Gradient Adaptive Convolution", "abstract": "  Visibility in hazy nighttime scenes is frequently reduced by multiple\nfactors, including low light, intense glow, light scattering, and the presence\nof multicolored light sources. Existing nighttime dehazing methods often\nstruggle with handling glow or low-light conditions, resulting in either\nexcessively dark visuals or unsuppressed glow outputs. In this paper, we\nenhance the visibility from a single nighttime haze image by suppressing glow\nand enhancing low-light regions. To handle glow effects, our framework learns\nfrom the rendered glow pairs. Specifically, a light source aware network is\nproposed to detect light sources of night images, followed by the APSF\n(Atmospheric Point Spread Function)-guided glow rendering. Our framework is\nthen trained on the rendered images, resulting in glow suppression. Moreover,\nwe utilize gradient-adaptive convolution, to capture edges and textures in hazy\nscenes. By leveraging extracted edges and textures, we enhance the contrast of\nthe scene without losing important structural details. To boost low-light\nintensity, our network learns an attention map, then adjusted by gamma\ncorrection. This attention has high values on low-light regions and low values\non haze and glow regions. Extensive evaluation on real nighttime haze images,\ndemonstrates the effectiveness of our method. Our experiments demonstrate that\nour method achieves a PSNR of 30.38dB, outperforming state-of-the-art methods\nby 13% on GTA5 nighttime haze dataset. Our data and code is available at\nhttps://github.com/jinyeying/nighttime_dehaze.\n", "date": "2023-08-03", "categories": []}
{"id": "2401.11256", "title": "Equivariant Multiscale Learned Invertible Reconstruction for Cone Beam\n  CT", "abstract": "  Cone Beam CT (CBCT) is an essential imaging modality nowadays, but the image\nquality of CBCT still lags behind the high quality standards established by the\nconventional Computed Tomography. We propose LIRE+, a learned iterative scheme\nfor fast and memory-efficient CBCT reconstruction, which is a substantially\nfaster and more parameter-efficient alternative to the recently proposed LIRE\nmethod. LIRE+ is a rotationally-equivariant multiscale learned invertible\nprimal-dual iterative scheme for CBCT reconstruction. Memory usage is optimized\nby relying on simple reversible residual networks in primal/dual cells and\npatch-wise computations inside the cells during forward and backward passes,\nwhile increased inference speed is achieved by making the primal-dual scheme\nmultiscale so that the reconstruction process starts at low resolution and with\nlow resolution primal/dual latent vectors. A LIRE+ model was trained and\nvalidated on a set of 260 + 22 thorax CT scans and tested using a set of 142\nthorax CT scans with additional evaluation with and without finetuning on an\nout-of-distribution set of 79 Head and Neck (HN) CT scans. Our method surpasses\nclassical and deep learning baselines, including LIRE, on the thorax test set.\nFor a similar inference time and with only 37 % of the parameter budget, LIRE+\nachieves a +0.2 dB PSNR improvement over LIRE, while being able to match the\nperformance of LIRE in 45 % less inference time and with 28 % of the parameter\nbudget. Rotational equivariance ensures robustness of LIRE+ to patient\norientation, while LIRE and other deep learning baselines suffer from\nsubstantial performance degradation when patient orientation is unusual. On the\nHN dataset in the absence of finetuning, LIRE+ is generally comparable to LIRE\nin performance apart from a few outlier cases, whereas after identical\nfinetuning LIRE+ demonstates a +1.02 dB PSNR improvement over LIRE.\n", "date": "2024-01-20", "categories": []}
{"id": "2401.11541", "title": "Multi-View Neural 3D Reconstruction of Micro-/Nanostructures with Atomic\n  Force Microscopy", "abstract": "  Atomic Force Microscopy (AFM) is a widely employed tool for micro-/nanoscale\ntopographic imaging. However, conventional AFM scanning struggles to\nreconstruct complex 3D micro-/nanostructures precisely due to limitations such\nas incomplete sample topography capturing and tip-sample convolution artifacts.\nHere, we propose a multi-view neural-network-based framework with AFM\n(MVN-AFM), which accurately reconstructs surface models of intricate\nmicro-/nanostructures. Unlike previous works, MVN-AFM does not depend on any\nspecially shaped probes or costly modifications to the AFM system. To achieve\nthis, MVN-AFM uniquely employs an iterative method to align multi-view data and\neliminate AFM artifacts simultaneously. Furthermore, we pioneer the application\nof neural implicit surface reconstruction in nanotechnology and achieve\nmarkedly improved results. Extensive experiments show that MVN-AFM effectively\neliminates artifacts present in raw AFM images and reconstructs various\nmicro-/nanostructures including complex geometrical microstructures printed via\nTwo-photon Lithography and nanoparticles such as PMMA nanospheres and ZIF-67\nnanocrystals. This work presents a cost-effective tool for micro-/nanoscale 3D\nanalysis.\n", "date": "2024-01-21", "categories": []}
{"id": "2312.06946", "title": "WaterHE-NeRF: Water-ray Tracing Neural Radiance Fields for Underwater\n  Scene Reconstruction", "abstract": "  Neural Radiance Field (NeRF) technology demonstrates immense potential in\nnovel viewpoint synthesis tasks, due to its physics-based volumetric rendering\nprocess, which is particularly promising in underwater scenes. Addressing the\nlimitations of existing underwater NeRF methods in handling light attenuation\ncaused by the water medium and the lack of real Ground Truth (GT) supervision,\nthis study proposes WaterHE-NeRF. We develop a new water-ray tracing field by\nRetinex theory that precisely encodes color, density, and illuminance\nattenuation in three-dimensional space. WaterHE-NeRF, through its illuminance\nattenuation mechanism, generates both degraded and clear multi-view images and\noptimizes image restoration by combining reconstruction loss with Wasserstein\ndistance. Additionally, the use of histogram equalization (HE) as pseudo-GT\nenhances the network's accuracy in preserving original details and color\ndistribution. Extensive experiments on real underwater datasets and synthetic\ndatasets validate the effectiveness of WaterHE-NeRF. Our code will be made\npublicly available.\n", "date": "2023-12-12", "categories": []}
{"id": "2308.11932", "title": "Synergistic Multiscale Detail Refinement via Intrinsic Supervision for\n  Underwater Image Enhancement", "abstract": "  Visually restoring underwater scenes primarily involves mitigating\ninterference from underwater media. Existing methods ignore the inherent\nscale-related characteristics in underwater scenes. Therefore, we present the\nsynergistic multi-scale detail refinement via intrinsic supervision (SMDR-IS)\nfor enhancing underwater scene details, which contain multi-stages. The\nlow-degradation stage from the original images furnishes the original stage\nwith multi-scale details, achieved through feature propagation using the\nAdaptive Selective Intrinsic Supervised Feature (ASISF) module. By using\nintrinsic supervision, the ASISF module can precisely control and guide feature\ntransmission across multi-degradation stages, enhancing multi-scale detail\nrefinement and minimizing the interference from irrelevant information in the\nlow-degradation stage. In multi-degradation encoder-decoder framework of\nSMDR-IS, we introduce the Bifocal Intrinsic-Context Attention Module (BICA).\nBased on the intrinsic supervision principles, BICA efficiently exploits\nmulti-scale scene information in images. BICA directs higher-resolution spaces\nby tapping into the insights of lower-resolution ones, underscoring the pivotal\nrole of spatial contextual relationships in underwater image restoration.\nThroughout training, the inclusion of a multi-degradation loss function can\nenhance the network, allowing it to adeptly extract information across diverse\nscales. When benchmarked against state-of-the-art methods, SMDR-IS consistently\nshowcases superior performance. The code is publicly available at:\nhttps://github.com/zhoujingchun03/SMDR-IS.\n", "date": "2023-08-23", "categories": []}
{"id": "2312.01689", "title": "Fast and accurate sparse-view CBCT reconstruction using meta-learned\n  neural attenuation field and hash-encoding regularization", "abstract": "  Cone beam computed tomography (CBCT) is an emerging medical imaging technique\nto visualize the internal anatomical structures of patients. During a CBCT\nscan, several projection images of different angles or views are collectively\nutilized to reconstruct a tomographic image. However, reducing the number of\nprojections in a CBCT scan while preserving the quality of a reconstructed\nimage is challenging due to the nature of an ill-posed inverse problem.\nRecently, a neural attenuation field (NAF) method was proposed by adopting a\nneural radiance field algorithm as a new way for CBCT reconstruction,\ndemonstrating fast and promising results using only 50 views. However,\ndecreasing the number of projections is still preferable to reduce potential\nradiation exposure, and a faster reconstruction time is required considering a\ntypical scan time. In this work, we propose a fast and accurate sparse-view\nCBCT reconstruction (FACT) method to provide better reconstruction quality and\nfaster optimization speed in the minimal number of view acquisitions ($<$ 50\nviews). In the FACT method, we meta-trained a neural network and a hash-encoder\nusing a few scans (= 15), and a new regularization technique is utilized to\nreconstruct the details of an anatomical structure. In conclusion, we have\nshown that the FACT method produced better, and faster reconstruction results\nover the other conventional algorithms based on CBCT scans of different body\nparts (chest, head, and abdomen) and CT vendors (Siemens, Phillips, and GE).\n", "date": "2023-12-04", "categories": []}
{"id": "2301.10260", "title": "Learned Interferometric Imaging for the SPIDER Instrument", "abstract": "  The Segmented Planar Imaging Detector for Electro-Optical Reconnaissance\n(SPIDER) is an optical interferometric imaging device that aims to offer an\nalternative to the large space telescope designs of today with reduced size,\nweight and power consumption. This is achieved through interferometric imaging.\nState-of-the-art methods for reconstructing images from interferometric\nmeasurements adopt proximal optimization techniques, which are computationally\nexpensive and require handcrafted priors. In this work we present two\ndata-driven approaches for reconstructing images from measurements made by the\nSPIDER instrument. These approaches use deep learning to learn prior\ninformation from training data, increasing the reconstruction quality, and\nsignificantly reducing the computation time required to recover images by\norders of magnitude. Reconstruction time is reduced to ${\\sim} 10$\nmilliseconds, opening up the possibility of real-time imaging with SPIDER for\nthe first time. Furthermore, we show that these methods can also be applied in\ndomains where training data is scarce, such as astronomical imaging, by\nleveraging transfer learning from domains where plenty of training data are\navailable.\n", "date": "2023-01-24", "categories": []}
{"id": "2308.09460", "title": "Accelerated Bayesian imaging by relaxed proximal-point Langevin sampling", "abstract": "  This paper presents a new accelerated proximal Markov chain Monte Carlo\nmethodology to perform Bayesian inference in imaging inverse problems with an\nunderlying convex geometry. The proposed strategy takes the form of a\nstochastic relaxed proximal-point iteration that admits two complementary\ninterpretations. For models that are smooth or regularised by Moreau-Yosida\nsmoothing, the algorithm is equivalent to an implicit midpoint discretisation\nof an overdamped Langevin diffusion targeting the posterior distribution of\ninterest. This discretisation is asymptotically unbiased for Gaussian targets\nand shown to converge in an accelerated manner for any target that is\n$\\kappa$-strongly log-concave (i.e., requiring in the order of $\\sqrt{\\kappa}$\niterations to converge, similarly to accelerated optimisation schemes),\ncomparing favorably to [M. Pereyra, L. Vargas Mieles, K.C. Zygalakis, SIAM J.\nImaging Sciences, 13,2 (2020), pp. 905-935] which is only provably accelerated\nfor Gaussian targets and has bias. For models that are not smooth, the\nalgorithm is equivalent to a Leimkuhler-Matthews discretisation of a Langevin\ndiffusion targeting a Moreau-Yosida approximation of the posterior distribution\nof interest, and hence achieves a significantly lower bias than conventional\nunadjusted Langevin strategies based on the Euler-Maruyama discretisation. For\ntargets that are $\\kappa$-strongly log-concave, the provided non-asymptotic\nconvergence analysis also identifies the optimal time step which maximizes the\nconvergence speed. The proposed methodology is demonstrated through a range of\nexperiments related to image deconvolution with Gaussian and Poisson noise,\nwith assumption-driven and data-driven convex priors. Source codes for the\nnumerical experiments of this paper are available from\nhttps://github.com/MI2G/accelerated-langevin-imla.\n", "date": "2023-08-18", "categories": []}
{"id": "2401.07213", "title": "Depth-agnostic Single Image Dehazing", "abstract": "  Single image dehazing is a challenging ill-posed problem. Existing datasets\nfor training deep learning-based methods can be generated by hand-crafted or\nsynthetic schemes. However, the former often suffers from small scales, while\nthe latter forces models to learn scene depth instead of haze distribution,\ndecreasing their dehazing ability. To overcome the problem, we propose a simple\nyet novel synthetic method to decouple the relationship between haze density\nand scene depth, by which a depth-agnostic dataset (DA-HAZE) is generated.\nMeanwhile, a Global Shuffle Strategy (GSS) is proposed for generating\ndifferently scaled datasets, thereby enhancing the generalization ability of\nthe model. Extensive experiments indicate that models trained on DA-HAZE\nachieve significant improvements on real-world benchmarks, with less\ndiscrepancy between SOTS and DA-SOTS (the test set of DA-HAZE). Additionally,\nDepth-agnostic dehazing is a more complicated task because of the lack of depth\nprior. Therefore, an efficient architecture with stronger feature modeling\nability and fewer computational costs is necessary. We revisit the U-Net-based\narchitectures for dehazing, in which dedicatedly designed blocks are\nincorporated. However, the performances of blocks are constrained by limited\nfeature fusion methods. To this end, we propose a Convolutional Skip Connection\n(CSC) module, allowing vanilla feature fusion methods to achieve promising\nresults with minimal costs. Extensive experimental results demonstrate that\ncurrent state-of-the-art methods. equipped with CSC can achieve better\nperformance and reasonable computational expense, whether the haze distribution\nis relevant to the scene depth.\n", "date": "2024-01-14", "categories": []}
{"id": "2306.08836", "title": "Probabilistic-based Feature Embedding of 4-D Light Fields for\n  Compressive Imaging and Denoising", "abstract": "  The high-dimensional nature of the 4-D light field (LF) poses great\nchallenges in achieving efficient and effective feature embedding, that\nseverely impacts the performance of downstream tasks. To tackle this crucial\nissue, in contrast to existing methods with empirically-designed architectures,\nwe propose a probabilistic-based feature embedding (PFE), which learns a\nfeature embedding architecture by assembling various low-dimensional\nconvolution patterns in a probability space for fully capturing spatial-angular\ninformation. Building upon the proposed PFE, we then leverage the intrinsic\nlinear imaging model of the coded aperture camera to construct a\ncycle-consistent 4-D LF reconstruction network from coded measurements.\nMoreover, we incorporate PFE into an iterative optimization framework for 4-D\nLF denoising. Our extensive experiments demonstrate the significant superiority\nof our methods on both real-world and synthetic 4-D LF images, both\nquantitatively and qualitatively, when compared with state-of-the-art methods.\nThe source code will be publicly available at\nhttps://github.com/lyuxianqiang/LFCA-CR-NET.\n", "date": "2023-06-15", "categories": []}
{"id": "2302.10630", "title": "LIT-Former: Linking In-plane and Through-plane Transformers for\n  Simultaneous CT Image Denoising and Deblurring", "abstract": "  This paper studies 3D low-dose computed tomography (CT) imaging. Although\nvarious deep learning methods were developed in this context, typically they\nfocus on 2D images and perform denoising due to low-dose and deblurring for\nsuper-resolution separately. Up to date, little work was done for simultaneous\nin-plane denoising and through-plane deblurring, which is important to obtain\nhigh-quality 3D CT images with lower radiation and faster imaging speed. For\nthis task, a straightforward method is to directly train an end-to-end 3D\nnetwork. However, it demands much more training data and expensive\ncomputational costs. Here, we propose to link in-plane and through-plane\ntransformers for simultaneous in-plane denoising and through-plane deblurring,\ntermed as LIT-Former, which can efficiently synergize in-plane and\nthrough-plane sub-tasks for 3D CT imaging and enjoy the advantages of both\nconvolution and transformer networks. LIT-Former has two novel designs:\nefficient multi-head self-attention modules (eMSM) and efficient convolutional\nfeedforward networks (eCFN). First, eMSM integrates in-plane 2D self-attention\nand through-plane 1D self-attention to efficiently capture global interactions\nof 3D self-attention, the core unit of transformer networks. Second, eCFN\nintegrates 2D convolution and 1D convolution to extract local information of 3D\nconvolution in the same fashion. As a result, the proposed LIT-Former synergize\nthese two subtasks, significantly reducing the computational complexity as\ncompared to 3D counterparts and enabling rapid convergence. Extensive\nexperimental results on simulated and clinical datasets demonstrate superior\nperformance over state-of-the-art models. The source code is made available at\nhttps://github.com/hao1635/LIT-Former.\n", "date": "2023-02-21", "categories": []}
{"id": "2311.14925", "title": "Coordinate-based Neural Network for Fourier Phase Retrieval", "abstract": "  Fourier phase retrieval is essential for high-definition imaging of nanoscale\nstructures across diverse fields, notably coherent diffraction imaging. This\nstudy presents the Single impliCit neurAl Network (SCAN), a tool built upon\ncoordinate neural networks meticulously designed for enhanced phase retrieval\nperformance. Remedying the drawbacks of conventional iterative methods which\nare easiliy trapped into local minimum solutions and sensitive to noise, SCAN\nadeptly connects object coordinates to their amplitude and phase within a\nunified network in an unsupervised manner. While many existing methods\nprimarily use Fourier magnitude in their loss function, our approach\nincorporates both the predicted magnitude and phase, enhancing retrieval\naccuracy. Comprehensive tests validate SCAN's superiority over traditional and\nother deep learning models regarding accuracy and noise robustness. We also\ndemonstrate that SCAN excels in the ptychography setting.\n", "date": "2023-11-25", "categories": []}
{"id": "2401.03220", "title": "MetaISP -- Exploiting Global Scene Structure for Accurate Multi-Device\n  Color Rendition", "abstract": "  Image signal processors (ISPs) are historically grown legacy software systems\nfor reconstructing color images from noisy raw sensor measurements. Each\nsmartphone manufacturer has developed its ISPs with its own characteristic\nheuristics for improving the color rendition, for example, skin tones and other\nvisually essential colors. The recent interest in replacing the historically\ngrown ISP systems with deep-learned pipelines to match DSLR's image quality\nimproves structural features in the image. However, these works ignore the\nsuperior color processing based on semantic scene analysis that distinguishes\nmobile phone ISPs from DSLRs. Here, we present MetaISP, a single model designed\nto learn how to translate between the color and local contrast characteristics\nof different devices. MetaISP takes the RAW image from device A as input and\ntranslates it to RGB images that inherit the appearance characteristics of\ndevices A, B, and C. We achieve this result by employing a lightweight deep\nlearning technique that conditions its output appearance based on the device of\ninterest. In this approach, we leverage novel attention mechanisms inspired by\ncross-covariance to learn global scene semantics. Additionally, we use the\nmetadata that typically accompanies RAW images and estimate scene illuminants\nwhen they are unavailable.\n", "date": "2024-01-06", "categories": []}
{"id": "2303.04940", "title": "Non-aligned supervision for Real Image Dehazing", "abstract": "  Removing haze from real-world images is challenging due to unpredictable\nweather conditions, resulting in the misalignment of hazy and clear image\npairs. In this paper, we propose an innovative dehazing framework that operates\nunder non-aligned supervision. This framework is grounded in the atmospheric\nscattering model, and consists of three interconnected networks: dehazing,\nairlight, and transmission networks. In particular, we explore a non-alignment\nscenario that a clear reference image, unaligned with the input hazy image, is\nutilized to supervise the dehazing network. To implement this, we present a\nmulti-scale reference loss that compares the feature representations between\nthe referred image and the dehazed output. Our scenario makes it easier to\ncollect hazy/clear image pairs in real-world environments, even under\nconditions of misalignment and shift views. To showcase the effectiveness of\nour scenario, we have collected a new hazy dataset including 415 image pairs\ncaptured by mobile Phone in both rural and urban areas, called \"Phone-Hazy\".\nFurthermore, we introduce a self-attention network based on mean and variance\nfor modeling real infinite airlight, using the dark channel prior as positional\nguidance. Additionally, a channel attention network is employed to estimate the\nthree-channel transmission. Experimental results demonstrate the superior\nperformance of our framework over existing state-of-the-art techniques in the\nreal-world image dehazing task. Phone-Hazy and code will be available at\nhttps://fanjunkai1.github.io/projectpage/NSDNet/index.html.\n", "date": "2023-03-08", "categories": []}
{"id": "2308.16742", "title": "Unsupervised CT Metal Artifact Reduction by Plugging Diffusion Priors in\n  Dual Domains", "abstract": "  During the process of computed tomography (CT), metallic implants often cause\ndisruptive artifacts in the reconstructed images, impeding accurate diagnosis.\nSeveral supervised deep learning-based approaches have been proposed for\nreducing metal artifacts (MAR). However, these methods heavily rely on training\nwith simulated data, as obtaining paired metal artifact CT and clean CT data in\nclinical settings is challenging. This limitation can lead to decreased\nperformance when applying these methods in clinical practice. Existing\nunsupervised MAR methods, whether based on learning or not, typically operate\nwithin a single domain, either in the image domain or the sinogram domain. In\nthis paper, we propose an unsupervised MAR method based on the diffusion model,\na generative model with a high capacity to represent data distributions.\nSpecifically, we first train a diffusion model using CT images without metal\nartifacts. Subsequently, we iteratively utilize the priors embedded within the\npre-trained diffusion model in both the sinogram and image domains to restore\nthe degraded portions caused by metal artifacts. This dual-domain processing\nempowers our approach to outperform existing unsupervised MAR methods,\nincluding another MAR method based on the diffusion model, which we have\nqualitatively and quantitatively validated using synthetic datasets. Moreover,\nour method demonstrates superior visual results compared to both supervised and\nunsupervised methods on clinical datasets.\n", "date": "2023-08-31", "categories": []}
{"id": "2401.02884", "title": "MsDC-DEQ-Net: Deep Equilibrium Model (DEQ) with Multi-scale Dilated\n  Convolution for Image Compressive Sensing (CS)", "abstract": "  Compressive sensing (CS) is a technique that enables the recovery of sparse\nsignals using fewer measurements than traditional sampling methods. To address\nthe computational challenges of CS reconstruction, our objective is to develop\nan interpretable and concise neural network model for reconstructing natural\nimages using CS. We achieve this by mapping one step of the iterative shrinkage\nthresholding algorithm (ISTA) to a deep network block, representing one\niteration of ISTA. To enhance learning ability and incorporate structural\ndiversity, we integrate aggregated residual transformations (ResNeXt) and\nsqueeze-and-excitation (SE) mechanisms into the ISTA block. This block serves\nas a deep equilibrium layer, connected to a semi-tensor product network\n(STP-Net) for convenient sampling and providing an initial reconstruction. The\nresulting model, called MsDC-DEQ-Net, exhibits competitive performance compared\nto state-of-the-art network-based methods. It significantly reduces storage\nrequirements compared to deep unrolling methods, using only one iteration block\ninstead of multiple iterations. Unlike deep unrolling models, MsDC-DEQ-Net can\nbe iteratively used, gradually improving reconstruction accuracy while\nconsidering computation trade-offs. Additionally, the model benefits from\nmulti-scale dilated convolutions, further enhancing performance.\n", "date": "2024-01-05", "categories": []}
{"id": "2401.01461", "title": "Efficient Hybrid Zoom using Camera Fusion on Mobile Phones", "abstract": "  DSLR cameras can achieve multiple zoom levels via shifting lens distances or\nswapping lens types. However, these techniques are not possible on smartphone\ndevices due to space constraints. Most smartphone manufacturers adopt a hybrid\nzoom system: commonly a Wide (W) camera at a low zoom level and a Telephoto (T)\ncamera at a high zoom level. To simulate zoom levels between W and T, these\nsystems crop and digitally upsample images from W, leading to significant\ndetail loss. In this paper, we propose an efficient system for hybrid zoom\nsuper-resolution on mobile devices, which captures a synchronous pair of W and\nT shots and leverages machine learning models to align and transfer details\nfrom T to W. We further develop an adaptive blending method that accounts for\ndepth-of-field mismatches, scene occlusion, flow uncertainty, and alignment\nerrors. To minimize the domain gap, we design a dual-phone camera rig to\ncapture real-world inputs and ground-truths for supervised training. Our method\ngenerates a 12-megapixel image in 500ms on a mobile platform and compares\nfavorably against state-of-the-art methods under extensive evaluation on\nreal-world scenarios.\n", "date": "2024-01-02", "categories": []}
{"id": "2401.01662", "title": "Simultaneous q-Space Sampling Optimization and Reconstruction for Fast\n  and High-fidelity Diffusion Magnetic Resonance Imaging", "abstract": "  Diffusion Magnetic Resonance Imaging (dMRI) plays a crucial role in the\nnoninvasive investigation of tissue microstructural properties and structural\nconnectivity in the \\textit{in vivo} human brain. However, to effectively\ncapture the intricate characteristics of water diffusion at various directions\nand scales, it is important to employ comprehensive q-space sampling.\nUnfortunately, this requirement leads to long scan times, limiting the clinical\napplicability of dMRI. To address this challenge, we propose SSOR, a\nSimultaneous q-Space sampling Optimization and Reconstruction framework. We\njointly optimize a subset of q-space samples using a continuous representation\nof spherical harmonic functions and a reconstruction network. Additionally, we\nintegrate the unique properties of diffusion magnetic resonance imaging (dMRI)\nin both the q-space and image domains by applying $l1$-norm and total-variation\nregularization. The experiments conducted on HCP data demonstrate that SSOR has\npromising strengths both quantitatively and qualitatively and exhibits\nrobustness to noise.\n", "date": "2024-01-03", "categories": []}
{"id": "2401.01097", "title": "Robust single-particle cryo-EM image denoising and restoration", "abstract": "  Cryo-electron microscopy (cryo-EM) has achieved near-atomic level resolution\nof biomolecules by reconstructing 2D micrographs. However, the resolution and\naccuracy of the reconstructed particles are significantly reduced due to the\nextremely low signal-to-noise ratio (SNR) and complex noise structure of\ncryo-EM images. In this paper, we introduce a diffusion model with\npost-processing framework to effectively denoise and restore single particle\ncryo-EM images. Our method outperforms the state-of-the-art (SOTA) denoising\nmethods by effectively removing structural noise that has not been addressed\nbefore. Additionally, more accurate and high-resolution three-dimensional\nreconstruction structures can be obtained from denoised cryo-EM images.\n", "date": "2024-01-02", "categories": []}
{"id": "2308.00942", "title": "On the use of deep learning for phase recovery", "abstract": "  Phase recovery (PR) refers to calculating the phase of the light field from\nits intensity measurements. As exemplified from quantitative phase imaging and\ncoherent diffraction imaging to adaptive optics, PR is essential for\nreconstructing the refractive index distribution or topography of an object and\ncorrecting the aberration of an imaging system. In recent years, deep learning\n(DL), often implemented through deep neural networks, has provided\nunprecedented support for computational imaging, leading to more efficient\nsolutions for various PR problems. In this review, we first briefly introduce\nconventional methods for PR. Then, we review how DL provides support for PR\nfrom the following three stages, namely, pre-processing, in-processing, and\npost-processing. We also review how DL is used in phase image processing.\nFinally, we summarize the work in DL for PR and outlook on how to better use DL\nto improve the reliability and efficiency in PR. Furthermore, we present a\nlive-updating resource (https://github.com/kqwang/phase-recovery) for readers\nto learn more about PR.\n", "date": "2023-08-02", "categories": []}
{"id": "2312.17561", "title": "Informative Rays Selection for Few-Shot Neural Radiance Fields", "abstract": "  Neural Radiance Fields (NeRF) have recently emerged as a powerful method for\nimage-based 3D reconstruction, but the lengthy per-scene optimization limits\ntheir practical usage, especially in resource-constrained settings. Existing\napproaches solve this issue by reducing the number of input views and\nregularizing the learned volumetric representation with either complex losses\nor additional inputs from other modalities. In this paper, we present KeyNeRF,\na simple yet effective method for training NeRF in few-shot scenarios by\nfocusing on key informative rays. Such rays are first selected at camera level\nby a view selection algorithm that promotes baseline diversity while\nguaranteeing scene coverage, then at pixel level by sampling from a probability\ndistribution based on local image entropy. Our approach performs favorably\nagainst state-of-the-art methods, while requiring minimal changes to existing\nNeRF codebases.\n", "date": "2023-12-29", "categories": []}
{"id": "2312.16247", "title": "Toward Accurate and Temporally Consistent Video Restoration from Raw\n  Data", "abstract": "  Denoising and demosaicking are two fundamental steps in reconstructing a\nclean full-color video from raw data, while performing video denoising and\ndemosaicking jointly, namely VJDD, could lead to better video restoration\nperformance than performing them separately. In addition to restoration\naccuracy, another key challenge to VJDD lies in the temporal consistency of\nconsecutive frames. This issue exacerbates when perceptual regularization terms\nare introduced to enhance video perceptual quality. To address these\nchallenges, we present a new VJDD framework by consistent and accurate latent\nspace propagation, which leverages the estimation of previous frames as prior\nknowledge to ensure consistent recovery of the current frame. A data temporal\nconsistency (DTC) loss and a relational perception consistency (RPC) loss are\naccordingly designed. Compared with the commonly used flow-based losses, the\nproposed losses can circumvent the error accumulation problem caused by\ninaccurate flow estimation and effectively handle intensity changes in videos,\nimproving much the temporal consistency of output videos while preserving\ntexture details. Extensive experiments demonstrate the leading VJDD performance\nof our method in term of restoration accuracy, perceptual quality and temporal\nconsistency. Codes and dataset are available at\n\\url{https://github.com/GuoShi28/VJDD}.\n", "date": "2023-12-25", "categories": []}
{"id": "2312.16959", "title": "Efficient Physics-Based Learned Reconstruction Methods for Real-Time 3D\n  Near-Field MIMO Radar Imaging", "abstract": "  Near-field multiple-input multiple-output (MIMO) radar imaging systems have\nrecently gained significant attention. In this paper, we develop novel\nnon-iterative deep learning-based reconstruction methods for real-time\nnear-field MIMO imaging. The goal is to achieve high image quality with low\ncomputational cost at compressive settings. The developed approaches have two\nstages. In the first approach, physics-based initial stage performs adjoint\noperation to back-project the measurements to the image-space, and deep neural\nnetwork (DNN)-based second stage converts the 3D backprojected measurements to\na magnitude-only reflectivity image. Since scene reflectivities often have\nrandom phase, DNN processes directly the magnitude of the adjoint result. As\nDNN, 3D U-Net is used to jointly exploit range and cross-range correlations. To\ncomparatively evaluate the significance of exploiting physics in a\nlearning-based approach, two additional approaches that replace the\nphysics-based first stage with fully connected layers are also developed as\npurely learning-based methods. The performance is also analyzed by changing the\nDNN architecture for the second stage to include complex-valued processing\n(instead of magnitude-only processing), 2D convolution kernels (instead of 3D),\nand ResNet architecture (instead of U-Net). Moreover, we develop a synthesizer\nto generate large-scale dataset for training with 3D extended targets. We\nillustrate the performance through experimental data and extensive simulations.\nThe results show the effectiveness of the developed physics-based learned\nreconstruction approach in terms of both run-time and image quality at highly\ncompressive settings. Our source codes and dataset are made available at\nGitHub.\n", "date": "2023-12-28", "categories": []}
{"id": "2312.16987", "title": "Image Quality, Uniformity and Computation Improvement of Compressive\n  Light Field Displays with U-Net", "abstract": "  We apply the U-Net model for compressive light field synthesis. Compared to\nmethods based on stacked CNN and iterative algorithms, this method offers\nbetter image quality, uniformity and less computation.\n", "date": "2023-12-28", "categories": []}
{"id": "2312.17234", "title": "Personalized Restoration via Dual-Pivot Tuning", "abstract": "  Generative diffusion models can serve as a prior which ensures that solutions\nof image restoration systems adhere to the manifold of natural images. However,\nfor restoring facial images, a personalized prior is necessary to accurately\nrepresent and reconstruct unique facial features of a given individual. In this\npaper, we propose a simple, yet effective, method for personalized restoration,\ncalled Dual-Pivot Tuning - a two-stage approach that personalize a blind\nrestoration system while maintaining the integrity of the general prior and the\ndistinct role of each component. Our key observation is that for optimal\npersonalization, the generative model should be tuned around a fixed text\npivot, while the guiding network should be tuned in a generic\n(non-personalized) manner, using the personalized generative model as a fixed\n``pivot\". This approach ensures that personalization does not interfere with\nthe restoration process, resulting in a natural appearance with high fidelity\nto the person's identity and the attributes of the degraded image. We evaluated\nour approach both qualitatively and quantitatively through extensive\nexperiments with images of widely recognized individuals, comparing it against\nrelevant baselines. Surprisingly, we found that our personalized prior not only\nachieves higher fidelity to identity with respect to the person's identity, but\nalso outperforms state-of-the-art generic priors in terms of general image\nquality. Project webpage: https://personalized-restoration.github.io\n", "date": "2023-12-28", "categories": []}
{"id": "2303.08977", "title": "DeblurSR: Event-Based Motion Deblurring Under the Spiking Representation", "abstract": "  We present DeblurSR, a novel motion deblurring approach that converts a\nblurry image into a sharp video. DeblurSR utilizes event data to compensate for\nmotion ambiguities and exploits the spiking representation to parameterize the\nsharp output video as a mapping from time to intensity. Our key contribution,\nthe Spiking Representation (SR), is inspired by the neuromorphic principles\ndetermining how biological neurons communicate with each other in living\norganisms. We discuss why the spikes can represent sharp edges and how the\nspiking parameters are interpreted from the neuromorphic perspective. DeblurSR\nhas higher output quality and requires fewer computing resources than\nstate-of-the-art event-based motion deblurring methods. We additionally show\nthat our approach easily extends to video super-resolution when combined with\nrecent advances in implicit neural representation. The implementation and\nanimated visualization of DeblurSR are available at\nhttps://github.com/chensong1995/DeblurSR.\n", "date": "2023-03-15", "categories": []}
{"id": "2308.03448", "title": "Make Explicit Calibration Implicit: Calibrate Denoiser Instead of the\n  Noise Model", "abstract": "  Explicit calibration-based methods have dominated RAW image denoising under\nextremely low-light environments. However, these methods are impeded by several\ncritical limitations: a) the explicit calibration process is both labor- and\ntime-intensive, b) challenge exists in transferring denoisers across different\ncamera models, and c) the disparity between synthetic and real noise is\nexacerbated by digital gain. To address these issues, we introduce a\ngroundbreaking pipeline named Lighting Every Darkness (LED), which is effective\nregardless of the digital gain or the camera sensor. LED eliminates the need\nfor explicit noise model calibration, instead utilizing an implicit fine-tuning\nprocess that allows quick deployment and requires minimal data. Structural\nmodifications are also included to reduce the discrepancy between synthetic and\nreal noise without extra computational demands. Our method surpasses existing\nmethods in various camera models, including new ones not in public datasets,\nwith just a few pairs per digital gain and only 0.5% of the typical iterations.\nFurthermore, LED also allows researchers to focus more on deep learning\nadvancements while still utilizing sensor engineering benefits. Code and\nrelated materials can be found in https://srameo.github.io/projects/led-iccv23/ .\n", "date": "2023-08-07", "categories": []}
{"id": "2308.12009", "title": "StofNet: Super-resolution Time of Flight Network", "abstract": "  Time of Flight (ToF) is a prevalent depth sensing technology in the fields of\nrobotics, medical imaging, and non-destructive testing. Yet, ToF sensing faces\nchallenges from complex ambient conditions making an inverse modelling from the\nsparse temporal information intractable. This paper highlights the potential of\nmodern super-resolution techniques to learn varying surroundings for a reliable\nand accurate ToF detection. Unlike existing models, we tailor an architecture\nfor sub-sample precise semi-global signal localization by combining\nsuper-resolution with an efficient residual contraction block to balance\nbetween fine signal details and large scale contextual information. We\nconsolidate research on ToF by conducting a benchmark comparison against six\nstate-of-the-art methods for which we employ two publicly available datasets.\nThis includes the release of our SToF-Chirp dataset captured by an airborne\nultrasound transducer. Results showcase the superior performance of our\nproposed StofNet in terms of precision, reliability and model complexity. Our\ncode is available at https://github.com/hahnec/stofnet.\n", "date": "2023-08-23", "categories": []}
{"id": "2312.15136", "title": "Towards End-to-End Structure Solutions from Information-Compromised\n  Diffraction Data via Generative Deep Learning", "abstract": "  The revolution in materials in the past century was built on a knowledge of\nthe atomic arrangements and the structure-property relationship. The sine qua\nnon for obtaining quantitative structural information is single crystal\ncrystallography. However, increasingly we need to solve structures in cases\nwhere the information content in our input signal is significantly degraded,\nfor example, due to orientational averaging of grains, finite size effects due\nto nanostructure, and mixed signals due to sample heterogeneity. Understanding\nthe structure property relationships in such situations is, if anything, more\nimportant and insightful, yet we do not have robust approaches for\naccomplishing it. In principle, machine learning (ML) and deep learning (DL)\nare promising approaches since they augment information in the degraded input\nsignal with prior knowledge learned from large databases of already known\nstructures. Here we present a novel ML approach, a variational query-based\nmulti-branch deep neural network that has the promise to be a robust but\ngeneral tool to address this problem end-to-end. We demonstrate the approach on\ncomputed powder x-ray diffraction (PXRD), along with partial chemical\ncomposition information, as input. We choose as a structural representation a\nmodified electron density we call the Cartesian mapped electron density (CMED),\nthat straightforwardly allows our ML model to learn material structures across\ndifferent chemistries, symmetries and crystal systems. When evaluated on\ntheoretically simulated data for the cubic and trigonal crystal systems, the\nsystem achieves up to $93.4\\%$ average similarity with the ground truth on\nunseen materials, both with known and partially-known chemical composition\ninformation, showing great promise for successful structure solution even from\ndegraded and incomplete input data.\n", "date": "2023-12-23", "categories": []}
{"id": "2312.15487", "title": "BSRAW: Improving Blind RAW Image Super-Resolution", "abstract": "  In smartphones and compact cameras, the Image Signal Processor (ISP)\ntransforms the RAW sensor image into a human-readable sRGB image. Most popular\nsuper-resolution methods depart from a sRGB image and upscale it further,\nimproving its quality. However, modeling the degradations in the sRGB domain is\ncomplicated because of the non-linear ISP transformations. Despite this known\nissue, only a few methods work directly with RAW images and tackle real-world\nsensor degradations. We tackle blind image super-resolution in the RAW domain.\nWe design a realistic degradation pipeline tailored specifically for training\nmodels with raw sensor data. Our approach considers sensor noise, defocus,\nexposure, and other common issues. Our BSRAW models trained with our pipeline\ncan upscale real-scene RAW images and improve their quality. As part of this\neffort, we also present a new DSLM dataset and benchmark for this task.\n", "date": "2023-12-24", "categories": []}
{"id": "2312.15575", "title": "Neural Born Series Operator for Biomedical Ultrasound Computed\n  Tomography", "abstract": "  Ultrasound Computed Tomography (USCT) provides a radiation-free option for\nhigh-resolution clinical imaging. Despite its potential, the computationally\nintensive Full Waveform Inversion (FWI) required for tissue property\nreconstruction limits its clinical utility. This paper introduces the Neural\nBorn Series Operator (NBSO), a novel technique designed to speed up wave\nsimulations, thereby facilitating a more efficient USCT image reconstruction\nprocess through an NBSO-based FWI pipeline. Thoroughly validated on\ncomprehensive brain and breast datasets, simulated under experimental USCT\nconditions, the NBSO proves to be accurate and efficient in both forward\nsimulation and image reconstruction. This advancement demonstrates the\npotential of neural operators in facilitating near real-time USCT\nreconstruction, making the clinical application of USCT increasingly viable and\npromising.\n", "date": "2023-12-25", "categories": []}
{"id": "2312.15970", "title": "Learning Deformable Hypothesis Sampling for Accurate PatchMatch\n  Multi-View Stereo", "abstract": "  This paper introduces a learnable Deformable Hypothesis Sampler\n(DeformSampler) to address the challenging issue of noisy depth estimation for\naccurate PatchMatch Multi-View Stereo (MVS). We observe that the heuristic\ndepth hypothesis sampling modes employed by PatchMatch MVS solvers are\ninsensitive to (i) the piece-wise smooth distribution of depths across the\nobject surface, and (ii) the implicit multi-modal distribution of depth\nprediction probabilities along the ray direction on the surface points.\nAccordingly, we develop DeformSampler to learn distribution-sensitive sample\nspaces to (i) propagate depths consistent with the scene's geometry across the\nobject surface, and (ii) fit a Laplace Mixture model that approaches the\npoint-wise probabilities distribution of the actual depths along the ray\ndirection. We integrate DeformSampler into a learnable PatchMatch MVS system to\nenhance depth estimation in challenging areas, such as piece-wise discontinuous\nsurface boundaries and weakly-textured regions. Experimental results on DTU and\nTanks \\& Temples datasets demonstrate its superior performance and\ngeneralization capabilities compared to state-of-the-art competitors. Code is\navailable at https://github.com/Geo-Tell/DS-PMNet.\n", "date": "2023-12-26", "categories": []}
{"id": "2312.14235", "title": "Neural Spline Fields for Burst Image Fusion and Layer Separation", "abstract": "  Each photo in an image burst can be considered a sample of a complex 3D\nscene: the product of parallax, diffuse and specular materials, scene motion,\nand illuminant variation. While decomposing all of these effects from a stack\nof misaligned images is a highly ill-conditioned task, the conventional\nalign-and-merge burst pipeline takes the other extreme: blending them into a\nsingle image. In this work, we propose a versatile intermediate representation:\na two-layer alpha-composited image plus flow model constructed with neural\nspline fields -- networks trained to map input coordinates to spline control\npoints. Our method is able to, during test-time optimization, jointly fuse a\nburst image capture into one high-resolution reconstruction and decompose it\ninto transmission and obstruction layers. Then, by discarding the obstruction\nlayer, we can perform a range of tasks including seeing through occlusions,\nreflection suppression, and shadow removal. Validated on complex synthetic and\nin-the-wild captures we find that, with no post-processing steps or learned\npriors, our generalizable model is able to outperform existing dedicated\nsingle-image and multi-view obstruction removal approaches.\n", "date": "2023-12-21", "categories": []}
{"id": "2312.14432", "title": "Scalable 3D Reconstruction From Single Particle X-Ray Diffraction Images\n  Based on Online Machine Learning", "abstract": "  X-ray free-electron lasers (XFELs) offer unique capabilities for measuring\nthe structure and dynamics of biomolecules, helping us understand the basic\nbuilding blocks of life. Notably, high-repetition-rate XFELs enable single\nparticle imaging (X-ray SPI) where individual, weakly scattering biomolecules\nare imaged under near-physiological conditions with the opportunity to access\nfleeting states that cannot be captured in cryogenic or crystallized\nconditions. Existing X-ray SPI reconstruction algorithms, which estimate the\nunknown orientation of a particle in each captured image as well as its shared\n3D structure, are inadequate in handling the massive datasets generated by\nthese emerging XFELs. Here, we introduce X-RAI, an online reconstruction\nframework that estimates the structure of a 3D macromolecule from large X-ray\nSPI datasets. X-RAI consists of a convolutional encoder, which amortizes pose\nestimation over large datasets, as well as a physics-based decoder, which\nemploys an implicit neural representation to enable high-quality 3D\nreconstruction in an end-to-end, self-supervised manner. We demonstrate that\nX-RAI achieves state-of-the-art performance for small-scale datasets in\nsimulation and challenging experimental settings and demonstrate its\nunprecedented ability to process large datasets containing millions of\ndiffraction images in an online fashion. These abilities signify a paradigm\nshift in X-ray SPI towards real-time capture and reconstruction.\n", "date": "2023-12-22", "categories": []}
{"id": "2302.11883", "title": "PIFON-EPT: MR-Based Electrical Property Tomography Using\n  Physics-Informed Fourier Networks", "abstract": "  We propose Physics-Informed Fourier Networks for Electrical Properties (EP)\nTomography (PIFON-EPT), a novel deep learning-based method for EP\nreconstruction using noisy and/or incomplete magnetic resonance (MR)\nmeasurements. Our approach leverages the Helmholtz equation to constrain two\nnetworks, responsible for the denoising and completion of the transmit fields,\nand the estimation of the object's EP, respectively. We embed a random Fourier\nfeatures mapping into our networks to enable efficient learning of\nhigh-frequency details encoded in the transmit fields. We demonstrated the\nefficacy of PIFON-EPT through several simulated experiments at 3 and 7 tesla\n(T) MR imaging, and showed that our method can reconstruct physically\nconsistent EP and transmit fields. Specifically, when only $20\\%$ of the noisy\nmeasured fields were used as inputs, PIFON-EPT reconstructed the EP of a\nphantom with $\\leq 5\\%$ error, and denoised and completed the measurements with\n$\\leq 1\\%$ error. Additionally, we adapted PIFON-EPT to solve the generalized\nHelmholtz equation that accounts for gradients of EP between inhomogeneities.\nThis yielded improved results at interfaces between different materials without\nexplicit knowledge of boundary conditions. PIFON-EPT is the first method that\ncan simultaneously reconstruct EP and transmit fields from incomplete noisy MR\nmeasurements, providing new opportunities for EPT research.\n", "date": "2023-02-23", "categories": []}
{"id": "2311.16512", "title": "CoSeR: Bridging Image and Language for Cognitive Super-Resolution", "abstract": "  Existing super-resolution (SR) models primarily focus on restoring local\ntexture details, often neglecting the global semantic information within the\nscene. This oversight can lead to the omission of crucial semantic details or\nthe introduction of inaccurate textures during the recovery process. In our\nwork, we introduce the Cognitive Super-Resolution (CoSeR) framework, empowering\nSR models with the capacity to comprehend low-resolution images. We achieve\nthis by marrying image appearance and language understanding to generate a\ncognitive embedding, which not only activates prior information from large\ntext-to-image diffusion models but also facilitates the generation of\nhigh-quality reference images to optimize the SR process. To further improve\nimage fidelity, we propose a novel condition injection scheme called\n\"All-in-Attention\", consolidating all conditional information into a single\nmodule. Consequently, our method successfully restores semantically correct and\nphotorealistic details, demonstrating state-of-the-art performance across\nmultiple benchmarks. Code: https://github.com/VINHYU/CoSeR\n", "date": "2023-11-27", "categories": []}
{"id": "2312.06723", "title": "Learning to See Low-Light Images via Feature Domain Adaptation", "abstract": "  Raw low light image enhancement (LLIE) has achieved much better performance\nthan the sRGB domain enhancement methods due to the merits of raw data.\nHowever, the ambiguity between noisy to clean and raw to sRGB mappings may\nmislead the single-stage enhancement networks. The two-stage networks avoid\nambiguity by decoupling the two mappings but usually have large computing\ncomplexity. To solve this problem, we propose a single-stage network empowered\nby Feature Domain Adaptation (FDA) to decouple the denoising and color mapping\ntasks in raw LLIE. The denoising encoder is supervised by the clean raw image,\nand then the denoised features are adapted for the color mapping task by an FDA\nmodule. We propose a Lineformer to serve as the FDA, which can well explore the\nglobal and local correlations with fewer line buffers (friendly to the\nline-based imaging process). During inference, the raw supervision branch is\nremoved. In this way, our network combines the advantage of a two-stage\nenhancement process with the efficiency of single-stage inference. Experiments\non four benchmark datasets demonstrate that our method achieves\nstate-of-the-art performance with fewer computing costs (60% FLOPs of the\ntwo-stage method DNF). Our codes will be released after the acceptance of this\nwork.\n", "date": "2023-12-11", "categories": []}
{"id": "2312.13310", "title": "Computational Spectral Imaging with Unified Encoding Model: A\n  Comparative Study and Beyond", "abstract": "  Computational spectral imaging is drawing increasing attention owing to the\nsnapshot advantage, and amplitude, phase, and wavelength encoding systems are\nthree types of representative implementations. Fairly comparing and\nunderstanding the performance of these systems is essential, but challenging\ndue to the heterogeneity in encoding design. To overcome this limitation, we\npropose the unified encoding model (UEM) that covers all physical systems using\nthe three encoding types. Specifically, the UEM comprises physical amplitude,\nphysical phase, and physical wavelength encoding models that can be combined\nwith a digital decoding model in a joint encoder-decoder optimization framework\nto compare the three systems under a unified experimental setup fairly.\nFurthermore, we extend the UEMs to ideal versions, namely, ideal amplitude,\nideal phase, and ideal wavelength encoding models, which are free from physical\nconstraints, to explore the full potential of the three types of computational\nspectral imaging systems. Finally, we conduct a holistic comparison of the\nthree types of computational spectral imaging systems and provide valuable\ninsights for designing and exploiting these systems in the future.\n", "date": "2023-12-20", "categories": []}
{"id": "2312.13820", "title": "Super-resolution of THz time-domain images based on low-rank\n  representation", "abstract": "  Terahertz time-domain spectroscopy (THz-TDS) employs sub-picosecond pulses to\nprobe dielectric properties of materials giving as a result a 3-dimensional\nhyperspectral data cube. The spatial resolution of THz images is primarily\nlimited by two sources: a non-zero THz beam waist and the acquisition step\nsize. Acquisition with a small step size allows for the visualisation of\nsmaller details in images at the expense of acquisition time, but the\nfrequency-dependent point-spread function remains the biggest bottleneck for\nTHz imaging. This work presents a super-resolution approach to restore THz\ntime-domain images acquired with medium-to-big step sizes. The results show the\noptimized and robust performance for different frequency bands (from 0.5 to 3.5\nTHz) obtaining higher resolution and additionally removing effects of blur at\nlower frequencies and noise at higher frequencies.\n", "date": "2023-12-21", "categories": []}
{"id": "2211.10525", "title": "Differentiable Uncalibrated Imaging", "abstract": "  We propose a differentiable imaging framework to address uncertainty in\nmeasurement coordinates such as sensor locations and projection angles. We\nformulate the problem as measurement interpolation at unknown nodes supervised\nthrough the forward operator. To solve it we apply implicit neural networks,\nalso known as neural fields, which are naturally differentiable with respect to\nthe input coordinates. We also develop differentiable spline interpolators\nwhich perform as well as neural networks, require less time to optimize and\nhave well-understood properties. Differentiability is key as it allows us to\njointly fit a measurement representation, optimize over the uncertain\nmeasurement coordinates, and perform image reconstruction which in turn ensures\nconsistent calibration. We apply our approach to 2D and 3D computed tomography,\nand show that it produces improved reconstructions compared to baselines that\ndo not account for the lack of calibration. The flexibility of the proposed\nframework makes it easy to extend to almost arbitrary imaging problems.\n", "date": "2022-11-18", "categories": []}
{"id": "2308.13739", "title": "Devignet: High-Resolution Vignetting Removal via a Dual Aggregated\n  Fusion Transformer With Adaptive Channel Expansion", "abstract": "  Vignetting commonly occurs as a degradation in images resulting from factors\nsuch as lens design, improper lens hood usage, and limitations in camera\nsensors. This degradation affects image details, color accuracy, and presents\nchallenges in computational photography. Existing vignetting removal algorithms\npredominantly rely on ideal physics assumptions and hand-crafted parameters,\nresulting in the ineffective removal of irregular vignetting and suboptimal\nresults. Moreover, the substantial lack of real-world vignetting datasets\nhinders the objective and comprehensive evaluation of vignetting removal. To\naddress these challenges, we present Vigset, a pioneering dataset for\nvignetting removal. Vigset includes 983 pairs of both vignetting and\nvignetting-free high-resolution ($5340\\times3697$) real-world images under\nvarious conditions. In addition, We introduce DeVigNet, a novel frequency-aware\nTransformer architecture designed for vignetting removal. Through the Laplacian\nPyramid decomposition, we propose the Dual Aggregated Fusion Transformer to\nhandle global features and remove vignetting in the low-frequency domain.\nAdditionally, we propose the Adaptive Channel Expansion Module to enhance\ndetails in the high-frequency domain. The experiments demonstrate that the\nproposed model outperforms existing state-of-the-art methods. The code, models,\nand dataset are available at \\url{https://github.com/CXH-Research/DeVigNet}.\n", "date": "2023-08-26", "categories": []}
{"id": "2312.12644", "title": "Rotational Augmented Noise2Inverse for Low-dose Computed Tomography\n  Reconstruction", "abstract": "  In this work, we present a novel self-supervised method for Low Dose Computed\nTomography (LDCT) reconstruction. Reducing the radiation dose to patients\nduring a CT scan is a crucial challenge since the quality of the reconstruction\nhighly degrades because of low photons or limited measurements. Supervised deep\nlearning methods have shown the ability to remove noise in images but require\naccurate ground truth which can be obtained only by performing additional\nhigh-radiation CT scans. Therefore, we propose a novel self-supervised\nframework for LDCT, in which ground truth is not required for training the\nconvolutional neural network (CNN). Based on the Noise2Inverse (N2I) method, we\nenforce in the training loss the equivariant property of rotation\ntransformation, which is induced by the CT imaging system, to improve the\nquality of the CT image in a lower dose. Numerical and experimental results\nshow that the reconstruction accuracy of N2I with sparse views is degrading\nwhile the proposed rotational augmented Noise2Inverse (RAN2I) method keeps\nbetter image quality over a different range of sampling angles. Finally, the\nquantitative results demonstrate that RAN2I achieves higher image quality\ncompared to N2I, and experimental results of RAN2I on real projection data show\ncomparable performance to supervised learning.\n", "date": "2023-12-19", "categories": []}
{"id": "2312.12961", "title": "Radar Fields: An Extension of Radiance Fields to SAR", "abstract": "  Radiance fields have been a major breakthrough in the field of inverse\nrendering, novel view synthesis and 3D modeling of complex scenes from\nmulti-view image collections. Since their introduction, it was shown that they\ncould be extended to other modalities such as LiDAR, radio frequencies, X-ray\nor ultrasound. In this paper, we show that, despite the important difference\nbetween optical and synthetic aperture radar (SAR) image formation models, it\nis possible to extend radiance fields to radar images thus presenting the first\n\"radar fields\". This allows us to learn surface models using only collections\nof radar images, similar to how regular radiance fields are learned and with\nthe same computational complexity on average. Thanks to similarities in how\nboth fields are defined, this work also shows a potential for hybrid methods\ncombining both optical and SAR images.\n", "date": "2023-12-20", "categories": []}
{"id": "2302.02854", "title": "NA-SODINN: a deep learning algorithm for exoplanet image detection based\n  on residual noise regimes", "abstract": "  Supervised deep learning was recently introduced in high-contrast imaging\n(HCI) through the SODINN algorithm, a convolutional neural network designed for\nexoplanet detection in angular differential imaging (ADI) datasets. The\nbenchmarking of HCI algorithms within the Exoplanet Imaging Data Challenge\n(EIDC) showed that (i) SODINN can produce a high number of false positives in\nthe final detection maps, and (ii) algorithms processing images in a more local\nmanner perform better. This work aims to improve the SODINN detection\nperformance by introducing new local processing approaches and adapting its\nlearning process accordingly. We propose NA-SODINN, a new deep learning binary\nclassifier based on a convolutional neural network (CNN) that better captures\nimage noise correlations in ADI-processed frames by identifying noise regimes.\nOur new approach was tested against its predecessor, as well as two\nSODINN-based hybrid models and a more standard annular-PCA approach, through\nlocal receiving operating characteristics (ROC) analysis of ADI sequences from\nthe VLT/SPHERE and Keck/NIRC-2 instruments. Results show that NA-SODINN\nenhances SODINN in both sensitivity and specificity, especially in the\nspeckle-dominated noise regime. NA-SODINN is also benchmarked against the\ncomplete set of submitted detection algorithms in EIDC, in which we show that\nits final detection score matches or outperforms the most powerful detection\nalgorithms.Throughout the supervised machine learning case, this study\nillustrates and reinforces the importance of adapting the task of detection to\nthe local content of processed images.\n", "date": "2023-02-06", "categories": []}
{"id": "2309.02318", "title": "TiAVox: Time-aware Attenuation Voxels for Sparse-view 4D DSA\n  Reconstruction", "abstract": "  Four-dimensional Digital Subtraction Angiography (4D DSA) plays a critical\nrole in the diagnosis of many medical diseases, such as Arteriovenous\nMalformations (AVM) and Arteriovenous Fistulas (AVF). Despite its significant\napplication value, the reconstruction of 4D DSA demands numerous views to\neffectively model the intricate vessels and radiocontrast flow, thereby\nimplying a significant radiation dose. To address this high radiation issue, we\npropose a Time-aware Attenuation Voxel (TiAVox) approach for sparse-view 4D DSA\nreconstruction, which paves the way for high-quality 4D imaging. Additionally,\n2D and 3D DSA imaging results can be generated from the reconstructed 4D DSA\nimages. TiAVox introduces 4D attenuation voxel grids, which reflect attenuation\nproperties from both spatial and temporal dimensions. It is optimized by\nminimizing discrepancies between the rendered images and sparse 2D DSA images.\nWithout any neural network involved, TiAVox enjoys specific physical\ninterpretability. The parameters of each learnable voxel represent the\nattenuation coefficients. We validated the TiAVox approach on both clinical and\nsimulated datasets, achieving a 31.23 Peak Signal-to-Noise Ratio (PSNR) for\nnovel view synthesis using only 30 views on the clinically sourced dataset,\nwhereas traditional Feldkamp-Davis-Kress methods required 133 views. Similarly,\nwith merely 10 views from the synthetic dataset, TiAVox yielded a PSNR of 34.32\nfor novel view synthesis and 41.40 for 3D reconstruction. We also executed\nablation studies to corroborate the essential components of TiAVox. The code\nwill be publically available.\n", "date": "2023-09-05", "categories": []}
{"id": "2312.09754", "title": "PPFM: Image denoising in photon-counting CT using single-step posterior\n  sampling Poisson flow generative models", "abstract": "  Diffusion and Poisson flow models have shown impressive performance in a wide\nrange of generative tasks, including low-dose CT image denoising. However, one\nlimitation in general, and for clinical applications in particular, is slow\nsampling. Due to their iterative nature, the number of function evaluations\n(NFE) required is usually on the order of $10-10^3$, both for conditional and\nunconditional generation. In this paper, we present posterior sampling Poisson\nflow generative models (PPFM), a novel image denoising technique for low-dose\nand photon-counting CT that produces excellent image quality whilst keeping\nNFE=1. Updating the training and sampling processes of Poisson flow generative\nmodels (PFGM)++, we learn a conditional generator which defines a trajectory\nbetween the prior noise distribution and the posterior distribution of\ninterest. We additionally hijack and regularize the sampling process to achieve\nNFE=1. Our results shed light on the benefits of the PFGM++ framework compared\nto diffusion models. In addition, PPFM is shown to perform favorably compared\nto current state-of-the-art diffusion-style models with NFE=1, consistency\nmodels, as well as popular deep learning and non-deep learning-based image\ndenoising techniques, on clinical low-dose CT images and clinical images from a\nprototype photon-counting CT system.\n", "date": "2023-12-15", "categories": []}
{"id": "2312.12064", "title": "MPI Planar Correction of Pulse Based ToF Cameras", "abstract": "  Time-of-Flight (ToF) cameras are becoming popular in a wide span of areas\nranging from consumer-grade electronic devices to safety-critical industrial\nrobots. This is mainly due to their high frame rate, relative good precision\nand the lowered costs. Although ToF cameras are in continuous development,\nespecially pulse-based variants, they still face different problems, including\nspurious noise over the points or multipath inference (MPI). The latter can\ncause deformed surfaces to manifest themselves on curved surfaces instead of\nplanar ones, making standard spatial data preprocessing, such as plane\nextraction, difficult. In this paper, we focus on the MPI reduction problem\nusing Feature Pyramid Networks (FPN) which allow the mitigation of this type of\nartifact for pulse-based ToF cameras. With our end-to-end network, we managed\nto attenuate the MPI effect on planar surfaces using a learning-based method on\nreal ToF data. Both the custom dataset used for our model training as well as\nthe code is available on the author's Github homepage.\n", "date": "2023-12-19", "categories": []}
{"id": "2307.03998", "title": "Lightweight Improved Residual Network for Efficient Inverse Tone Mapping", "abstract": "  The display devices like HDR10 televisions are increasingly prevalent in our\ndaily life for visualizing high dynamic range (HDR) images. But the majority of\nmedia images on the internet remain in 8-bit standard dynamic range (SDR)\nformat. Therefore, converting SDR images to HDR ones by inverse tone mapping\n(ITM) is crucial to unlock the full potential of abundant media images.\nHowever, existing ITM methods are usually developed with complex network\narchitectures requiring huge computational costs. In this paper, we propose a\nlightweight Improved Residual Network (IRNet) by enhancing the power of popular\nresidual block for efficient ITM. Specifically, we propose a new Improved\nResidual Block (IRB) to extract and fuse multi-layer features for fine-grained\nHDR image reconstruction. Experiments on three benchmark datasets demonstrate\nthat our IRNet achieves state-of-the-art performance on both the ITM and joint\nSR-ITM tasks. The code, models and data will be publicly available at\nhttps://github.com/ThisisVikki/ITM-baseline.\n", "date": "2023-07-08", "categories": []}
{"id": "2309.16899", "title": "On the Contractivity of Plug-and-Play Operators", "abstract": "  In plug-and-play (PnP) regularization, the proximal operator in algorithms\nsuch as ISTA and ADMM is replaced by a powerful denoiser. This formal\nsubstitution works surprisingly well in practice. In fact, PnP has been shown\nto give state-of-the-art results for various imaging applications. The\nempirical success of PnP has motivated researchers to understand its\ntheoretical underpinnings and, in particular, its convergence. It was shown in\nprior work that for kernel denoisers such as the nonlocal means, PnP-ISTA\nprovably converges under some strong assumptions on the forward model. The\npresent work is motivated by the following questions: Can we relax the\nassumptions on the forward model? Can the convergence analysis be extended to\nPnP-ADMM? Can we estimate the convergence rate? In this letter, we resolve\nthese questions using the contraction mapping theorem: (i) for symmetric\ndenoisers, we show that (under mild conditions) PnP-ISTA and PnP-ADMM exhibit\nlinear convergence; and (ii) for kernel denoisers, we show that PnP-ISTA and\nPnP-ADMM converge linearly for image inpainting. We validate our theoretical\nfindings using reconstruction experiments.\n", "date": "2023-09-28", "categories": []}
{"id": "2312.10604", "title": "A Dual Domain Multi-exposure Image Fusion Network based on the\n  Spatial-Frequency Integration", "abstract": "  Multi-exposure image fusion aims to generate a single high-dynamic image by\nintegrating images with different exposures. Existing deep learning-based\nmulti-exposure image fusion methods primarily focus on spatial domain fusion,\nneglecting the global modeling ability of the frequency domain. To effectively\nleverage the global illumination modeling ability of the frequency domain, we\npropose a novelty perspective on multi-exposure image fusion via the\nSpatial-Frequency Integration Framework, named MEF-SFI. Initially, we revisit\nthe properties of the Fourier transform on the 2D image, and verify the\nfeasibility of multi-exposure image fusion on the frequency domain where the\namplitude and phase component is able to guide the integration of the\nillumination information. Subsequently, we present the deep Fourier-based\nmulti-exposure image fusion framework, which consists of a spatial path and\nfrequency path for local and global modeling separately. Specifically, we\nintroduce a Spatial-Frequency Fusion Block to facilitate efficient interaction\nbetween dual domains and capture complementary information from input images\nwith different exposures. Finally, we combine a dual domain loss function to\nensure the retention of complementary information in both the spatial and\nfrequency domains. Extensive experiments on the PQA-MEF dataset demonstrate\nthat our method achieves visual-appealing fusion results against\nstate-of-the-art multi-exposure image fusion approaches. Our code is available\nat https://github.com/SSyangguang/MEF-freq.\n", "date": "2023-12-17", "categories": []}
{"id": "2312.11119", "title": "Hyperspectral Image Reconstruction via Combinatorial Embedding of\n  Cross-Channel Spatio-Spectral Clues", "abstract": "  Existing learning-based hyperspectral reconstruction methods show limitations\nin fully exploiting the information among the hyperspectral bands. As such, we\npropose to investigate the chromatic inter-dependencies in their respective\nhyperspectral embedding space. These embedded features can be fully exploited\nby querying the inter-channel correlations in a combinatorial manner, with the\nunique and complementary information efficiently fused into the final\nprediction. We found such independent modeling and combinatorial excavation\nmechanisms are extremely beneficial to uncover marginal spectral features,\nespecially in the long wavelength bands. In addition, we have proposed a\nspatio-spectral attention block and a spectrum-fusion attention module, which\ngreatly facilitates the excavation and fusion of information at both\nsemantically long-range levels and fine-grained pixel levels across all\ndimensions. Extensive quantitative and qualitative experiments show that our\nmethod (dubbed CESST) achieves SOTA performance. Code for this project is at:\nhttps://github.com/AlexYangxx/CESST.\n", "date": "2023-12-18", "categories": []}
{"id": "2212.07786", "title": "Convergent Data-driven Regularizations for CT Reconstruction", "abstract": "  The reconstruction of images from their corresponding noisy Radon transform\nis a typical example of an ill-posed linear inverse problem as arising in the\napplication of computerized tomography (CT). As the (naive) solution does not\ndepend on the measured data continuously, regularization is needed to\nre-establish a continuous dependence. In this work, we investigate simple, but\nyet still provably convergent approaches to learning linear regularization\nmethods from data. More specifically, we analyze two approaches: One generic\nlinear regularization that learns how to manipulate the singular values of the\nlinear operator in an extension of our previous work, and one tailored approach\nin the Fourier domain that is specific to CT-reconstruction. We prove that such\napproaches become convergent regularization methods as well as the fact that\nthe reconstructions they provide are typically much smoother than the training\ndata they were trained on. Finally, we compare the spectral as well as the\nFourier-based approaches for CT-reconstruction numerically, discuss their\nadvantages and disadvantages and investigate the effect of discretization\nerrors at different resolutions.\n", "date": "2022-12-14", "categories": []}
{"id": "2312.09488", "title": "Sequence adaptive field-imperfection estimation (SAFE): retrospective\n  estimation and correction of $B_1^+$ and $B_0$ inhomogeneities for enhanced\n  MRF quantification", "abstract": "  $B_1^+$ and $B_0$ field-inhomogeneities can significantly reduce accuracy and\nrobustness of MRF's quantitative parameter estimates. Additional $B_1^+$ and\n$B_0$ calibration scans can mitigate this but add scan time and cannot be\napplied retrospectively to previously collected data. Here, we proposed a\ncalibration-free sequence-adaptive deep-learning framework, to estimate and\ncorrect for $B_1^+$ and $B_0$ effects of any MRF sequence. We demonstrate its\ncapability on arbitrary MRF sequences at 3T, where no training data were\npreviously obtained. Such approach can be applied to any previously-acquired\nand future MRF-scans. The flexibility in directly applying this framework to\nother quantitative sequences is also highlighted.\n", "date": "2023-12-15", "categories": []}
{"id": "2305.12852", "title": "Cycle Consistency-based Uncertainty Quantification of Neural Networks in\n  Inverse Imaging Problems", "abstract": "  Uncertainty estimation is critical for numerous applications of deep neural\nnetworks and draws growing attention from researchers. Here, we demonstrate an\nuncertainty quantification approach for deep neural networks used in inverse\nproblems based on cycle consistency. We build forward-backward cycles using the\nphysical forward model available and a trained deep neural network solving the\ninverse problem at hand, and accordingly derive uncertainty estimators through\nregression analysis on the consistency of these forward-backward cycles. We\ntheoretically analyze cycle consistency metrics and derive their relationship\nwith respect to uncertainty, bias, and robustness of the neural network\ninference. To demonstrate the effectiveness of these cycle consistency-based\nuncertainty estimators, we classified corrupted and out-of-distribution input\nimage data using some of the widely used image deblurring and super-resolution\nneural networks as testbeds. The blind testing of our method outperformed other\nmodels in identifying unseen input data corruption and distribution shifts.\nThis work provides a simple-to-implement and rapid uncertainty quantification\nmethod that can be universally applied to various neural networks used for\nsolving inverse problems.\n", "date": "2023-05-22", "categories": []}
{"id": "2305.16321", "title": "Eclipse: Disambiguating Illumination and Materials using Unintended\n  Shadows", "abstract": "  Decomposing an object's appearance into representations of its materials and\nthe surrounding illumination is difficult, even when the object's 3D shape is\nknown beforehand. This problem is especially challenging for diffuse objects:\nit is ill-conditioned because diffuse materials severely blur incoming light,\nand it is ill-posed because diffuse materials under high-frequency lighting can\nbe indistinguishable from shiny materials under low-frequency lighting. We show\nthat it is possible to recover precise materials and illumination -- even from\ndiffuse objects -- by exploiting unintended shadows, like the ones cast onto an\nobject by the photographer who moves around it. These shadows are a nuisance in\nmost previous inverse rendering pipelines, but here we exploit them as signals\nthat improve conditioning and help resolve material-lighting ambiguities. We\npresent a method based on differentiable Monte Carlo ray tracing that uses\nimages of an object to jointly recover its spatially-varying materials, the\nsurrounding illumination environment, and the shapes of the unseen light\noccluders who inadvertently cast shadows upon it.\n", "date": "2023-05-25", "categories": []}
{"id": "2312.07631", "title": "AI-driven projection tomography with multicore fibre-optic cell rotation", "abstract": "  Optical tomography has emerged as a non-invasive imaging method, providing\nthree-dimensional insights into subcellular structures and thereby enabling a\ndeeper understanding of cellular functions, interactions, and processes.\nConventional optical tomography methods are constrained by a limited\nillumination scanning range, leading to anisotropic resolution and incomplete\nimaging of cellular structures. To overcome this problem, we employ a compact\nmulti-core fibre-optic cell rotator system that facilitates precise optical\nmanipulation of cells within a microfluidic chip, achieving full-angle\nprojection tomography with isotropic resolution. Moreover, we demonstrate an\nAI-driven tomographic reconstruction workflow, which can be a paradigm shift\nfrom conventional computational methods, often demanding manual processing, to\na fully autonomous process. The performance of the proposed cell rotation\ntomography approach is validated through the three-dimensional reconstruction\nof cell phantoms and HL60 human cancer cells. The versatility of this\nlearning-based tomographic reconstruction workflow paves the way for its broad\napplication across diverse tomographic imaging modalities, including but not\nlimited to flow cytometry tomography and acoustic rotation tomography.\nTherefore, this AI-driven approach can propel advancements in cell biology,\naiding in the inception of pioneering therapeutics, and augmenting early-stage\ncancer diagnostics.\n", "date": "2023-12-12", "categories": []}
{"id": "2312.07911", "title": "Projective Parallel Single-Pixel Imaging: 3D Structured Light Scanning\n  Under Global Illumination", "abstract": "  We present projective parallel single-pixel imaging (pPSI), a 3D photography\nmethod that provides a robust and efficient way to analyze the light transport\nbehavior and enables separation of light effect due to global illumination,\nthereby achieving 3D structured light scanning under global illumination. The\nlight transport behavior is described by the light transport coefficients\n(LTC), which contain complete information for a projector camera pair, and is a\n4D data set. However, the capture of LTC is generally time consuming. The 4D\nLTC in pPSI are reduced to projection functions, thereby enabling a highly\nefficient data capture process. We introduce the local maximum constraint,\nwhich provides constraint for the location of candidate correspondence matching\npoints when projections are captured. Local slice extension (LSE) method is\nintroduced to accelerate the capture of projection functions. Optimization is\nconducted for pPSI under several situations. The number of projection functions\nrequired for pPSI is optimized and the influence of capture ratio in LSE on the\naccuracy of the correspondence matching points is investigated. Discussions and\nexperiments include two typical kinds of global illuminations:\ninter-reflections and subsurface scattering. The proposed method is validated\nwith several challenging scenarios, and outperforms the state-of-the-art\nmethods.\n", "date": "2023-12-13", "categories": []}
{"id": "2312.08118", "title": "Neural Radiance Fields for Transparent Object Using Visual Hull", "abstract": "  Unlike opaque object, novel view synthesis of transparent object is a\nchallenging task, because transparent object refracts light of background\ncausing visual distortions on the transparent object surface along the\nviewpoint change. Recently introduced Neural Radiance Fields (NeRF) is a view\nsynthesis method. Thanks to its remarkable performance improvement, lots of\nfollowing applications based on NeRF in various topics have been developed.\nHowever, if an object with a different refractive index is included in a scene\nsuch as transparent object, NeRF shows limited performance because refracted\nlight ray at the surface of the transparent object is not appropriately\nconsidered. To resolve the problem, we propose a NeRF-based method consisting\nof the following three steps: First, we reconstruct a three-dimensional shape\nof a transparent object using visual hull. Second, we simulate the refraction\nof the rays inside of the transparent object according to Snell's law. Last, we\nsample points through refracted rays and put them into NeRF. Experimental\nevaluation results demonstrate that our method addresses the limitation of\nconventional NeRF with transparent objects.\n", "date": "2023-12-13", "categories": []}
{"id": "2212.05909", "title": "NFResNet: Multi-scale and U-shaped Networks for Deblurring", "abstract": "  Multi-Scale and U-shaped Networks are widely used in various image\nrestoration problems, including deblurring. Keeping in mind the wide range of\napplications, we present a comparison of these architectures and their effects\non image deblurring. We also introduce a new block called as NFResblock. It\nconsists of a Fast Fourier Transformation layer and a series of modified\nNon-Linear Activation Free Blocks. Based on these architectures and additions,\nwe introduce NFResnet and NFResnet+, which are modified multi-scale and U-Net\narchitectures, respectively. We also use three different loss functions to\ntrain these architectures: Charbonnier Loss, Edge Loss, and Frequency\nReconstruction Loss. Extensive experiments on the Deep Video Deblurring\ndataset, along with ablation studies for each component, have been presented in\nthis paper. The proposed architectures achieve a considerable increase in Peak\nSignal to Noise (PSNR) ratio and Structural Similarity Index (SSIM) value.\n", "date": "2022-12-12", "categories": []}
{"id": "2312.06850", "title": "NDELS: A Novel Approach for Nighttime Dehazing, Low-Light Enhancement,\n  and Light Suppression", "abstract": "  This paper tackles the intricate challenge of improving the quality of\nnighttime images under hazy and low-light conditions. Overcoming issues\nincluding nonuniform illumination glows, texture blurring, glow effects, color\ndistortion, noise disturbance, and overall, low light have proven daunting.\nDespite the inherent difficulties, this paper introduces a pioneering solution\nnamed Nighttime Dehazing, Low-Light Enhancement, and Light Suppression (NDELS).\nNDELS utilizes a unique network that combines three essential processes to\nenhance visibility, brighten low-light regions, and effectively suppress glare\nfrom bright light sources. In contrast to limited progress in nighttime\ndehazing, unlike its daytime counterpart, NDELS presents a comprehensive and\ninnovative approach. The efficacy of NDELS is rigorously validated through\nextensive comparisons with eight state-of-the-art algorithms across four\ndiverse datasets. Experimental results showcase the superior performance of our\nmethod, demonstrating its outperformance in terms of overall image quality,\nincluding color and edge enhancement. Quantitative (PSNR, SSIM) and qualitative\nmetrics (CLIPIQA, MANIQA, TRES), measure these results.\n", "date": "2023-12-11", "categories": []}
{"id": "2312.07102", "title": "Calibration-free quantitative phase imaging in multi-core fiber\n  endoscopes using end-to-end deep learning", "abstract": "  Quantitative phase imaging (QPI) through multi-core fibers (MCFs) has been an\nemerging in vivo label-free endoscopic imaging modality with minimal\ninvasiveness. However, the computational demands of conventional iterative\nphase retrieval algorithms have limited their real-time imaging potential. We\ndemonstrate a learning-based MCF phase imaging method, that significantly\nreduced the phase reconstruction time to 5.5 ms, enabling video-rate imaging at\n181 fps. Moreover, we introduce an innovative optical system that automatically\ngenerated the first open-source dataset tailored for MCF phase imaging,\ncomprising 50,176 paired speckle and phase images. Our trained deep neural\nnetwork (DNN) demonstrates robust phase reconstruction performance in\nexperiments with a mean fidelity of up to 99.8\\%. Such an efficient fiber phase\nimaging approach can broaden the applications of QPI in hard-to-reach areas.\n", "date": "2023-12-12", "categories": []}
{"id": "2312.05707", "title": "Non-Cartesian Self-Supervised Physics-Driven Deep Learning\n  Reconstruction for Highly-Accelerated Multi-Echo Spiral fMRI", "abstract": "  Functional MRI (fMRI) is an important tool for non-invasive studies of brain\nfunction. Over the past decade, multi-echo fMRI methods that sample multiple\necho times has become popular with potential to improve quantification. While\nthese acquisitions are typically performed with Cartesian trajectories,\nnon-Cartesian trajectories, in particular spiral acquisitions, hold promise for\ndenser sampling of echo times. However, such acquisitions require very high\nacceleration rates for sufficient spatiotemporal resolutions. In this work, we\npropose to use a physics-driven deep learning (PD-DL) reconstruction to\naccelerate multi-echo spiral fMRI by 10-fold. We modify a self-supervised\nlearning algorithm for optimized training with non-Cartesian trajectories and\nuse it to train the PD-DL network. Results show that the proposed\nself-supervised PD-DL reconstruction achieves high spatio-temporal resolution\nwith meaningful BOLD analysis.\n", "date": "2023-12-09", "categories": []}
{"id": "2312.04679", "title": "ConVRT: Consistent Video Restoration Through Turbulence with Test-time\n  Optimization of Neural Video Representations", "abstract": "  tmospheric turbulence presents a significant challenge in long-range imaging.\nCurrent restoration algorithms often struggle with temporal inconsistency, as\nwell as limited generalization ability across varying turbulence levels and\nscene content different than the training data. To tackle these issues, we\nintroduce a self-supervised method, Consistent Video Restoration through\nTurbulence (ConVRT) a test-time optimization method featuring a neural video\nrepresentation designed to enhance temporal consistency in restoration. A key\ninnovation of ConVRT is the integration of a pretrained vision-language model\n(CLIP) for semantic-oriented supervision, which steers the restoration towards\nsharp, photorealistic images in the CLIP latent space. We further develop a\nprincipled selection strategy of text prompts, based on their statistical\ncorrelation with a perceptual metric. ConVRT's test-time optimization allows it\nto adapt to a wide range of real-world turbulence conditions, effectively\nleveraging the insights gained from pre-trained models on simulated data.\nConVRT offers a comprehensive and effective solution for mitigating real-world\nturbulence in dynamic videos.\n", "date": "2023-12-07", "categories": []}
{"id": "2312.05006", "title": "Decoupling Degradation and Content Processing for Adverse Weather Image\n  Restoration", "abstract": "  Adverse weather image restoration strives to recover clear images from those\naffected by various weather types, such as rain, haze, and snow. Each weather\ntype calls for a tailored degradation removal approach due to its unique impact\non images. Conversely, content reconstruction can employ a uniform approach, as\nthe underlying image content remains consistent. Although previous techniques\ncan handle multiple weather types within a single network, they neglect the\ncrucial distinction between these two processes, limiting the quality of\nrestored images. This work introduces a novel adverse weather image restoration\nmethod, called DDCNet, which decouples the degradation removal and content\nreconstruction process at the feature level based on their channel statistics.\nSpecifically, we exploit the unique advantages of the Fourier transform in both\nthese two processes: (1) the degradation information is mainly located in the\namplitude component of the Fourier domain, and (2) the Fourier domain contains\nglobal information. The former facilitates channel-dependent degradation\nremoval operation, allowing the network to tailor responses to various adverse\nweather types; the latter, by integrating Fourier's global properties into\nchannel-independent content features, enhances network capacity for consistent\nglobal content reconstruction. We further augment the degradation removal\nprocess with a degradation mapping loss function. Extensive experiments\ndemonstrate our method achieves state-of-the-art performance in multiple\nadverse weather removal benchmarks.\n", "date": "2023-12-08", "categories": []}
{"id": "2312.04327", "title": "Learning to sample in Cartesian MRI", "abstract": "  Despite its exceptional soft tissue contrast, Magnetic Resonance Imaging\n(MRI) faces the challenge of long scanning times compared to other modalities\nlike X-ray radiography. Shortening scanning times is crucial in clinical\nsettings, as it increases patient comfort, decreases examination costs and\nimproves throughput. Recent advances in compressed sensing (CS) and deep\nlearning allow accelerated MRI acquisition by reconstructing high-quality\nimages from undersampled data. While reconstruction algorithms have received\nmost of the focus, designing acquisition trajectories to optimize\nreconstruction quality remains an open question. This thesis explores two\napproaches to address this gap in the context of Cartesian MRI. First, we\npropose two algorithms, lazy LBCS and stochastic LBCS, that significantly\nimprove upon G\\\"ozc\\\"u et al.'s greedy learning-based CS (LBCS) approach. These\nalgorithms scale to large, clinically relevant scenarios like multi-coil 3D MR\nand dynamic MRI, previously inaccessible to LBCS. Additionally, we demonstrate\nthat generative adversarial networks (GANs) can serve as a natural criterion\nfor adaptive sampling by leveraging variance in the measurement domain to guide\nacquisition. Second, we delve into the underlying structures or assumptions\nthat enable mask design algorithms to perform well in practice. Our experiments\nreveal that state-of-the-art deep reinforcement learning (RL) approaches, while\ncapable of adaptation and long-horizon planning, offer only marginal\nimprovements over stochastic LBCS, which is neither adaptive nor does long-term\nplanning. Altogether, our findings suggest that stochastic LBCS and similar\nmethods represent promising alternatives to deep RL. They shine in particular\nby their scalability and computational efficiency and could be key in the\ndeployment of optimized acquisition trajectories in Cartesian MRI.\n", "date": "2023-12-07", "categories": []}
{"id": "2309.04437", "title": "Single View Refractive Index Tomography with Neural Fields", "abstract": "  Refractive Index Tomography is the inverse problem of reconstructing the\ncontinuously-varying 3D refractive index in a scene using 2D projected image\nmeasurements. Although a purely refractive field is not directly visible, it\nbends light rays as they travel through space, thus providing a signal for\nreconstruction. The effects of such fields appear in many scientific computer\nvision settings, ranging from refraction due to transparent cells in microscopy\nto the lensing of distant galaxies caused by dark matter in astrophysics.\nReconstructing these fields is particularly difficult due to the complex\nnonlinear effects of the refractive field on observed images. Furthermore,\nwhile standard 3D reconstruction and tomography settings typically have access\nto observations of the scene from many viewpoints, many refractive index\ntomography problem settings only have access to images observed from a single\nviewpoint. We introduce a method that leverages prior knowledge of light\nsources scattered throughout the refractive medium to help disambiguate the\nsingle-view refractive index tomography problem. We differentiably trace curved\nrays through a neural field representation of the refractive field, and\noptimize its parameters to best reproduce the observed image. We demonstrate\nthe efficacy of our approach by reconstructing simulated refractive fields,\nanalyze the effects of light source distribution on the recovered field, and\ntest our method on a simulated dark matter mapping problem where we\nsuccessfully recover the 3D refractive field caused by a realistic dark matter\ndistribution.\n", "date": "2023-09-08", "categories": []}
{"id": "2306.07581", "title": "Binary Radiance Fields", "abstract": "  In this paper, we propose \\textit{binary radiance fields} (BiRF), a\nstorage-efficient radiance field representation employing binary feature\nencoding that encodes local features using binary encoding parameters in a\nformat of either $+1$ or $-1$. This binarization strategy lets us represent the\nfeature grid with highly compact feature encoding and a dramatic reduction in\nstorage size. Furthermore, our 2D-3D hybrid feature grid design enhances the\ncompactness of feature encoding as the 3D grid includes main components while\n2D grids capture details. In our experiments, binary radiance field\nrepresentation successfully outperforms the reconstruction performance of\nstate-of-the-art (SOTA) efficient radiance field models with lower storage\nallocation. In particular, our model achieves impressive results in static\nscene reconstruction, with a PSNR of 32.03 dB for Synthetic-NeRF scenes, 34.48\ndB for Synthetic-NSVF scenes, 28.20 dB for Tanks and Temples scenes while only\nutilizing 0.5 MB of storage space, respectively. We hope the proposed binary\nradiance field representation will make radiance fields more accessible without\na storage bottleneck.\n", "date": "2023-06-13", "categories": []}
{"id": "2309.09039", "title": "Microscale 3-D Capacitance Tomography with a CMOS Sensor Array", "abstract": "  Electrical capacitance tomography (ECT) is a nonoptical imaging technique in\nwhich a map of the interior permittivity of a volume is estimated by making\ncapacitance measurements at its boundary and solving an inverse problem. While\nprevious ECT demonstrations have often been at centimeter scales, ECT is not\nlimited to macroscopic systems. In this paper, we demonstrate ECT imaging of\npolymer microspheres and bacterial biofilms using a CMOS microelectrode array,\nachieving spatial resolution of 10 microns. Additionally, we propose a deep\nlearning architecture and an improved multi-objective training scheme for\nreconstructing out-of-plane permittivity maps from the sensor measurements.\nExperimental results show that the proposed approach is able to resolve\nmicroscopic 3-D structures, achieving 91.5% prediction accuracy on the\nmicrosphere dataset and 82.7% on the biofilm dataset, including an average of\n4.6% improvement over baseline computational methods.\n", "date": "2023-09-16", "categories": []}
{"id": "2312.00852", "title": "Beyond First-Order Tweedie: Solving Inverse Problems using Latent\n  Diffusion", "abstract": "  Sampling from the posterior distribution poses a major computational\nchallenge in solving inverse problems using latent diffusion models. Common\nmethods rely on Tweedie's first-order moments, which are known to induce a\nquality-limiting bias. Existing second-order approximations are impractical due\nto prohibitive computational costs, making standard reverse diffusion processes\nintractable for posterior sampling. This paper introduces Second-order Tweedie\nsampler from Surrogate Loss (STSL), a novel sampler that offers efficiency\ncomparable to first-order Tweedie with a tractable reverse process using\nsecond-order approximation. Our theoretical results reveal that the\nsecond-order approximation is lower bounded by our surrogate loss that only\nrequires $O(1)$ compute using the trace of the Hessian, and by the lower bound\nwe derive a new drift term to make the reverse process tractable. Our method\nsurpasses SoTA solvers PSLD and P2L, achieving 4X and 8X reduction in neural\nfunction evaluations, respectively, while notably enhancing sampling quality on\nFFHQ, ImageNet, and COCO benchmarks. In addition, we show STSL extends to\ntext-guided image editing and addresses residual distortions present from\ncorrupted images in leading text-guided image editing methods. To our best\nknowledge, this is the first work to offer an efficient second-order\napproximation in solving inverse problems using latent diffusion and editing\nreal-world images with corruptions.\n", "date": "2023-12-01", "categories": []}
{"id": "2312.01638", "title": "J-Net: Improved U-Net for Terahertz Image Super-Resolution", "abstract": "  Terahertz (THz) waves are electromagnetic waves in the 0.1 to 10 THz\nfrequency range, and THz imaging is utilized in a range of applications,\nincluding security inspections, biomedical fields, and the non-destructive\nexamination of materials. However, THz images have low resolution due to the\nlong wavelength of THz waves. Therefore, improving the resolution of THz images\nis one of the current hot research topics. We propose a novel network\narchitecture called J-Net which is improved version of U-Net to solve the THz\nimage super-resolution. It employs the simple baseline blocks which can extract\nlow resolution (LR) image features and learn the mapping of LR images to\nhighresolution (HR) images efficiently. All training was conducted using the\nDIV2K+Flickr2K dataset, and we employed the peak signal-to-noise ratio (PSNR)\nfor quantitative comparison. In our comparisons with other THz image\nsuper-resolution methods, JNet achieved a PSNR of 32.52 dB, surpassing other\ntechniques by more than 1 dB. J-Net also demonstrates superior performance on\nreal THz images compared to other methods. Experiments show that the proposed\nJ-Net achieves better PSNR and visual improvement compared with other THz image\nsuper-resolution methods.\n", "date": "2023-12-04", "categories": []}
{"id": "2312.01761", "title": "Light Field Imaging in the Restrictive Object Space based on Flexible\n  Angular Plane", "abstract": "  In some applications, the object space of light field imaging system is\nrestrictive, such as industrial and medical endoscopes. If the traditional\nlight field imaging system is used in the restrictive object space (ROS)\ndirectly but without any specific considerations, the ROS will lead to severe\nmicrolens image distortions and then affects light field decoding, calibration\nand 3D reconstruction. The light field imaging in restrictive object space\n(ROS-LF) is complicated but significant. In this paper, we first deduce that\nthe reason of the microlens image deviation is the position variation of the\nangular plane, then we propose the flexible angular plane for ROS-LF, while in\nthe traditional light field the angular plane always coincides with the main\nlens plane. Subsequently, we propose the microlens image non-distortion\nprinciple for ROS-LF and introduce the ROS-LF imaging principle. We demonstrate\nthat the difference is an aperture constant term between the ROS-LF and\ntraditional light field imaging models. At last, we design a ROS-LF simulated\nsystem and calibrate it to verify principles proposed in this paper.\n", "date": "2023-12-04", "categories": []}
{"id": "2302.11259", "title": "Transfer Learning Enhanced Full Waveform Inversion", "abstract": "  We propose a way to favorably employ neural networks in the field of\nnon-destructive testing using Full Waveform Inversion (FWI). The presented\nmethodology discretizes the unknown material distribution in the domain with a\nneural network within an adjoint optimization. To further increase efficiency\nof the FWI, pretrained neural networks are used to provide a good starting\npoint for the inversion. This reduces the number of iterations in the Full\nWaveform Inversion for specific, yet generalizable settings.\n", "date": "2023-02-22", "categories": []}
{"id": "2312.00252", "title": "PyNeRF: Pyramidal Neural Radiance Fields", "abstract": "  Neural Radiance Fields (NeRFs) can be dramatically accelerated by spatial\ngrid representations. However, they do not explicitly reason about scale and so\nintroduce aliasing artifacts when reconstructing scenes captured at different\ncamera distances. Mip-NeRF and its extensions propose scale-aware renderers\nthat project volumetric frustums rather than point samples but such approaches\nrely on positional encodings that are not readily compatible with grid methods.\nWe propose a simple modification to grid-based models by training model heads\nat different spatial grid resolutions. At render time, we simply use coarser\ngrids to render samples that cover larger volumes. Our method can be easily\napplied to existing accelerated NeRF methods and significantly improves\nrendering quality (reducing error rates by 20-90% across synthetic and\nunbounded real-world scenes) while incurring minimal performance overhead (as\neach model head is quick to evaluate). Compared to Mip-NeRF, we reduce error\nrates by 20% while training over 60x faster.\n", "date": "2023-11-30", "categories": []}
{"id": "2312.00387", "title": "Partition-based K-space Synthesis for Multi-contrast Parallel Imaging", "abstract": "  Multi-contrast magnetic resonance imaging is a significant and essential\nmedical imaging technique.However, multi-contrast imaging has longer\nacquisition time and is easy to cause motion artifacts. In particular, the\nacquisition time for a T2-weighted image is prolonged due to its longer\nrepetition time (TR). On the contrary, T1-weighted image has a shorter TR.\nTherefore,utilizing complementary information across T1 and T2-weighted image\nis a way to decrease the overall imaging time. Previous T1-assisted T2\nreconstruction methods have mostly focused on image domain using whole-based\nimage fusion approaches. The image domain reconstruction method has the defects\nof high computational complexity and limited flexibility. To address this\nissue, we propose a novel multi-contrast imaging method called partition-based\nk-space synthesis (PKS) which can achieve super reconstruction quality of\nT2-weighted image by feature fusion. Concretely, we first decompose\nfully-sampled T1 k-space data and under-sampled T2 k-space data into two\nsub-data, separately. Then two new objects are constructed by combining the two\nsub-T1/T2 data. After that, the two new objects as the whole data to realize\nthe reconstruction of T2-weighted image. Finally, the objective T2 is\nsynthesized by extracting the sub-T2 data of each part. Experimental results\nshowed that our combined technique can achieve comparable or better results\nthan using traditional k-space parallel imaging(SAKE) that processes each\ncontrast independently.\n", "date": "2023-12-01", "categories": []}
{"id": "2311.18012", "title": "Bayesian Imaging for Radio Interferometry with Score-Based Priors", "abstract": "  The inverse imaging task in radio interferometry is a key limiting factor to\nretrieving Bayesian uncertainties in radio astronomy in a computationally\neffective manner. We use a score-based prior derived from optical images of\ngalaxies to recover images of protoplanetary disks from the DSHARP survey. We\ndemonstrate that our method produces plausible posterior samples despite the\nmisspecified galaxy prior. We show that our approach produces results which are\ncompetitive with existing radio interferometry imaging algorithms.\n", "date": "2023-11-29", "categories": []}
{"id": "2311.18710", "title": "Meta-Prior: Meta learning for Adaptive Inverse Problem Solvers", "abstract": "  Deep neural networks have become a foundational tool for addressing imaging\ninverse problems. They are typically trained for a specific task, with a\nsupervised loss to learn a mapping from the observations to the image to\nrecover. However, real-world imaging challenges often lack ground truth data,\nrendering traditional supervised approaches ineffective. Moreover, for each new\nimaging task, a new model needs to be trained from scratch, wasting time and\nresources. To overcome these limitations, we introduce a novel approach based\non meta-learning. Our method trains a meta-model on a diverse set of imaging\ntasks that allows the model to be efficiently fine-tuned for specific tasks\nwith few fine-tuning steps. We show that the proposed method extends to the\nunsupervised setting, where no ground truth data is available. In its bilevel\nformulation, the outer level uses a supervised loss, that evaluates how well\nthe fine-tuned model performs, while the inner loss can be either supervised or\nunsupervised, relying only on the measurement operator. This allows the\nmeta-model to leverage a few ground truth samples for each task while being\nable to generalize to new imaging tasks. We show that in simple settings, this\napproach recovers the Bayes optimal estimator, illustrating the soundness of\nour approach. We also demonstrate our method's effectiveness on various tasks,\nincluding image processing and magnetic resonance imaging.\n", "date": "2023-11-30", "categories": []}
{"id": "2311.18810", "title": "Convergence of Nonconvex PnP-ADMM with MMSE Denoisers", "abstract": "  Plug-and-Play Alternating Direction Method of Multipliers (PnP-ADMM) is a\nwidely-used algorithm for solving inverse problems by integrating physical\nmeasurement models and convolutional neural network (CNN) priors. PnP-ADMM has\nbeen theoretically proven to converge for convex data-fidelity terms and\nnonexpansive CNNs. It has however been observed that PnP-ADMM often empirically\nconverges even for expansive CNNs. This paper presents a theoretical\nexplanation for the observed stability of PnP-ADMM based on the interpretation\nof the CNN prior as a minimum mean-squared error (MMSE) denoiser. Our\nexplanation parallels a similar argument recently made for the iterative\nshrinkage/thresholding algorithm variant of PnP (PnP-ISTA) and relies on the\nconnection between MMSE denoisers and proximal operators. We also numerically\nevaluate the performance gap between PnP-ADMM using a nonexpansive DnCNN\ndenoiser and expansive DRUNet denoiser, thus motivating the use of expansive\nCNNs.\n", "date": "2023-11-30", "categories": []}
{"id": "2305.11120", "title": "A Compound Gaussian Least Squares Algorithm and Unrolled Network for\n  Linear Inverse Problems", "abstract": "  For solving linear inverse problems, particularly of the type that appears in\ntomographic imaging and compressive sensing, this paper develops two new\napproaches. The first approach is an iterative algorithm that minimizes a\nregularized least squares objective function where the regularization is based\non a compound Gaussian prior distribution. The compound Gaussian prior subsumes\nmany of the commonly used priors in image reconstruction, including those of\nsparsity-based approaches. The developed iterative algorithm gives rise to the\npaper's second new approach, which is a deep neural network that corresponds to\nan \"unrolling\" or \"unfolding\" of the iterative algorithm. Unrolled deep neural\nnetworks have interpretable layers and outperform standard deep learning\nmethods. This paper includes a detailed computational theory that provides\ninsight into the construction and performance of both algorithms. The\nconclusion is that both algorithms outperform other state-of-the-art approaches\nto tomographic image formation and compressive sensing, especially in the\ndifficult regime of low training.\n", "date": "2023-05-18", "categories": []}
{"id": "2308.03818", "title": "A sparse coding approach to inverse problems with application to\n  microwave tomography", "abstract": "  Inverse imaging problems that are ill-posed can be encountered across\nmultiple domains of science and technology, ranging from medical diagnosis to\nastronomical studies. To reconstruct images from incomplete and distorted data,\nit is necessary to create algorithms that can take into account both, the\nphysical mechanisms responsible for generating these measurements and the\nintrinsic characteristics of the images being analyzed. In this work, the\nsparse representation of images is reviewed, which is a realistic, compact and\neffective generative model for natural images inspired by the visual system of\nmammals. It enables us to address ill-posed linear inverse problems by training\nthe model on a vast collection of images. Moreover, we extend the application\nof sparse coding to solve the non-linear and ill-posed problem in microwave\ntomography imaging, which could lead to a significant improvement of the\nstate-of-the-arts algorithms.\n", "date": "2023-08-07", "categories": []}
{"id": "2311.13187", "title": "NeISF: Neural Incident Stokes Field for Geometry and Material Estimation", "abstract": "  Multi-view inverse rendering is the problem of estimating the scene\nparameters such as shapes, materials, or illuminations from a sequence of\nimages captured under different viewpoints. Many approaches, however, assume\nsingle light bounce and thus fail to recover challenging scenarios like\ninter-reflections. On the other hand, simply extending those methods to\nconsider multi-bounced light requires more assumptions to alleviate the\nambiguity. To address this problem, we propose Neural Incident Stokes Fields\n(NeISF), a multi-view inverse rendering framework that reduces ambiguities\nusing polarization cues. The primary motivation for using polarization cues is\nthat it is the accumulation of multi-bounced light, providing rich information\nabout geometry and material. Based on this knowledge, the proposed incident\nStokes field efficiently models the accumulated polarization effect with the\naid of an original physically-based differentiable polarimetric renderer.\nLastly, experimental results show that our method outperforms the existing\nworks in synthetic and real scenarios.\n", "date": "2023-11-22", "categories": []}
{"id": "2311.17251", "title": "SubZero: Subspace Zero-Shot MRI Reconstruction", "abstract": "  Recently introduced zero-shot self-supervised learning (ZS-SSL) has shown\npotential in accelerated MRI in a scan-specific scenario, which enabled\nhigh-quality reconstructions without access to a large training dataset. ZS-SSL\nhas been further combined with the subspace model to accelerate 2D T2-shuffling\nacquisitions. In this work, we propose a parallel network framework and\nintroduce an attention mechanism to improve subspace-based zero-shot\nself-supervised learning and enable higher acceleration factors. We name our\nmethod SubZero and demonstrate that it can achieve improved performance\ncompared with current methods in T1 and T2 mapping acquisitions.\n", "date": "2023-11-28", "categories": []}
{"id": "2311.17657", "title": "Volumetric Cloud Field Reconstruction", "abstract": "  Volumetric phenomena, such as clouds and fog, present a significant challenge\nfor 3D reconstruction systems due to their translucent nature and their complex\ninteractions with light. Conventional techniques for reconstructing scattering\nvolumes rely on controlled setups, limiting practical applications. This paper\nintroduces an approach to reconstructing volumes from a few input stereo pairs.\nWe propose a novel deep learning framework that integrates a deep stereo model\nwith a 3D Convolutional Neural Network (3D CNN) and an advection module,\ncapable of capturing the shape and dynamics of volumes. The stereo depths are\nused to carve empty space around volumes, providing the 3D CNN with a prior for\ncoping with the lack of input views. Refining our output, the advection module\nleverages the temporal evolution of the medium, providing a mechanism to infer\nmotion and improve temporal consistency. The efficacy of our system is\ndemonstrated through its ability to estimate density and velocity fields of\nlarge-scale volumes, in this case, clouds, from a sparse set of stereo image\npairs.\n", "date": "2023-11-29", "categories": []}
{"id": "2308.14610", "title": "PolarRec: Radio Interferometric Data Reconstruction with Polar\n  Coordinate Representation", "abstract": "  In radio astronomy, visibility data, which are measurements of wave signals\nfrom radio telescopes, are transformed into images for observation of distant\ncelestial objects. However, these resultant images usually contain both real\nsources and artifacts, due to signal sparsity and other factors. One way to\nobtain cleaner images is to reconstruct samples into dense forms before\nimaging. Unfortunately, existing reconstruction methods often miss some\ncomponents of visibility in frequency domain, so blurred object edges and\npersistent artifacts remain in the images. Furthermore, the computation\noverhead is high on irregular visibility samples due to the data skew. To\naddress these problems, we propose PolarRec, a transformer-encoder-conditioned\nreconstruction pipeline with visibility samples converted into the polar\ncoordinate representation. This representation matches the way in which radio\ntelescopes observe a celestial area as the Earth rotates. As a result,\nvisibility samples distribute in the polar system more uniformly than in the\nCartesian space. Therefore, we propose to use radial distance in the loss\nfunction, to help reconstruct complete visibility effectively. Also, we group\nvisibility samples by their polar angles and propose a group-based encoding\nscheme to improve the efficiency. Our experiments demonstrate that PolarRec\nmarkedly improves imaging results by faithfully reconstructing all frequency\ncomponents in the visibility domain while significantly reducing the\ncomputation cost in visibility data encoding. We believe this high-quality and\nhigh-efficiency imaging of PolarRec will better facilitate astronomers to\nconduct their research.\n", "date": "2023-08-28", "categories": []}
{"id": "2311.16652", "title": "Augmenting x-ray single particle imaging reconstruction with\n  self-supervised machine learning", "abstract": "  The development of X-ray Free Electron Lasers (XFELs) has opened numerous\nopportunities to probe atomic structure and ultrafast dynamics of various\nmaterials. Single Particle Imaging (SPI) with XFELs enables the investigation\nof biological particles in their natural physiological states with unparalleled\ntemporal resolution, while circumventing the need for cryogenic conditions or\ncrystallization. However, reconstructing real-space structures from\nreciprocal-space x-ray diffraction data is highly challenging due to the\nabsence of phase and orientation information, which is further complicated by\nweak scattering signals and considerable fluctuations in the number of photons\nper pulse. In this work, we present an end-to-end, self-supervised machine\nlearning approach to recover particle orientations and estimate reciprocal\nspace intensities from diffraction images only. Our method demonstrates great\nrobustness under demanding experimental conditions with significantly enhanced\nreconstruction capabilities compared with conventional algorithms, and\nsignifies a paradigm shift in SPI as currently practiced at XFELs.\n", "date": "2023-11-28", "categories": []}
{"id": "2311.16923", "title": "Super-Resolution through StyleGAN Regularized Latent Search: A\n  Realism-Fidelity Trade-off", "abstract": "  This paper addresses the problem of super-resolution: constructing a highly\nresolved (HR) image from a low resolved (LR) one. Recent unsupervised\napproaches search the latent space of a StyleGAN pre-trained on HR images, for\nthe image that best downscales to the input LR image. However, they tend to\nproduce out-of-domain images and fail to accurately reconstruct HR images that\nare far from the original domain. Our contribution is twofold. Firstly, we\nintroduce a new regularizer to constrain the search in the latent space,\nensuring that the inverted code lies in the original image manifold. Secondly,\nwe further enhanced the reconstruction through expanding the image prior around\nthe optimal latent code. Our results show that the proposed approach recovers\nrealistic high-quality images for large magnification factors. Furthermore, for\nlow magnification factors, it can still reconstruct details that the generator\ncould not have produced otherwise. Altogether, our approach achieves a good\ntrade-off between fidelity and realism for the super-resolution task.\n", "date": "2023-11-28", "categories": []}
{"id": "2311.15369", "title": "TD-Net: A Tri-domain network for sparse-view CT reconstruction", "abstract": "  Sparse-view CT reconstruction, aimed at reducing X-ray radiation risks,\nfrequently suffers from image quality degradation, manifested as noise and\nartifacts. Existing post-processing and dual-domain techniques, although\neffective in radiation reduction, often lead to over-smoothed results,\ncompromising diagnostic clarity. Addressing this, we introduce TD-Net, a\npioneering tri-domain approach that unifies sinogram, image, and frequency\ndomain optimizations. By incorporating Frequency Supervision Module(FSM),\nTD-Net adeptly preserves intricate details, overcoming the prevalent\nover-smoothing issue. Extensive evaluations demonstrate TD-Net's superior\nperformance in reconstructing high-quality CT images from sparse views,\nefficiently balancing radiation safety and image fidelity. The enhanced\ncapabilities of TD-Net in varied noise scenarios highlight its potential as a\nbreakthrough in medical imaging.\n", "date": "2023-11-26", "categories": []}
{"id": "2311.13626", "title": "Physics-driven generative adversarial networks empower single-pixel\n  infrared hyperspectral imaging", "abstract": "  A physics-driven generative adversarial network (GAN) was established here\nfor single-pixel hyperspectral imaging (HSI) in the infrared spectrum, to\neliminate the extensive data training work required by traditional data-driven\nmodel. Within the GAN framework, the physical process of single-pixel imaging\n(SPI) was integrated into the generator, and the actual and estimated\none-dimensional (1D) bucket signals were employed as constraints in the\nobjective function to update the network's parameters and optimize the\ngenerator with the assistance of the discriminator. In comparison to\nsingle-pixel infrared HSI methods based on compressed sensing and\nphysics-driven convolution neural networks, our physics-driven GAN-based\nsingle-pixel infrared HSI can achieve higher imaging performance but with fewer\nmeasurements. We believe that this physics-driven GAN will promote practical\napplications of computational imaging, especially various SPI-based techniques.\n", "date": "2023-11-22", "categories": []}
{"id": "2311.13779", "title": "Detection and Identification Accuracy of PCA-Accelerated Real-Time\n  Processing of Hyperspectral Imagery", "abstract": "  Real-time or near real-time hyperspectral detection and identification are\nextremely useful and needed in many fields. These data sets can be quite large,\nand the algorithms can require numerous computations that slow the process\ndown. A common way of speeding up the process is to use principal component\nanalysis (PCA) for dimension reduction. In the reduced dimensional space,\nprovided by a subset of the principal components, fewer computations are needed\nto process the data resulting in a faster run time. In this paper, we propose a\nway to further decrease the time required to use PCA by investigating how many\nprincipal components may be omitted with minimal impact on the detection rate.\nUsing ACE to perform the detection, and then probability, and spectral fit for\nidentification, we find that the number of principal components can be reduced\nby a substantial amount before seeing a noticeable change in detection rates.\n", "date": "2023-11-23", "categories": []}
{"id": "2311.13963", "title": "Investigating the use of publicly available natural videos to learn\n  Dynamic MR image reconstruction", "abstract": "  Purpose: To develop and assess a deep learning (DL) pipeline to learn dynamic\nMR image reconstruction from publicly available natural videos (Inter4K).\n  Materials and Methods: Learning was performed for a range of DL architectures\n(VarNet, 3D UNet, FastDVDNet) and corresponding sampling patterns (Cartesian,\nradial, spiral) either from true multi-coil cardiac MR data (N=692) or from\npseudo-MR data simulated from Inter4K natural videos (N=692). Real-time\nundersampled dynamic MR images were reconstructed using DL networks trained\nwith cardiac data and natural videos, and compressed sensing (CS). Differences\nwere assessed in simulations (N=104 datasets) in terms of MSE, PSNR, and SSIM\nand prospectively for cardiac (short axis, four chambers, N=20) and speech\n(N=10) data in terms of subjective image quality ranking, SNR and Edge\nsharpness. Friedman Chi Square tests with post-hoc Nemenyi analysis were\nperformed to assess statistical significance.\n  Results: For all simulation metrics, DL networks trained with cardiac data\noutperformed DL networks trained with natural videos, which outperformed CS\n(p<0.05). However, in prospective experiments DL reconstructions using both\ntraining datasets were ranked similarly (and higher than CS) and presented no\nstatistical differences in SNR and Edge Sharpness for most conditions.\nAdditionally, high SSIM was measured between the DL methods with cardiac data\nand natural videos (SSIM>0.85).\n  Conclusion: The developed pipeline enabled learning dynamic MR reconstruction\nfrom natural videos preserving DL reconstruction advantages such as high\nquality fast and ultra-fast reconstructions while overcoming some limitations\n(data scarcity or sharing). The natural video dataset, code and pre-trained\nnetworks are made readily available on github.\n  Key Words: real-time; dynamic MRI; deep learning; image reconstruction;\nmachine learning;\n", "date": "2023-11-23", "categories": []}
{"id": "2311.13177", "title": "Volumetric Reconstruction Resolves Off-Resonance Artifacts in Static and\n  Dynamic PROPELLER MRI", "abstract": "  Off-resonance artifacts in magnetic resonance imaging (MRI) are visual\ndistortions that occur when the actual resonant frequencies of spins within the\nimaging volume differ from the expected frequencies used to encode spatial\ninformation. These discrepancies can be caused by a variety of factors,\nincluding magnetic field inhomogeneities, chemical shifts, or susceptibility\ndifferences within the tissues. Such artifacts can manifest as blurring,\nghosting, or misregistration of the reconstructed image, and they often\ncompromise its diagnostic quality. We propose to resolve these artifacts by\nlifting the 2D MRI reconstruction problem to 3D, introducing an additional\n\"spectral\" dimension to model this off-resonance. Our approach is inspired by\nrecent progress in modeling radiance fields, and is capable of reconstructing\nboth static and dynamic MR images as well as separating fat and water, which is\nof independent clinical interest. We demonstrate our approach in the context of\nPROPELLER (Periodically Rotated Overlapping ParallEL Lines with Enhanced\nReconstruction) MRI acquisitions, which are popular for their robustness to\nmotion artifacts. Our method operates in a few minutes on a single GPU, and to\nour knowledge is the first to correct for chemical shift in gradient echo\nPROPELLER MRI reconstruction without additional measurements or pretraining\ndata.\n", "date": "2023-11-22", "categories": []}
{"id": "2309.04342", "title": "Revealing the preference for correcting separated aberrations in joint\n  optic-image design", "abstract": "  The joint design of the optical system and the downstream algorithm is a\nchallenging and promising task. Due to the demand for balancing the global\noptimal of imaging systems and the computational cost of physical simulation,\nexisting methods cannot achieve efficient joint design of complex systems such\nas smartphones and drones. In this work, starting from the perspective of the\noptical design, we characterize the optics with separated aberrations.\nAdditionally, to bridge the hardware and software without gradients, an image\nsimulation system is presented to reproduce the genuine imaging procedure of\nlenses with large field-of-views. As for aberration correction, we propose a\nnetwork to perceive and correct the spatially varying aberrations and validate\nits superiority over state-of-the-art methods. Comprehensive experiments reveal\nthat the preference for correcting separated aberrations in joint design is as\nfollows: longitudinal chromatic aberration, lateral chromatic aberration,\nspherical aberration, field curvature, and coma, with astigmatism coming last.\nDrawing from the preference, a 10% reduction in the total track length of the\nconsumer-level mobile phone lens module is accomplished. Moreover, this\nprocedure spares more space for manufacturing deviations, realizing\nextreme-quality enhancement of computational photography. The optimization\nparadigm provides innovative insight into the practical joint design of\nsophisticated optical systems and post-processing algorithms.\n", "date": "2023-09-08", "categories": []}
{"id": "2311.12792", "title": "Intrinsic Image Decomposition via Ordinal Shading", "abstract": "  Intrinsic decomposition is a fundamental mid-level vision problem that plays\na crucial role in various inverse rendering and computational photography\npipelines. Generating highly accurate intrinsic decompositions is an inherently\nunder-constrained task that requires precisely estimating continuous-valued\nshading and albedo. In this work, we achieve high-resolution intrinsic\ndecomposition by breaking the problem into two parts. First, we present a dense\nordinal shading formulation using a shift- and scale-invariant loss in order to\nestimate ordinal shading cues without restricting the predictions to obey the\nintrinsic model. We then combine low- and high-resolution ordinal estimations\nusing a second network to generate a shading estimate with both global\ncoherency and local details. We encourage the model to learn an accurate\ndecomposition by computing losses on the estimated shading as well as the\nalbedo implied by the intrinsic model. We develop a straightforward method for\ngenerating dense pseudo ground truth using our model's predictions and\nmulti-illumination data, enabling generalization to in-the-wild imagery. We\npresent an exhaustive qualitative and quantitative analysis of our predicted\nintrinsic components against state-of-the-art methods. Finally, we demonstrate\nthe real-world applicability of our estimations by performing otherwise\ndifficult editing tasks such as recoloring and relighting.\n", "date": "2023-11-21", "categories": []}
{"id": "2305.10925", "title": "Unsupervised Hyperspectral Pansharpening via Low-rank Diffusion Model", "abstract": "  Hyperspectral pansharpening is a process of merging a high-resolution\npanchromatic (PAN) image and a low-resolution hyperspectral (LRHS) image to\ncreate a single high-resolution hyperspectral (HRHS) image. Existing\nBayesian-based HS pansharpening methods require designing handcraft image prior\nto characterize the image features, and deep learning-based HS pansharpening\nmethods usually require a large number of paired training data and suffer from\npoor generalization ability. To address these issues, in this work, we propose\na low-rank diffusion model for hyperspectral pansharpening by simultaneously\nleveraging the power of the pre-trained deep diffusion model and better\ngeneralization ability of Bayesian methods. Specifically, we assume that the\nHRHS image can be recovered from the product of two low-rank tensors, i.e., the\nbase tensor and the coefficient matrix. The base tensor lies on the image field\nand has a low spectral dimension. Thus, we can conveniently utilize a\npre-trained remote sensing diffusion model to capture its image structures.\nAdditionally, we derive a simple yet quite effective way to pre-estimate the\ncoefficient matrix from the observed LRHS image, which preserves the spectral\ninformation of the HRHS. Experimental results demonstrate that the proposed\nmethod performs better than some popular traditional approaches and gains\nbetter generalization ability than some DL-based methods. The code is released\nin https://github.com/xyrui/PLRDiff.\n", "date": "2023-05-18", "categories": []}
{"id": "2311.10950", "title": "Single-shot Phase Retrieval from a Fractional Fourier Transform\n  Perspective", "abstract": "  The realm of classical phase retrieval concerns itself with the arduous task\nof recovering a signal from its Fourier magnitude measurements, which are\nfraught with inherent ambiguities. A single-exposure intensity measurement is\ncommonly deemed insufficient for the reconstruction of the primal signal, given\nthat the absent phase component is imperative for the inverse transformation.\nIn this work, we present a novel single-shot phase retrieval paradigm from a\nfractional Fourier transform (FrFT) perspective, which involves integrating the\nFrFT-based physical measurement model within a self-supervised reconstruction\nscheme. Specifically, the proposed FrFT-based measurement model addresses the\naliasing artifacts problem in the numerical calculation of Fresnel diffraction,\nfeaturing adaptability to both short-distance and long-distance propagation\nscenarios. Moreover, the intensity measurement in the FrFT domain proves highly\neffective in alleviating the ambiguities of phase retrieval and relaxing the\nprevious conditions on oversampled or multiple measurements in the Fourier\ndomain. Furthermore, the proposed self-supervised reconstruction approach\nharnesses the fast discrete algorithm of FrFT alongside untrained neural\nnetwork priors, thereby attaining preeminent results. Through numerical\nsimulations, we demonstrate that both amplitude and phase objects can be\neffectively retrieved from a single-shot intensity measurement using the\nproposed approach and provide a promising technique for support-free coherent\ndiffraction imaging.\n", "date": "2023-11-18", "categories": []}
{"id": "2311.11417", "title": "DiffSCI: Zero-Shot Snapshot Compressive Imaging via Iterative Spectral\n  Diffusion Model", "abstract": "  This paper endeavors to advance the precision of snapshot compressive imaging\n(SCI) reconstruction for multispectral image (MSI). To achieve this, we\nintegrate the advantageous attributes of established SCI techniques and an\nimage generative model, propose a novel structured zero-shot diffusion model,\ndubbed DiffSCI. DiffSCI leverages the structural insights from the deep prior\nand optimization-based methodologies, complemented by the generative\ncapabilities offered by the contemporary denoising diffusion model.\nSpecifically, firstly, we employ a pre-trained diffusion model, which has been\ntrained on a substantial corpus of RGB images, as the generative denoiser\nwithin the Plug-and-Play framework for the first time. This integration allows\nfor the successful completion of SCI reconstruction, especially in the case\nthat current methods struggle to address effectively. Secondly, we\nsystematically account for spectral band correlations and introduce a robust\nmethodology to mitigate wavelength mismatch, thus enabling seamless adaptation\nof the RGB diffusion model to MSIs. Thirdly, an accelerated algorithm is\nimplemented to expedite the resolution of the data subproblem. This\naugmentation not only accelerates the convergence rate but also elevates the\nquality of the reconstruction process. We present extensive testing to show\nthat DiffSCI exhibits discernible performance enhancements over prevailing\nself-supervised and zero-shot approaches, surpassing even supervised\ntransformer counterparts across both simulated and real datasets. Our code will\nbe available.\n", "date": "2023-11-19", "categories": []}
{"id": "2311.05146", "title": "OW-SLR: Overlapping Windows on Semi-Local Region for Image\n  Super-Resolution", "abstract": "  There has been considerable progress in implicit neural representation to\nupscale an image to any arbitrary resolution. However, existing methods are\nbased on defining a function to predict the Red, Green and Blue (RGB) value\nfrom just four specific loci. Relying on just four loci is insufficient as it\nleads to losing fine details from the neighboring region(s). We show that by\ntaking into account the semi-local region leads to an improvement in\nperformance. In this paper, we propose applying a new technique called\nOverlapping Windows on Semi-Local Region (OW-SLR) to an image to obtain any\narbitrary resolution by taking the coordinates of the semi-local region around\na point in the latent space. This extracted detail is used to predict the RGB\nvalue of a point. We illustrate the technique by applying the algorithm to the\nOptical Coherence Tomography-Angiography (OCT-A) images and show that it can\nupscale them to random resolution. This technique outperforms the existing\nstate-of-the-art methods when applied to the OCT500 dataset. OW-SLR provides\nbetter results for classifying healthy and diseased retinal images such as\ndiabetic retinopathy and normals from the given set of OCT-A images. The\nproject page is available at https://rishavbb.github.io/ow-slr/index.html\n", "date": "2023-11-09", "categories": []}
{"id": "2311.09223", "title": "Non-line-of-sight imaging in the presence of scattering media using\n  phasor fields", "abstract": "  Non-line-of-sight (NLOS) imaging aims to reconstruct partially or completely\noccluded scenes. Recent approaches have demonstrated high-quality\nreconstructions of complex scenes with arbitrary reflectance, occlusions, and\nsignificant multi-path effects. However, previous works focused on surface\nscattering only, which reduces the generality in more challenging scenarios\nsuch as scenes submerged in scattering media. In this work, we investigate\ncurrent state-of-the-art NLOS imaging methods based on phasor fields to\nreconstruct scenes submerged in scattering media. We empirically analyze the\ncapability of phasor fields in reconstructing complex synthetic scenes\nsubmerged in thick scattering media. We also apply the method to real scenes,\nshowing that it performs similarly to recent diffuse optical tomography\nmethods.\n", "date": "2023-08-25", "categories": []}
{"id": "2311.09639", "title": "On the Quantification of Image Reconstruction Uncertainty without\n  Training Data", "abstract": "  Computational imaging plays a pivotal role in determining hidden information\nfrom sparse measurements. A robust inverse solver is crucial to fully\ncharacterize the uncertainty induced by these measurements, as it allows for\nthe estimation of the complete posterior of unrecoverable targets. This, in\nturn, facilitates a probabilistic interpretation of observational data for\ndecision-making. In this study, we propose a deep variational framework that\nleverages a deep generative model to learn an approximate posterior\ndistribution to effectively quantify image reconstruction uncertainty without\nthe need for training data. We parameterize the target posterior using a\nflow-based model and minimize their Kullback-Leibler (KL) divergence to achieve\naccurate uncertainty estimation. To bolster stability, we introduce a robust\nflow-based model with bi-directional regularization and enhance expressivity\nthrough gradient boosting. Additionally, we incorporate a space-filling design\nto achieve substantial variance reduction on both latent prior space and target\nposterior space. We validate our method on several benchmark tasks and two\nreal-world applications, namely fastMRI and black hole image reconstruction.\nOur results indicate that our method provides reliable and high-quality image\nreconstruction with robust uncertainty estimation.\n", "date": "2023-11-16", "categories": []}
{"id": "2211.06891", "title": "Residual Degradation Learning Unfolding Framework with Mixing Priors\n  across Spectral and Spatial for Compressive Spectral Imaging", "abstract": "  To acquire a snapshot spectral image, coded aperture snapshot spectral\nimaging (CASSI) is proposed. A core problem of the CASSI system is to recover\nthe reliable and fine underlying 3D spectral cube from the 2D measurement. By\nalternately solving a data subproblem and a prior subproblem, deep unfolding\nmethods achieve good performance. However, in the data subproblem, the used\nsensing matrix is ill-suited for the real degradation process due to the device\nerrors caused by phase aberration, distortion; in the prior subproblem, it is\nimportant to design a suitable model to jointly exploit both spatial and\nspectral priors. In this paper, we propose a Residual Degradation Learning\nUnfolding Framework (RDLUF), which bridges the gap between the sensing matrix\nand the degradation process. Moreover, a Mix$S^2$ Transformer is designed via\nmixing priors across spectral and spatial to strengthen the spectral-spatial\nrepresentation capability. Finally, plugging the Mix$S^2$ Transformer into the\nRDLUF leads to an end-to-end trainable neural network RDLUF-Mix$S^2$.\nExperimental results establish the superior performance of the proposed method\nover existing ones.\n", "date": "2022-11-13", "categories": []}
{"id": "2311.03062", "title": "Imaging through multimode fibres with physical prior", "abstract": "  Imaging through perturbed multimode fibres based on deep learning has been\nwidely researched. However, existing methods mainly use target-speckle pairs in\ndifferent configurations. It is challenging to reconstruct targets without\ntrained networks. In this paper, we propose a physics-assisted, unsupervised,\nlearning-based fibre imaging scheme. The role of the physical prior is to\nsimplify the mapping relationship between the speckle pattern and the target\nimage, thereby reducing the computational complexity. The unsupervised network\nlearns target features according to the optimized direction provided by the\nphysical prior. Therefore, the reconstruction process of the online learning\nonly requires a few speckle patterns and unpaired targets. The proposed scheme\nalso increases the generalization ability of the learning-based method in\nperturbed multimode fibres. Our scheme has the potential to extend the\napplication of multimode fibre imaging.\n", "date": "2023-11-06", "categories": []}
{"id": "2303.07271", "title": "Provably Convergent Plug-and-Play Quasi-Newton Methods", "abstract": "  Plug-and-Play (PnP) methods are a class of efficient iterative methods that\naim to combine data fidelity terms and deep denoisers using classical\noptimization algorithms, such as ISTA or ADMM, with applications in inverse\nproblems and imaging. Provable PnP methods are a subclass of PnP methods with\nconvergence guarantees, such as fixed point convergence or convergence to\ncritical points of some energy function. Many existing provable PnP methods\nimpose heavy restrictions on the denoiser or fidelity function, such as\nnon-expansiveness or strict convexity, respectively. In this work, we propose a\nnovel algorithmic approach incorporating quasi-Newton steps into a provable PnP\nframework based on proximal denoisers, resulting in greatly accelerated\nconvergence while retaining light assumptions on the denoiser. By\ncharacterizing the denoiser as the proximal operator of a weakly convex\nfunction, we show that the fixed points of the proposed quasi-Newton PnP\nalgorithm are critical points of a weakly convex function. Numerical\nexperiments on image deblurring and super-resolution demonstrate 2--8x faster\nconvergence as compared to other provable PnP methods with similar\nreconstruction quality.\n", "date": "2023-03-09", "categories": []}
{"id": "2306.11977", "title": "Encoding Enhanced Complex CNN for Accurate and Highly Accelerated MRI", "abstract": "  Magnetic resonance imaging (MRI) using hyperpolarized noble gases provides a\nway to visualize the structure and function of human lung, but the long imaging\ntime limits its broad research and clinical applications. Deep learning has\ndemonstrated great potential for accelerating MRI by reconstructing images from\nundersampled data. However, most existing deep conventional neural networks\n(CNN) directly apply square convolution to k-space data without considering the\ninherent properties of k-space sampling, limiting k-space learning efficiency\nand image reconstruction quality. In this work, we propose an encoding enhanced\n(EN2) complex CNN for highly undersampled pulmonary MRI reconstruction. EN2\nemploys convolution along either the frequency or phase-encoding direction,\nresembling the mechanisms of k-space sampling, to maximize the utilization of\nthe encoding correlation and integrity within a row or column of k-space. We\nalso employ complex convolution to learn rich representations from the complex\nk-space data. In addition, we develop a feature-strengthened modularized unit\nto further boost the reconstruction performance. Experiments demonstrate that\nour approach can accurately reconstruct hyperpolarized 129Xe and 1H lung MRI\nfrom 6-fold undersampled k-space data and provide lung function measurements\nwith minimal biases compared with fully-sampled image. These results\ndemonstrate the effectiveness of the proposed algorithmic components and\nindicate that the proposed approach could be used for accelerated pulmonary MRI\nin research and clinical lung disease patient care.\n", "date": "2023-06-21", "categories": []}
{"id": "2311.05958", "title": "A Neural Height-Map Approach for the Binocular Photometric Stereo\n  Problem", "abstract": "  In this work we propose a novel, highly practical, binocular photometric\nstereo (PS) framework, which has same acquisition speed as single view PS,\nhowever significantly improves the quality of the estimated geometry.\n  As in recent neural multi-view shape estimation frameworks such as NeRF,\nSIREN and inverse graphics approaches to multi-view photometric stereo (e.g.\nPS-NeRF) we formulate shape estimation task as learning of a differentiable\nsurface and texture representation by minimising surface normal discrepancy for\nnormals estimated from multiple varying light images for two views as well as\ndiscrepancy between rendered surface intensity and observed images. Our method\ndiffers from typical multi-view shape estimation approaches in two key ways.\nFirst, our surface is represented not as a volume but as a neural heightmap\nwhere heights of points on a surface are computed by a deep neural network.\nSecond, instead of predicting an average intensity as PS-NeRF or introducing\nlambertian material assumptions as Guo et al., we use a learnt BRDF and perform\nnear-field per point intensity rendering.\n  Our method achieves the state-of-the-art performance on the DiLiGenT-MV\ndataset adapted to binocular stereo setup as well as a new binocular\nphotometric stereo dataset - LUCES-ST.\n", "date": "2023-11-10", "categories": []}
{"id": "2311.04081", "title": "Learning Super-Resolution Ultrasound Localization Microscopy from\n  Radio-Frequency Data", "abstract": "  Ultrasound Localization Microscopy (ULM) enables imaging of vascular\nstructures in the micrometer range by accumulating contrast agent particle\nlocations over time. Precise and efficient target localization accuracy remains\nan active research topic in the ULM field to further push the boundaries of\nthis promising medical imaging technology. Existing work incorporates\nDelay-And-Sum (DAS) beamforming into particle localization pipelines, which\nultimately determines the ULM image resolution capability. In this paper we\npropose to feed unprocessed Radio-Frequency (RF) data into a super-resolution\nnetwork while bypassing DAS beamforming and its limitations. To facilitate\nthis, we demonstrate label projection and inverse point transformation between\nB-mode and RF coordinate space as required by our approach. We assess our\nmethod against state-of-the-art techniques based on a public dataset featuring\nin silico and in vivo data. Results from our RF-trained network suggest that\nexcluding DAS beamforming offers a great potential to optimize on the ULM\nresolution performance.\n", "date": "2023-11-07", "categories": []}
{"id": "2309.15533", "title": "Uncertainty Quantification via Neural Posterior Principal Components", "abstract": "  Uncertainty quantification is crucial for the deployment of image restoration\nmodels in safety-critical domains, like autonomous driving and biological\nimaging. To date, methods for uncertainty visualization have mainly focused on\nper-pixel estimates. Yet, a heatmap of per-pixel variances is typically of\nlittle practical use, as it does not capture the strong correlations between\npixels. A more natural measure of uncertainty corresponds to the variances\nalong the principal components (PCs) of the posterior distribution.\nTheoretically, the PCs can be computed by applying PCA on samples generated\nfrom a conditional generative model for the input image. However, this requires\ngenerating a very large number of samples at test time, which is painfully slow\nwith the current state-of-the-art (diffusion) models. In this work, we present\na method for predicting the PCs of the posterior distribution for any input\nimage, in a single forward pass of a neural network. Our method can either wrap\naround a pre-trained model that was trained to minimize the mean square error\n(MSE), or can be trained from scratch to output both a predicted image and the\nposterior PCs. We showcase our method on multiple inverse problems in imaging,\nincluding denoising, inpainting, super-resolution, and biological\nimage-to-image translation. Our method reliably conveys instance-adaptive\nuncertainty directions, achieving uncertainty quantification comparable with\nposterior samplers while being orders of magnitude faster. Code and examples\nare available at https://eliasnehme.github.io/NPPC/\n", "date": "2023-09-27", "categories": []}
{"id": "2309.01865", "title": "BigFUSE: Global Context-Aware Image Fusion in Dual-View Light-Sheet\n  Fluorescence Microscopy with Image Formation Prior", "abstract": "  Light-sheet fluorescence microscopy (LSFM), a planar illumination technique\nthat enables high-resolution imaging of samples, experiences defocused image\nquality caused by light scattering when photons propagate through thick\ntissues. To circumvent this issue, dualview imaging is helpful. It allows\nvarious sections of the specimen to be scanned ideally by viewing the sample\nfrom opposing orientations. Recent image fusion approaches can then be applied\nto determine in-focus pixels by comparing image qualities of two views locally\nand thus yield spatially inconsistent focus measures due to their limited\nfield-of-view. Here, we propose BigFUSE, a global context-aware image fuser\nthat stabilizes image fusion in LSFM by considering the global impact of photon\npropagation in the specimen while determining focus-defocus based on local\nimage qualities. Inspired by the image formation prior in dual-view LSFM, image\nfusion is considered as estimating a focus-defocus boundary using Bayes\nTheorem, where (i) the effect of light scattering onto focus measures is\nincluded within Likelihood; and (ii) the spatial consistency regarding\nfocus-defocus is imposed in Prior. The expectation-maximum algorithm is then\nadopted to estimate the focus-defocus boundary. Competitive experimental\nresults show that BigFUSE is the first dual-view LSFM fuser that is able to\nexclude structured artifacts when fusing information, highlighting its\nabilities of automatic image fusion.\n", "date": "2023-09-05", "categories": []}
{"id": "2311.01216", "title": "Convergent plug-and-play with proximal denoiser and unconstrained\n  regularization parameter", "abstract": "  In this work, we present new proofs of convergence for Plug-and-Play (PnP)\nalgorithms. PnP methods are efficient iterative algorithms for solving image\ninverse problems where regularization is performed by plugging a pre-trained\ndenoiser in a proximal algorithm, such as Proximal Gradient Descent (PGD) or\nDouglas-Rachford Splitting (DRS). Recent research has explored convergence by\nincorporating a denoiser that writes exactly as a proximal operator. However,\nthe corresponding PnP algorithm has then to be run with stepsize equal to $1$.\nThe stepsize condition for nonconvex convergence of the proximal algorithm in\nuse then translates to restrictive conditions on the regularization parameter\nof the inverse problem. This can severely degrade the restoration capacity of\nthe algorithm. In this paper, we present two remedies for this limitation.\nFirst, we provide a novel convergence proof for PnP-DRS that does not impose\nany restrictions on the regularization parameter. Second, we examine a relaxed\nversion of the PGD algorithm that converges across a broader range of\nregularization parameters. Our experimental study, conducted on deblurring and\nsuper-resolution experiments, demonstrate that both of these solutions enhance\nthe accuracy of image restoration.\n", "date": "2023-11-02", "categories": []}
{"id": "2311.01292", "title": "Joint 3D Shape and Motion Estimation from Rolling Shutter Light-Field\n  Images", "abstract": "  In this paper, we propose an approach to address the problem of 3D\nreconstruction of scenes from a single image captured by a light-field camera\nequipped with a rolling shutter sensor. Our method leverages the 3D information\ncues present in the light-field and the motion information provided by the\nrolling shutter effect. We present a generic model for the imaging process of\nthis sensor and a two-stage algorithm that minimizes the re-projection error\nwhile considering the position and motion of the camera in a motion-shape\nbundle adjustment estimation strategy. Thereby, we provide an instantaneous 3D\nshape-and-pose-and-velocity sensing paradigm. To the best of our knowledge,\nthis is the first study to leverage this type of sensor for this purpose. We\nalso present a new benchmark dataset composed of different light-fields showing\nrolling shutter effects, which can be used as a common base to improve the\nevaluation and tracking the progress in the field. We demonstrate the\neffectiveness and advantages of our approach through several experiments\nconducted for different scenes and types of motions. The source code and\ndataset are publicly available at: https://github.com/ICB-Vision-AI/RSLF\n", "date": "2023-11-02", "categories": []}
{"id": "2311.00186", "title": "Image Restoration with Point Spread Function Regularization and Active\n  Learning", "abstract": "  Large-scale astronomical surveys can capture numerous images of celestial\nobjects, including galaxies and nebulae. Analysing and processing these images\ncan reveal intricate internal structures of these objects, allowing researchers\nto conduct comprehensive studies on their morphology, evolution, and physical\nproperties. However, varying noise levels and point spread functions can hamper\nthe accuracy and efficiency of information extraction from these images. To\nmitigate these effects, we propose a novel image restoration algorithm that\nconnects a deep learning-based restoration algorithm with a high-fidelity\ntelescope simulator. During the training stage, the simulator generates images\nwith different levels of blur and noise to train the neural network based on\nthe quality of restored images. After training, the neural network can directly\nrestore images obtained by the telescope, as represented by the simulator. We\nhave tested the algorithm using real and simulated observation data and have\nfound that it effectively enhances fine structures in blurry images and\nincreases the quality of observation images. This algorithm can be applied to\nlarge-scale sky survey data, such as data obtained by LSST, Euclid, and CSST,\nto further improve the accuracy and efficiency of information extraction,\npromoting advances in the field of astronomical research.\n", "date": "2023-10-31", "categories": []}
{"id": "2310.20265", "title": "Low-Dose CT Image Enhancement Using Deep Learning", "abstract": "  The application of ionizing radiation for diagnostic imaging is common around\nthe globe. However, the process of imaging, itself, remains to be a relatively\nhazardous operation. Therefore, it is preferable to use as low a dose of\nionizing radiation as possible, particularly in computed tomography (CT)\nimaging systems, where multiple x-ray operations are performed for the\nreconstruction of slices of body tissues. A popular method for radiation dose\nreduction in CT imaging is known as the quarter-dose technique, which reduces\nthe x-ray dose but can cause a loss of image sharpness. Since CT image\nreconstruction from directional x-rays is a nonlinear process, it is\nanalytically difficult to correct the effect of dose reduction on image\nquality. Recent and popular deep-learning approaches provide an intriguing\npossibility of image enhancement for low-dose artifacts. Some recent works\npropose combinations of multiple deep-learning and classical methods for this\npurpose, which over-complicate the process. However, it is observed here that\nthe straight utilization of the well-known U-NET provides very successful\nresults for the correction of low-dose artifacts. Blind tests with actual\nradiologists reveal that the U-NET enhanced quarter-dose CT images not only\nprovide an immense visual improvement over the low-dose versions, but also\nbecome diagnostically preferable images, even when compared to their full-dose\nCT versions.\n", "date": "2023-10-31", "categories": []}
{"id": "2310.20279", "title": "Machine learning refinement of in situ images acquired by low electron\n  dose LC-TEM", "abstract": "  We study a machine learning (ML) technique for refining images acquired\nduring in situ observation using liquid-cell transmission electron microscopy\n(LC-TEM). Our model is constructed using a U-Net architecture and a ResNet\nencoder. For training our ML model, we prepared an original image dataset that\ncontained pairs of images of samples acquired with and without a solution\npresent. The former images were used as noisy images and the latter images were\nused as corresponding ground truth images. The number of pairs of image sets\nwas $1,204$ and the image sets included images acquired at several different\nmagnifications and electron doses. The trained model converted a noisy image\ninto a clear image. The time necessary for the conversion was on the order of\n10ms, and we applied the model to in situ observations using the software Gatan\nDigitalMicrograph (DM). Even if a nanoparticle was not visible in a view window\nin the DM software because of the low electron dose, it was visible in a\nsuccessive refined image generated by our ML model.\n", "date": "2023-10-31", "categories": []}
{"id": "2307.07288", "title": "Implicit Neural Feature Fusion Function for Multispectral and\n  Hyperspectral Image Fusion", "abstract": "  Multispectral and Hyperspectral Image Fusion (MHIF) is a practical task that\naims to fuse a high-resolution multispectral image (HR-MSI) and a\nlow-resolution hyperspectral image (LR-HSI) of the same scene to obtain a\nhigh-resolution hyperspectral image (HR-HSI). Benefiting from powerful\ninductive bias capability, CNN-based methods have achieved great success in the\nMHIF task. However, they lack certain interpretability and require convolution\nstructures be stacked to enhance performance. Recently, Implicit Neural\nRepresentation (INR) has achieved good performance and interpretability in 2D\ntasks due to its ability to locally interpolate samples and utilize multimodal\ncontent such as pixels and coordinates. Although INR-based approaches show\npromise, they require extra construction of high-frequency information\n(\\emph{e.g.,} positional encoding). In this paper, inspired by previous work of\nMHIF task, we realize that HR-MSI could serve as a high-frequency detail\nauxiliary input, leading us to propose a novel INR-based hyperspectral fusion\nfunction named Implicit Neural Feature Fusion Function (INF). As an elaborate\nstructure, it solves the MHIF task and addresses deficiencies in the INR-based\napproaches. Specifically, our INF designs a Dual High-Frequency Fusion (DHFF)\nstructure that obtains high-frequency information twice from HR-MSI and LR-HSI,\nthen subtly fuses them with coordinate information. Moreover, the proposed INF\nincorporates a parameter-free method named INR with cosine similarity (INR-CS)\nthat uses cosine similarity to generate local weights through feature vectors.\nBased on INF, we construct an Implicit Neural Fusion Network (INFN) that\nachieves state-of-the-art performance for MHIF tasks of two public datasets,\n\\emph{i.e.,} CAVE and Harvard. The code will soon be made available on GitHub.\n", "date": "2023-07-14", "categories": []}
{"id": "2310.19011", "title": "Efficient Test-Time Adaptation for Super-Resolution with Second-Order\n  Degradation and Reconstruction", "abstract": "  Image super-resolution (SR) aims to learn a mapping from low-resolution (LR)\nto high-resolution (HR) using paired HR-LR training images. Conventional SR\nmethods typically gather the paired training data by synthesizing LR images\nfrom HR images using a predetermined degradation model, e.g., Bicubic\ndown-sampling. However, the realistic degradation type of test images may\nmismatch with the training-time degradation type due to the dynamic changes of\nthe real-world scenarios, resulting in inferior-quality SR images. To address\nthis, existing methods attempt to estimate the degradation model and train an\nimage-specific model, which, however, is quite time-consuming and impracticable\nto handle rapidly changing domain shifts. Moreover, these methods largely\nconcentrate on the estimation of one degradation type (e.g., blur degradation),\noverlooking other degradation types like noise and JPEG in real-world test-time\nscenarios, thus limiting their practicality. To tackle these problems, we\npresent an efficient test-time adaptation framework for SR, named SRTTA, which\nis able to quickly adapt SR models to test domains with different/unknown\ndegradation types. Specifically, we design a second-order degradation scheme to\nconstruct paired data based on the degradation type of the test image, which is\npredicted by a pre-trained degradation classifier. Then, we adapt the SR model\nby implementing feature-level reconstruction learning from the initial test\nimage to its second-order degraded counterparts, which helps the SR model\ngenerate plausible HR images. Extensive experiments are conducted on newly\nsynthesized corrupted DIV2K datasets with 8 different degradations and several\nreal-world datasets, demonstrating that our SRTTA framework achieves an\nimpressive improvement over existing methods with satisfying speed. The source\ncode is available at https://github.com/DengZeshuai/SRTTA.\n", "date": "2023-10-29", "categories": []}
{"id": "2302.08478", "title": "Kernelized Back-Projection Networks for Blind Super Resolution", "abstract": "  Since non-blind Super Resolution (SR) fails to super-resolve Low-Resolution\n(LR) images degraded by arbitrary degradations, SR with the degradation model\nis required. However, this paper reveals that non-blind SR that is trained\nsimply with various blur kernels exhibits comparable performance as those with\nthe degradation model for blind SR. This result motivates us to revisit\nhigh-performance non-blind SR and extend it to blind SR with blur kernels. This\npaper proposes two SR networks by integrating kernel estimation and SR branches\nin an iterative end-to-end manner. In the first model, which is called the\nKernel Conditioned Back-Projection Network (KCBPN), the low-dimensional kernel\nrepresentations are estimated for conditioning the SR branch. In our second\nmodel, the Kernelized BackProjection Network (KBPN), a raw kernel is estimated\nand directly employed for modeling the image degradation. The estimated kernel\nis employed not only for back-propagating its residual but also for\nforward-propagating the residual to iterative stages. This forward-propagation\nencourages these stages to learn a variety of different features in different\nstages by focusing on pixels with large residuals in each stage. Experimental\nresults validate the effectiveness of our proposed networks for kernel\nestimation and SR. We will release the code for this work.\n", "date": "2023-02-16", "categories": []}
{"id": "2304.06706", "title": "Zip-NeRF: Anti-Aliased Grid-Based Neural Radiance Fields", "abstract": "  Neural Radiance Field training can be accelerated through the use of\ngrid-based representations in NeRF's learned mapping from spatial coordinates\nto colors and volumetric density. However, these grid-based approaches lack an\nexplicit understanding of scale and therefore often introduce aliasing, usually\nin the form of jaggies or missing scene content. Anti-aliasing has previously\nbeen addressed by mip-NeRF 360, which reasons about sub-volumes along a cone\nrather than points along a ray, but this approach is not natively compatible\nwith current grid-based techniques. We show how ideas from rendering and signal\nprocessing can be used to construct a technique that combines mip-NeRF 360 and\ngrid-based models such as Instant NGP to yield error rates that are 8% - 77%\nlower than either prior technique, and that trains 24x faster than mip-NeRF\n360.\n", "date": "2023-04-13", "categories": []}
{"id": "2305.12672", "title": "Block Coordinate Plug-and-Play Methods for Blind Inverse Problems", "abstract": "  Plug-and-play (PnP) prior is a well-known class of methods for solving\nimaging inverse problems by computing fixed-points of operators combining\nphysical measurement models and learned image denoisers. While PnP methods have\nbeen extensively used for image recovery with known measurement operators,\nthere is little work on PnP for solving blind inverse problems. We address this\ngap by presenting a new block-coordinate PnP (BC-PnP) method that efficiently\nsolves this joint estimation problem by introducing learned denoisers as priors\non both the unknown image and the unknown measurement operator. We present a\nnew convergence theory for BC-PnP compatible with blind inverse problems by\nconsidering nonconvex data-fidelity terms and expansive denoisers. Our theory\nanalyzes the convergence of BC-PnP to a stationary point of an implicit\nfunction associated with an approximate minimum mean-squared error (MMSE)\ndenoiser. We numerically validate our method on two blind inverse problems:\nautomatic coil sensitivity estimation in magnetic resonance imaging (MRI) and\nblind image deblurring. Our results show that BC-PnP provides an efficient and\nprincipled framework for using denoisers as PnP priors for jointly estimating\nmeasurement operators and images.\n", "date": "2023-05-22", "categories": []}
{"id": "2310.17817", "title": "Bayesian imaging inverse problem with SA-Roundtrip prior via HMC-pCN\n  sampler", "abstract": "  Bayesian inference with deep generative prior has received considerable\ninterest for solving imaging inverse problems in many scientific and\nengineering fields. The selection of the prior distribution is learned from,\nand therefore an important representation learning of, available prior\nmeasurements. The SA-Roundtrip, a novel deep generative prior, is introduced to\nenable controlled sampling generation and identify the data's intrinsic\ndimension. This prior incorporates a self-attention structure within a\nbidirectional generative adversarial network. Subsequently, Bayesian inference\nis applied to the posterior distribution in the low-dimensional latent space\nusing the Hamiltonian Monte Carlo with preconditioned Crank-Nicolson (HMC-pCN)\nalgorithm, which is proven to be ergodic under specific conditions. Experiments\nconducted on computed tomography (CT) reconstruction with the MNIST and\nTomoPhantom datasets reveal that the proposed method outperforms\nstate-of-the-art comparisons, consistently yielding a robust and superior point\nestimator along with precise uncertainty quantification.\n", "date": "2023-10-24", "categories": []}
{"id": "2310.17997", "title": "Deep Learning Enables Large Depth-of-Field Images for\n  Sub-Diffraction-Limit Scanning Superlens Microscopy", "abstract": "  Scanning electron microscopy (SEM) is indispensable in diverse applications\nranging from microelectronics to food processing because it provides large\ndepth-of-field images with a resolution beyond the optical diffraction limit.\nHowever, the technology requires coating conductive films on insulator samples\nand a vacuum environment. We use deep learning to obtain the mapping\nrelationship between optical super-resolution (OSR) images and SEM domain\nimages, which enables the transformation of OSR images into SEM-like large\ndepth-of-field images. Our custom-built scanning superlens microscopy (SSUM)\nsystem, which requires neither coating samples by conductive films nor a vacuum\nenvironment, is used to acquire the OSR images with features down to ~80 nm.\nThe peak signal-to-noise ratio (PSNR) and structural similarity index measure\nvalues indicate that the deep learning method performs excellently in\nimage-to-image translation, with a PSNR improvement of about 0.74 dB over the\noptical super-resolution images. The proposed method provides a high level of\ndetail in the reconstructed results, indicating that it has broad applicability\nto chip-level defect detection, biological sample analysis, forensics, and\nvarious other fields.\n", "date": "2023-10-27", "categories": []}
{"id": "2310.17188", "title": "Blind Image Super-resolution with Rich Texture-Aware Codebooks", "abstract": "  Blind super-resolution (BSR) methods based on high-resolution (HR)\nreconstruction codebooks have achieved promising results in recent years.\nHowever, we find that a codebook based on HR reconstruction may not effectively\ncapture the complex correlations between low-resolution (LR) and HR images. In\ndetail, multiple HR images may produce similar LR versions due to complex blind\ndegradations, causing the HR-dependent only codebooks having limited texture\ndiversity when faced with confusing LR inputs. To alleviate this problem, we\npropose the Rich Texture-aware Codebook-based Network (RTCNet), which consists\nof the Degradation-robust Texture Prior Module (DTPM) and the Patch-aware\nTexture Prior Module (PTPM). DTPM effectively mines the cross-resolution\ncorrelation of textures between LR and HR images by exploiting the\ncross-resolution correspondence of textures. PTPM uses patch-wise semantic\npre-training to correct the misperception of texture similarity in the\nhigh-level semantic regularization. By taking advantage of this, RTCNet\neffectively gets rid of the misalignment of confusing textures between HR and\nLR in the BSR scenarios. Experiments show that RTCNet outperforms\nstate-of-the-art methods on various benchmarks by up to 0.16 ~ 0.46dB.\n", "date": "2023-10-26", "categories": []}
{"id": "2307.16508", "title": "Towards General Low-Light Raw Noise Synthesis and Modeling", "abstract": "  Modeling and synthesizing low-light raw noise is a fundamental problem for\ncomputational photography and image processing applications. Although most\nrecent works have adopted physics-based models to synthesize noise, the\nsignal-independent noise in low-light conditions is far more complicated and\nvaries dramatically across camera sensors, which is beyond the description of\nthese models. To address this issue, we introduce a new perspective to\nsynthesize the signal-independent noise by a generative model. Specifically, we\nsynthesize the signal-dependent and signal-independent noise in a physics- and\nlearning-based manner, respectively. In this way, our method can be considered\nas a general model, that is, it can simultaneously learn different noise\ncharacteristics for different ISO levels and generalize to various sensors.\nSubsequently, we present an effective multi-scale discriminator termed Fourier\ntransformer discriminator (FTD) to distinguish the noise distribution\naccurately. Additionally, we collect a new low-light raw denoising (LRD)\ndataset for training and benchmarking. Qualitative validation shows that the\nnoise generated by our proposed noise model can be highly similar to the real\nnoise in terms of distribution. Furthermore, extensive denoising experiments\ndemonstrate that our method performs favorably against state-of-the-art methods\non different sensors.\n", "date": "2023-07-31", "categories": []}
{"id": "2310.15850", "title": "Posterior Estimation for Dynamic PET imaging using Conditional\n  Variational Inference", "abstract": "  This work aims efficiently estimating the posterior distribution of kinetic\nparameters for dynamic positron emission tomography (PET) imaging given a\nmeasurement of time of activity curve. Considering the inherent information\nloss from parametric imaging to measurement space with the forward kinetic\nmodel, the inverse mapping is ambiguous. The conventional (but expensive)\nsolution can be the Markov Chain Monte Carlo (MCMC) sampling, which is known to\nproduce unbiased asymptotical estimation. We propose a deep-learning-based\nframework for efficient posterior estimation. Specifically, we counteract the\ninformation loss in the forward process by introducing latent variables. Then,\nwe use a conditional variational autoencoder (CVAE) and optimize its evidence\nlower bound. The well-trained decoder is able to infer the posterior with a\ngiven measurement and the sampled latent variables following a simple\nmultivariate Gaussian distribution. We validate our CVAE-based method using\nunbiased MCMC as the reference for low-dimensional data (a single brain region)\nwith the simplified reference tissue model.\n", "date": "2023-10-24", "categories": []}
{"id": "2305.18878", "title": "BPF Algorithms for Multiple Source-Translation Computed Tomography\n  Reconstruction", "abstract": "  Micro-computed tomography (micro-CT) is a widely used state-of-the-art\ninstrument employed to study the morphological structures of objects in various\nfields. However, its small field-of-view (FOV) cannot meet the pressing demand\nfor imaging relatively large objects at high spatial resolutions. Recently, we\ndevised a novel scanning mode called multiple source translation CT (mSTCT)\nthat effectively enlarges the FOV of the micro-CT and correspondingly developed\na virtual projection-based filtered backprojection (V-FBP) algorithm for\nreconstruction. Although V-FBP skillfully solves the truncation problem in\nmSTCT, it requires densely sampled projections to arrive at high-resolution\nreconstruction, which reduces imaging efficiency. In this paper, we developed\ntwo backprojection-filtration (BPF)-based algorithms for mSTCT, i.e., S-BPF\n(derivatives along source) and D-BPF (derivatives along detector). D-BPF can\nachieve high-resolution reconstruction with fewer projections than V-FBP and\nS-BPF. Through simulated and real experiments conducted in this paper, we\ndemonstrate that D-BPF can reduce source sampling by 75% compared with V-FBP at\nthe same spatial resolution, which makes mSTCT more feasible in practice.\nMeanwhile, S-BPF can yield more stable results than D-BPF, which is similar to\nV-FBP.\n", "date": "2023-05-30", "categories": []}
{"id": "2309.15563", "title": "Guided Frequency Loss for Image Restoration", "abstract": "  Image Restoration has seen remarkable progress in recent years. Many\ngenerative models have been adapted to tackle the known restoration cases of\nimages. However, the interest in benefiting from the frequency domain is not\nwell explored despite its major factor in these particular cases of image\nsynthesis. In this study, we propose the Guided Frequency Loss (GFL), which\nhelps the model to learn in a balanced way the image's frequency content\nalongside the spatial content. It aggregates three major components that work\nin parallel to enhance learning efficiency; a Charbonnier component, a\nLaplacian Pyramid component, and a Gradual Frequency component. We tested GFL\non the Super Resolution and the Denoising tasks. We used three different\ndatasets and three different architectures for each of them. We found that the\nGFL loss improved the PSNR metric in most implemented experiments. Also, it\nimproved the training of the Super Resolution models in both SwinIR and SRGAN.\nIn addition, the utility of the GFL loss increased better on constrained data\ndue to the less stochasticity in the high frequencies' components among\nsamples.\n", "date": "2023-09-27", "categories": []}
{"id": "2310.15119", "title": "Compressed Sensing of Generative Sparse-latent (GSL) Signals", "abstract": "  We consider reconstruction of an ambient signal in a compressed sensing (CS)\nsetup where the ambient signal has a neural network based generative model. The\ngenerative model has a sparse-latent input and we refer to the generated\nambient signal as generative sparse-latent signal (GSL). The proposed sparsity\ninducing reconstruction algorithm is inherently non-convex, and we show that a\ngradient based search provides a good reconstruction performance. We evaluate\nour proposed algorithm using simulated data.\n", "date": "2023-10-16", "categories": []}
{"id": "2310.12990", "title": "Wave-informed dictionary learning for high-resolution imaging in complex\n  media", "abstract": "  We propose an approach for imaging in scattering media when large and diverse\ndata sets are available. It has two steps. Using a dictionary learning\nalgorithm the first step estimates the true Green's function vectors as columns\nin an unordered sensing matrix. The array data comes from many sparse sets of\nsources whose location and strength are not known to us. In the second step,\nthe columns of the estimated sensing matrix are ordered for imaging using\nMulti-Dimensional Scaling with connectivity information derived from\ncross-correlations of its columns, as in time reversal. For these two steps to\nwork together we need data from large arrays of receivers so the columns of the\nsensing matrix are incoherent for the first step, as well as from sub-arrays so\nthat they are coherent enough to obtain the connectivity needed in the second\nstep. Through simulation experiments, we show that the proposed approach is\nable to provide images in complex media whose resolution is that of a\nhomogeneous medium.\n", "date": "2023-09-22", "categories": []}
{"id": "2310.13670", "title": "ManifoldNeRF: View-dependent Image Feature Supervision for Few-shot\n  Neural Radiance Fields", "abstract": "  Novel view synthesis has recently made significant progress with the advent\nof Neural Radiance Fields (NeRF). DietNeRF is an extension of NeRF that aims to\nachieve this task from only a few images by introducing a new loss function for\nunknown viewpoints with no input images. The loss function assumes that a\npre-trained feature extractor should output the same feature even if input\nimages are captured at different viewpoints since the images contain the same\nobject. However, while that assumption is ideal, in reality, it is known that\nas viewpoints continuously change, also feature vectors continuously change.\nThus, the assumption can harm training. To avoid this harmful training, we\npropose ManifoldNeRF, a method for supervising feature vectors at unknown\nviewpoints using interpolated features from neighboring known viewpoints. Since\nthe method provides appropriate supervision for each unknown viewpoint by the\ninterpolated features, the volume representation is learned better than\nDietNeRF. Experimental results show that the proposed method performs better\nthan others in a complex scene. We also experimented with several subsets of\nviewpoints from a set of viewpoints and identified an effective set of\nviewpoints for real environments. This provided a basic policy of viewpoint\npatterns for real-world application. The code is available at\nhttps://github.com/haganelego/ManifoldNeRF_BMVC2023\n", "date": "2023-10-20", "categories": []}
{"id": "2307.12672", "title": "Global k-Space Interpolation for Dynamic MRI Reconstruction using Masked\n  Image Modeling", "abstract": "  In dynamic Magnetic Resonance Imaging (MRI), k-space is typically\nundersampled due to limited scan time, resulting in aliasing artifacts in the\nimage domain. Hence, dynamic MR reconstruction requires not only modeling\nspatial frequency components in the x and y directions of k-space but also\nconsidering temporal redundancy. Most previous works rely on image-domain\nregularizers (priors) to conduct MR reconstruction. In contrast, we focus on\ninterpolating the undersampled k-space before obtaining images with Fourier\ntransform. In this work, we connect masked image modeling with k-space\ninterpolation and propose a novel Transformer-based k-space Global\nInterpolation Network, termed k-GIN. Our k-GIN learns global dependencies among\nlow- and high-frequency components of 2D+t k-space and uses it to interpolate\nunsampled data. Further, we propose a novel k-space Iterative Refinement Module\n(k-IRM) to enhance the high-frequency components learning. We evaluate our\napproach on 92 in-house 2D+t cardiac MR subjects and compare it to MR\nreconstruction methods with image-domain regularizers. Experiments show that\nour proposed k-space interpolation method quantitatively and qualitatively\noutperforms baseline methods. Importantly, the proposed approach achieves\nsubstantially higher robustness and generalizability in cases of\nhighly-undersampled MR data. For video presentation, poster, GIF results and\ncode please check our project page:\nhttps://jzpeterpan.github.io/k-gin.github.io/.\n", "date": "2023-07-24", "categories": []}
{"id": "2310.05647", "title": "Exploiting Manifold Structured Data Priors for Improved MR\n  Fingerprinting Reconstruction", "abstract": "  Estimating tissue parameter maps with high accuracy and precision from highly\nundersampled measurements presents one of the major challenges in MR\nfingerprinting (MRF). Many existing works project the recovered voxel\nfingerprints onto the Bloch manifold to improve reconstruction performance.\nHowever, little research focuses on exploiting the latent manifold structure\npriors among fingerprints. To fill this gap, we propose a novel MRF\nreconstruction framework based on manifold structured data priors. Since it is\ndifficult to directly estimate the fingerprint manifold structure, we model the\ntissue parameters as points on a low-dimensional parameter manifold. We reveal\nthat the fingerprint manifold shares the same intrinsic topology as the\nparameter manifold, although being embedded in different Euclidean spaces. To\nexploit the non-linear and non-local redundancies in MRF data, we divide the\nMRF data into spatial patches, and the similarity measurement among data\npatches can be accurately obtained using the Euclidean distance between the\ncorresponding patches in the parameter manifold. The measured similarity is\nthen used to construct the graph Laplacian operator, which represents the\nfingerprint manifold structure. Thus, the fingerprint manifold structure is\nintroduced in the reconstruction framework by using the low-dimensional\nparameter manifold. Additionally, we incorporate the locally low-rank prior in\nthe reconstruction framework to further utilize the local correlations within\neach patch for improved reconstruction performance. We also adopt a\nGPU-accelerated NUFFT library to accelerate reconstruction in non-Cartesian\nsampling scenarios. Experimental results demonstrate that our method can\nachieve significantly improved reconstruction performance with reduced\ncomputational time over the state-of-the-art methods.\n", "date": "2023-10-09", "categories": []}
{"id": "2310.11276", "title": "Video Super-Resolution Using a Grouped Residual in Residual Network", "abstract": "  Super-resolution (SR) is the technique of increasing the nominal resolution\nof image / video content accompanied with quality improvement. Video\nsuper-resolution (VSR) can be considered as the generalization of single image\nsuper-resolution (SISR). This generalization should be such that more detail is\ncreated in the output using adjacent input frames. In this paper, we propose a\ngrouped residual in residual network (GRRN) for VSR. By adjusting the\nhyperparameters of the proposed structure, we train three networks with\ndifferent numbers of parameters and compare their quantitative and qualitative\nresults with the existing methods. Although based on some quantitative\ncriteria, GRRN does not provide better results than the existing methods, in\nterms of the quality of the output image it has acceptable performance.\n", "date": "2023-10-17", "categories": []}
{"id": "2212.01312", "title": "Hybrid adiabatic quantum computing for tomographic image reconstruction\n  -- opportunities and limitations", "abstract": "  Our goal is to reconstruct tomographic images with few measurements and a low\nsignal-to-noise ratio. In clinical imaging, this helps to improve patient\ncomfort and reduce radiation exposure. As quantum computing advances, we\npropose to use an adiabatic quantum computer and associated hybrid methods to\nsolve the reconstruction problem. Tomographic reconstruction is an ill-posed\ninverse problem. We test our reconstruction technique for image size, noise\ncontent, and underdetermination of the measured projection data. We then\npresent the reconstructed binary and integer-valued images of up to 32 by 32\npixels. The demonstrated method competes with traditional reconstruction\nalgorithms and is superior in terms of robustness to noise and reconstructions\nfrom few projections. We postulate that hybrid quantum computing will soon\nreach maturity for real applications in tomographic reconstruction. Finally, we\npoint out the current limitations regarding the problem size and\ninterpretability of the algorithm.\n", "date": "2022-12-02", "categories": []}
{"id": "2304.08342", "title": "NF-ULA: Langevin Monte Carlo with Normalizing Flow Prior for Imaging\n  Inverse Problems", "abstract": "  Bayesian methods for solving inverse problems are a powerful alternative to\nclassical methods since the Bayesian approach offers the ability to quantify\nthe uncertainty in the solution. In recent years, data-driven techniques for\nsolving inverse problems have also been remarkably successful, due to their\nsuperior representation ability. In this work, we incorporate data-based models\ninto a class of Langevin-based sampling algorithms for Bayesian inference in\nimaging inverse problems. In particular, we introduce NF-ULA (Normalizing\nFlow-based Unadjusted Langevin algorithm), which involves learning a\nnormalizing flow (NF) as the image prior. We use NF to learn the prior because\na tractable closed-form expression for the log prior enables the\ndifferentiation of it using autograd libraries. Our algorithm only requires a\nnormalizing flow-based generative network, which can be pre-trained\nindependently of the considered inverse problem and the forward operator. We\nperform theoretical analysis by investigating the well-posedness and\nnon-asymptotic convergence of the resulting NF-ULA algorithm. The efficacy of\nthe proposed NF-ULA algorithm is demonstrated in various image restoration\nproblems such as image deblurring, image inpainting, and limited-angle X-ray\ncomputed tomography (CT) reconstruction. NF-ULA is found to perform better than\ncompeting methods for severely ill-posed inverse problems.\n", "date": "2023-04-17", "categories": []}
{"id": "2306.14268", "title": "Adaptive Window Pruning for Efficient Local Motion Deblurring", "abstract": "  Local motion blur commonly occurs in real-world photography due to the mixing\nbetween moving objects and stationary backgrounds during exposure. Existing\nimage deblurring methods predominantly focus on global deblurring,\ninadvertently affecting the sharpness of backgrounds in locally blurred images\nand wasting unnecessary computation on sharp pixels, especially for\nhigh-resolution images. This paper aims to adaptively and efficiently restore\nhigh-resolution locally blurred images. We propose a local motion deblurring\nvision Transformer (LMD-ViT) built on adaptive window pruning Transformer\nblocks (AdaWPT). To focus deblurring on local regions and reduce computation,\nAdaWPT prunes unnecessary windows, only allowing the active windows to be\ninvolved in the deblurring processes. The pruning operation relies on the\nblurriness confidence predicted by a confidence predictor that is trained\nend-to-end using a reconstruction loss with Gumbel-Softmax re-parameterization\nand a pruning loss guided by annotated blur masks. Our method removes local\nmotion blur effectively without distorting sharp regions, demonstrated by its\nexceptional perceptual and quantitative improvements compared to\nstate-of-the-art methods. In addition, our approach substantially reduces FLOPs\nby 66% and achieves more than a twofold increase in inference speed compared to\nTransformer-based deblurring methods. We will make our code and annotated blur\nmasks publicly available.\n", "date": "2023-06-25", "categories": []}
{"id": "2310.09625", "title": "JSMoCo: Joint Coil Sensitivity and Motion Correction in Parallel MRI\n  with a Self-Calibrating Score-Based Diffusion Model", "abstract": "  Magnetic Resonance Imaging (MRI) stands as a powerful modality in clinical\ndiagnosis. However, it is known that MRI faces challenges such as long\nacquisition time and vulnerability to motion-induced artifacts. Despite the\nsuccess of many existing motion correction algorithms, there has been limited\nresearch focused on correcting motion artifacts on the estimated coil\nsensitivity maps for fast MRI reconstruction. Existing methods might suffer\nfrom severe performance degradation due to error propagation resulting from the\ninaccurate coil sensitivity maps estimation. In this work, we propose to\njointly estimate the motion parameters and coil sensitivity maps for\nunder-sampled MRI reconstruction, referred to as JSMoCo. However, joint\nestimation of motion parameters and coil sensitivities results in a highly\nill-posed inverse problem due to an increased number of unknowns. To address\nthis, we introduce score-based diffusion models as powerful priors and leverage\nthe MRI physical principles to efficiently constrain the solution space for\nthis optimization problem. Specifically, we parameterize the rigid motion as\nthree trainable variables and model coil sensitivity maps as polynomial\nfunctions. Leveraging the physical knowledge, we then employ Gibbs sampler for\njoint estimation, ensuring system consistency between sensitivity maps and\ndesired images, avoiding error propagation from pre-estimated sensitivity maps\nto the reconstructed images. We conduct comprehensive experiments to evaluate\nthe performance of JSMoCo on the fastMRI dataset. The results show that our\nmethod is capable of reconstructing high-quality MRI images from\nsparsely-sampled k-space data, even affected by motion. It achieves this by\naccurately estimating both motion parameters and coil sensitivities,\neffectively mitigating motion-related challenges during MRI reconstruction.\n", "date": "2023-10-14", "categories": []}
{"id": "2310.10033", "title": "Deep Unfolding Network for Image Compressed Sensing by Content-adaptive\n  Gradient Updating and Deformation-invariant Non-local Modeling", "abstract": "  Inspired by certain optimization solvers, the deep unfolding network (DUN)\nhas attracted much attention in recent years for image compressed sensing (CS).\nHowever, there still exist the following two issues: 1) In existing DUNs, most\nhyperparameters are usually content independent, which greatly limits their\nadaptability for different input contents. 2) In each iteration, a plain\nconvolutional neural network is usually adopted, which weakens the perception\nof wider context prior and therefore depresses the expressive ability. In this\npaper, inspired by the traditional Proximal Gradient Descent (PGD) algorithm, a\nnovel DUN for image compressed sensing (dubbed DUN-CSNet) is proposed to solve\nthe above two issues. Specifically, for the first issue, a novel content\nadaptive gradient descent network is proposed, in which a well-designed step\nsize generation sub-network is developed to dynamically allocate the\ncorresponding step sizes for different textures of input image by generating a\ncontent-aware step size map, realizing a content-adaptive gradient updating.\nFor the second issue, considering the fact that many similar patches exist in\nan image but have undergone a deformation, a novel deformation-invariant\nnon-local proximal mapping network is developed, which can adaptively build the\nlong-range dependencies between the nonlocal patches by deformation-invariant\nnon-local modeling, leading to a wider perception on context priors. Extensive\nexperiments manifest that the proposed DUN-CSNet outperforms existing\nstate-of-the-art CS methods by large margins.\n", "date": "2023-10-16", "categories": []}
{"id": "2310.10044", "title": "Hyperspectral Image Fusion via Logarithmic Low-rank Tensor Ring\n  Decomposition", "abstract": "  Integrating a low-spatial-resolution hyperspectral image (LR-HSI) with a\nhigh-spatial-resolution multispectral image (HR-MSI) is recognized as a valid\nmethod for acquiring HR-HSI. Among the current fusion approaches, the tensor\nring (TR) decomposition-based method has received growing attention owing to\nits superior performance on preserving the spatial-spectral correlation.\nFurthermore, the low-rank property in some TR factors has been exploited via\nthe matrix nuclear norm regularization along mode-2. On the other hand, the\ntensor nuclear norm (TNN)-based approaches have recently demonstrated to be\nmore efficient on keeping high-dimensional low-rank structures in tensor\nrecovery. Here, we study the low-rankness of TR factors from the TNN\nperspective and consider the mode-2 logarithmic TNN (LTNN) on each TR factor. A\nnovel fusion model is proposed by incorporating this LTNN regularization and\nthe weighted total variation which is to promote the continuity of HR-HSI in\nthe spatial-spectral domain. Meanwhile, we have devised a highly efficient\nproximal alternating minimization algorithm to solve the proposed model. The\nexperimental results indicate that our method improves the visual quality and\nexceeds the existing state-of-the-art fusion approaches with respect to various\nquantitative metrics.\n", "date": "2023-10-16", "categories": []}
{"id": "2310.10088", "title": "PUCA: Patch-Unshuffle and Channel Attention for Enhanced Self-Supervised\n  Image Denoising", "abstract": "  Although supervised image denoising networks have shown remarkable\nperformance on synthesized noisy images, they often fail in practice due to the\ndifference between real and synthesized noise. Since clean-noisy image pairs\nfrom the real world are extremely costly to gather, self-supervised learning,\nwhich utilizes noisy input itself as a target, has been studied. To prevent a\nself-supervised denoising model from learning identical mapping, each output\npixel should not be influenced by its corresponding input pixel; This\nrequirement is known as J-invariance. Blind-spot networks (BSNs) have been a\nprevalent choice to ensure J-invariance in self-supervised image denoising.\nHowever, constructing variations of BSNs by injecting additional operations\nsuch as downsampling can expose blinded information, thereby violating\nJ-invariance. Consequently, convolutions designed specifically for BSNs have\nbeen allowed only, limiting architectural flexibility. To overcome this\nlimitation, we propose PUCA, a novel J-invariant U-Net architecture, for\nself-supervised denoising. PUCA leverages patch-unshuffle/shuffle to\ndramatically expand receptive fields while maintaining J-invariance and dilated\nattention blocks (DABs) for global context incorporation. Experimental results\ndemonstrate that PUCA achieves state-of-the-art performance, outperforming\nexisting methods in self-supervised image denoising.\n", "date": "2023-10-16", "categories": []}
{"id": "2310.08956", "title": "LRRU: Long-short Range Recurrent Updating Networks for Depth Completion", "abstract": "  Existing deep learning-based depth completion methods generally employ\nmassive stacked layers to predict the dense depth map from sparse input data.\nAlthough such approaches greatly advance this task, their accompanied huge\ncomputational complexity hinders their practical applications. To accomplish\ndepth completion more efficiently, we propose a novel lightweight deep network\nframework, the Long-short Range Recurrent Updating (LRRU) network. Without\nlearning complex feature representations, LRRU first roughly fills the sparse\ninput to obtain an initial dense depth map, and then iteratively updates it\nthrough learned spatially-variant kernels. Our iterative update process is\ncontent-adaptive and highly flexible, where the kernel weights are learned by\njointly considering the guidance RGB images and the depth map to be updated,\nand large-to-small kernel scopes are dynamically adjusted to capture\nlong-to-short range dependencies. Our initial depth map has coarse but complete\nscene depth information, which helps relieve the burden of directly regressing\nthe dense depth from sparse ones, while our proposed method can effectively\nrefine it to an accurate depth map with less learnable parameters and inference\ntime. Experimental results demonstrate that our proposed LRRU variants achieve\nstate-of-the-art performance across different parameter regimes. In particular,\nthe LRRU-Base model outperforms competing approaches on the NYUv2 dataset, and\nranks 1st on the KITTI depth completion benchmark at the time of submission.\nProject page: https://npucvr.github.io/LRRU/.\n", "date": "2023-10-13", "categories": []}
{"id": "2306.11014", "title": "Physics Constrained Unsupervised Deep Learning for Rapid, High\n  Resolution Scanning Coherent Diffraction Reconstruction", "abstract": "  By circumventing the resolution limitations of optics, coherent diffractive\nimaging (CDI) and ptychography are making their way into scientific fields\nranging from X-ray imaging to astronomy. Yet, the need for time consuming\niterative phase recovery hampers real-time imaging. While supervised deep\nlearning strategies have increased reconstruction speed, they sacrifice image\nquality. Furthermore, these methods' demand for extensive labeled training data\nis experimentally burdensome. Here, we propose an unsupervised physics-informed\nneural network reconstruction method, PtychoPINN, that retains the factor of\n100-to-1000 speedup of deep learning-based reconstruction while improving\nreconstruction quality by combining the diffraction forward map with real-space\nconstraints from overlapping measurements. In particular, PtychoPINN\nsignificantly advances generalizability, accuracy (with a typical 10 dB PSNR\nincrease), and linear resolution (2- to 6-fold gain). This blend of performance\nand speed offers exciting prospects for high-resolution real-time imaging in\nhigh-throughput environments such as X-ray free electron lasers (XFELs) and\ndiffraction-limited light sources.\n", "date": "2023-06-19", "categories": []}
{"id": "2308.03807", "title": "Nest-DGIL: Nesterov-optimized Deep Geometric Incremental Learning for CS\n  Image Reconstruction", "abstract": "  Proximal gradient-based optimization is one of the most common strategies to\nsolve inverse problem of images, and it is easy to implement. However, these\ntechniques often generate heavy artifacts in image reconstruction. One of the\nmost popular refinement methods is to fine-tune the regularization parameter to\nalleviate such artifacts, but it may not always be sufficient or applicable due\nto increased computational costs. In this work, we propose a deep geometric\nincremental learning framework based on the second Nesterov proximal gradient\noptimization. The proposed end-to-end network not only has the powerful\nlearning ability for high-/low-frequency image features, but also can\ntheoretically guarantee that geometric texture details will be reconstructed\nfrom preliminary linear reconstruction. Furthermore, it can avoid the risk of\nintermediate reconstruction results falling outside the geometric decomposition\ndomains and achieve fast convergence. Our reconstruction framework is\ndecomposed into four modules including general linear reconstruction, cascade\ngeometric incremental restoration, Nesterov acceleration, and post-processing.\nIn the image restoration step, a cascade geometric incremental learning module\nis designed to compensate for missing texture information from different\ngeometric spectral decomposition domains. Inspired by the overlap-tile\nstrategy, we also develop a post-processing module to remove the block effect\nin patch-wise-based natural image reconstruction. All parameters in the\nproposed model are learnable, an adaptive initialization technique of physical\nparameters is also employed to make model flexibility and ensure converging\nsmoothly. We compare the reconstruction performance of the proposed method with\nexisting state-of-the-art methods to demonstrate its superiority. Our source\ncodes are available at https://github.com/fanxiaohong/Nest-DGIL.\n", "date": "2023-08-06", "categories": []}
{"id": "2310.08002", "title": "MLP-AMDC: An MLP Architecture for Adaptive-Mask-based Dual-Camera\n  snapshot hyperspectral imaging", "abstract": "  Coded Aperture Snapshot Spectral Imaging (CASSI) system has great advantages\nover traditional methods in dynamically acquiring Hyper-Spectral Image (HSI),\nbut there are the following problems. 1) Traditional mask relies on random\npatterns or analytical design, both of which limit the performance improvement\nof CASSI. 2) Existing high-quality reconstruction algorithms are slow in\nreconstruction and can only reconstruct scene information offline. To address\nthe above two problems, this paper designs the AMDC-CASSI system, introducing\nRGB camera with CASSI based on Adaptive-Mask as multimodal input to improve the\nreconstruction quality. The existing SOTA reconstruction schemes are based on\ntransformer, but the operation of self-attention pulls down the operation\nefficiency of the network. In order to improve the inference speed of the\nreconstruction network, this paper proposes An MLP Architecture for\nAdaptive-Mask-based Dual-Camera (MLP-AMDC) to replace the transformer structure\nof the network. Numerous experiments have shown that MLP performs no less well\nthan transformer-based structures for HSI reconstruction, while MLP greatly\nimproves the network inference speed and has less number of parameters and\noperations, our method has a 8 db improvement over SOTA and at least a 5-fold\nimprovement in reconstruction speed. (https://github.com/caizeyu1992/MLP-AMDC.)\n", "date": "2023-10-12", "categories": []}
{"id": "2303.16254", "title": "CryoFormer: Continuous Heterogeneous Cryo-EM Reconstruction using\n  Transformer-based Neural Representations", "abstract": "  Cryo-electron microscopy (cryo-EM) allows for the high-resolution\nreconstruction of 3D structures of proteins and other biomolecules. Successful\nreconstruction of both shape and movement greatly helps understand the\nfundamental processes of life. However, it is still challenging to reconstruct\nthe continuous motions of 3D structures from hundreds of thousands of noisy and\nrandomly oriented 2D cryo-EM images. Recent advancements use Fourier domain\ncoordinate-based neural networks to continuously model 3D conformations, yet\nthey often struggle to capture local flexible regions accurately. We propose\nCryoFormer, a new approach for continuous heterogeneous cryo-EM reconstruction.\nOur approach leverages an implicit feature volume directly in the real domain\nas the 3D representation. We further introduce a novel query-based deformation\ntransformer decoder to improve the reconstruction quality. Our approach is\ncapable of refining pre-computed pose estimations and locating flexible\nregions. In experiments, our method outperforms current approaches on three\npublic datasets (1 synthetic and 2 experimental) and a new synthetic dataset of\nPEDV spike protein. The code and new synthetic dataset will be released for\nbetter reproducibility of our results. Project page:\nhttps://cryoformer.github.io.\n", "date": "2023-03-28", "categories": []}
{"id": "2310.07255", "title": "ADASR: An Adversarial Auto-Augmentation Framework for Hyperspectral and\n  Multispectral Data Fusion", "abstract": "  Deep learning-based hyperspectral image (HSI) super-resolution, which aims to\ngenerate high spatial resolution HSI (HR-HSI) by fusing hyperspectral image\n(HSI) and multispectral image (MSI) with deep neural networks (DNNs), has\nattracted lots of attention. However, neural networks require large amounts of\ntraining data, hindering their application in real-world scenarios. In this\nletter, we propose a novel adversarial automatic data augmentation framework\nADASR that automatically optimizes and augments HSI-MSI sample pairs to enrich\ndata diversity for HSI-MSI fusion. Our framework is sample-aware and optimizes\nan augmentor network and two downsampling networks jointly by adversarial\nlearning so that we can learn more robust downsampling networks for training\nthe upsampling network. Extensive experiments on two public classical\nhyperspectral datasets demonstrate the effectiveness of our ADASR compared to\nthe state-of-the-art methods.\n", "date": "2023-10-11", "categories": []}
{"id": "2310.05158", "title": "ITRE: Low-light Image Enhancement Based on Illumination Transmission\n  Ratio Estimation", "abstract": "  Noise, artifacts, and over-exposure are significant challenges in the field\nof low-light image enhancement. Existing methods often struggle to address\nthese issues simultaneously. In this paper, we propose a novel Retinex-based\nmethod, called ITRE, which suppresses noise and artifacts from the origin of\nthe model, prevents over-exposure throughout the enhancement process.\nSpecifically, we assume that there must exist a pixel which is least disturbed\nby low light within pixels of same color. First, clustering the pixels on the\nRGB color space to find the Illumination Transmission Ratio (ITR) matrix of the\nwhole image, which determines that noise is not over-amplified easily. Next, we\nconsider ITR of the image as the initial illumination transmission map to\nconstruct a base model for refined transmission map, which prevents artifacts.\nAdditionally, we design an over-exposure module that captures the fundamental\ncharacteristics of pixel over-exposure and seamlessly integrate it into the\nbase model. Finally, there is a possibility of weak enhancement when\ninter-class distance of pixels with same color is too small. To counteract\nthis, we design a Robust-Guard module that safeguards the robustness of the\nimage enhancement process. Extensive experiments demonstrate the effectiveness\nof our approach in suppressing noise, preventing artifacts, and controlling\nover-exposure level simultaneously. Our method performs superiority in\nqualitative and quantitative performance evaluations by comparing with\nstate-of-the-art methods.\n", "date": "2023-10-08", "categories": []}
{"id": "2310.04299", "title": "Convergent ADMM Plug and Play PET Image Reconstruction", "abstract": "  In this work, we investigate hybrid PET reconstruction algorithms based on\ncoupling a model-based variational reconstruction and the application of a\nseparately learnt Deep Neural Network operator (DNN) in an ADMM Plug and Play\nframework. Following recent results in optimization, fixed point convergence of\nthe scheme can be achieved by enforcing an additional constraint on network\nparameters during learning. We propose such an ADMM algorithm and show in a\nrealistic [18F]-FDG synthetic brain exam that the proposed scheme indeed lead\nexperimentally to convergence to a meaningful fixed point. When the proposed\nconstraint is not enforced during learning of the DNN, the proposed ADMM\nalgorithm was observed experimentally not to converge.\n", "date": "2023-10-06", "categories": []}
{"id": "2310.02340", "title": "Learning Interpretable Deep Disentangled Neural Networks for\n  Hyperspectral Unmixing", "abstract": "  Although considerable effort has been dedicated to improving the solution to\nthe hyperspectral unmixing problem, non-idealities such as complex radiation\nscattering and endmember variability negatively impact the performance of most\nexisting algorithms and can be very challenging to address. Recently, deep\nlearning-based frameworks have been explored for hyperspectral umixing due to\ntheir flexibility and powerful representation capabilities. However, such\ntechniques either do not address the non-idealities of the unmixing problem, or\nrely on black-box models which are not interpretable. In this paper, we propose\na new interpretable deep learning method for hyperspectral unmixing that\naccounts for nonlinearity and endmember variability. The proposed method\nleverages a probabilistic variational deep-learning framework, where\ndisentanglement learning is employed to properly separate the abundances and\nendmembers. The model is learned end-to-end using stochastic backpropagation,\nand trained using a self-supervised strategy which leverages benefits from\nsemi-supervised learning techniques. Furthermore, the model is carefully\ndesigned to provide a high degree of interpretability. This includes modeling\nthe abundances as a Dirichlet distribution, the endmembers using\nlow-dimensional deep latent variable representations, and using two-stream\nneural networks composed of additive piecewise-linear/nonlinear components.\nExperimental results on synthetic and real datasets illustrate the performance\nof the proposed method compared to state-of-the-art algorithms.\n", "date": "2023-10-03", "categories": []}
{"id": "2309.03335", "title": "SADIR: Shape-Aware Diffusion Models for 3D Image Reconstruction", "abstract": "  3D image reconstruction from a limited number of 2D images has been a\nlong-standing challenge in computer vision and image analysis. While deep\nlearning-based approaches have achieved impressive performance in this area,\nexisting deep networks often fail to effectively utilize the shape structures\nof objects presented in images. As a result, the topology of reconstructed\nobjects may not be well preserved, leading to the presence of artifacts such as\ndiscontinuities, holes, or mismatched connections between different parts. In\nthis paper, we propose a shape-aware network based on diffusion models for 3D\nimage reconstruction, named SADIR, to address these issues. In contrast to\nprevious methods that primarily rely on spatial correlations of image\nintensities for 3D reconstruction, our model leverages shape priors learned\nfrom the training data to guide the reconstruction process. To achieve this, we\ndevelop a joint learning network that simultaneously learns a mean shape under\ndeformation models. Each reconstructed image is then considered as a deformed\nvariant of the mean shape. We validate our model, SADIR, on both brain and\ncardiac magnetic resonance images (MRIs). Experimental results show that our\nmethod outperforms the baselines with lower reconstruction error and better\npreservation of the shape structure of objects within the images.\n", "date": "2023-09-06", "categories": []}
{"id": "2306.15203", "title": "Unsupervised Polychromatic Neural Representation for CT Metal Artifact\n  Reduction", "abstract": "  Emerging neural reconstruction techniques based on tomography (e.g., NeRF,\nNeAT, and NeRP) have started showing unique capabilities in medical imaging. In\nthis work, we present a novel Polychromatic neural representation (Polyner) to\ntackle the challenging problem of CT imaging when metallic implants exist\nwithin the human body. CT metal artifacts arise from the drastic variation of\nmetal's attenuation coefficients at various energy levels of the X-ray\nspectrum, leading to a nonlinear metal effect in CT measurements. Recovering CT\nimages from metal-affected measurements hence poses a complicated nonlinear\ninverse problem where empirical models adopted in previous metal artifact\nreduction (MAR) approaches lead to signal loss and strongly aliased\nreconstructions. Polyner instead models the MAR problem from a nonlinear\ninverse problem perspective. Specifically, we first derive a polychromatic\nforward model to accurately simulate the nonlinear CT acquisition process.\nThen, we incorporate our forward model into the implicit neural representation\nto accomplish reconstruction. Lastly, we adopt a regularizer to preserve the\nphysical properties of the CT images across different energy levels while\neffectively constraining the solution space. Our Polyner is an unsupervised\nmethod and does not require any external training data. Experimenting with\nmultiple datasets shows that our Polyner achieves comparable or better\nperformance than supervised methods on in-domain datasets while demonstrating\nsignificant performance improvements on out-of-domain datasets. To the best of\nour knowledge, our Polyner is the first unsupervised MAR method that\noutperforms its supervised counterparts. The code for this work is available\nat: https://github.com/iwuqing/Polyner.\n", "date": "2023-06-27", "categories": []}
{"id": "2307.05890", "title": "FreeSeed: Frequency-band-aware and Self-guided Network for Sparse-view\n  CT Reconstruction", "abstract": "  Sparse-view computed tomography (CT) is a promising solution for expediting\nthe scanning process and mitigating radiation exposure to patients, the\nreconstructed images, however, contain severe streak artifacts, compromising\nsubsequent screening and diagnosis. Recently, deep learning-based image\npost-processing methods along with their dual-domain counterparts have shown\npromising results. However, existing methods usually produce over-smoothed\nimages with loss of details due to (1) the difficulty in accurately modeling\nthe artifact patterns in the image domain, and (2) the equal treatment of each\npixel in the loss function. To address these issues, we concentrate on the\nimage post-processing and propose a simple yet effective FREquency-band-awarE\nand SElf-guidED network, termed FreeSeed, which can effectively remove artifact\nand recover missing detail from the contaminated sparse-view CT images.\nSpecifically, we first propose a frequency-band-aware artifact modeling network\n(FreeNet), which learns artifact-related frequency-band attention in Fourier\ndomain for better modeling the globally distributed streak artifact on the\nsparse-view CT images. We then introduce a self-guided artifact refinement\nnetwork (SeedNet), which leverages the predicted artifact to assist FreeNet in\ncontinuing to refine the severely corrupted details. Extensive experiments\ndemonstrate the superior performance of FreeSeed and its dual-domain\ncounterpart over the state-of-the-art sparse-view CT reconstruction methods.\nSource code is made available at https://github.com/Masaaki-75/freeseed.\n", "date": "2023-07-12", "categories": []}
{"id": "2310.00133", "title": "Prior Mismatch and Adaptation in PnP-ADMM with a Nonconvex Convergence\n  Analysis", "abstract": "  Plug-and-Play (PnP) priors is a widely-used family of methods for solving\nimaging inverse problems by integrating physical measurement models with image\npriors specified using image denoisers. PnP methods have been shown to achieve\nstate-of-the-art performance when the prior is obtained using powerful deep\ndenoisers. Despite extensive work on PnP, the topic of distribution mismatch\nbetween the training and testing data has often been overlooked in the PnP\nliterature. This paper presents a set of new theoretical and numerical results\non the topic of prior distribution mismatch and domain adaptation for\nalternating direction method of multipliers (ADMM) variant of PnP. Our\ntheoretical result provides an explicit error bound for PnP-ADMM due to the\nmismatch between the desired denoiser and the one used for inference. Our\nanalysis contributes to the work in the area by considering the mismatch under\nnonconvex data-fidelity terms and expansive denoisers. Our first set of\nnumerical results quantifies the impact of the prior distribution mismatch on\nthe performance of PnP-ADMM on the problem of image super-resolution. Our\nsecond set of numerical results considers a simple and effective domain\nadaption strategy that closes the performance gap due to the use of mismatched\ndenoisers. Our results suggest the relative robustness of PnP-ADMM to prior\ndistribution mismatch, while also showing that the performance gap can be\nsignificantly reduced with few training samples from the desired distribution.\n", "date": "2023-09-29", "categories": []}
{"id": "2310.01110", "title": "Prompt-tuning latent diffusion models for inverse problems", "abstract": "  We propose a new method for solving imaging inverse problems using\ntext-to-image latent diffusion models as general priors. Existing methods using\nlatent diffusion models for inverse problems typically rely on simple null text\nprompts, which can lead to suboptimal performance. To address this limitation,\nwe introduce a method for prompt tuning, which jointly optimizes the text\nembedding on-the-fly while running the reverse diffusion process. This allows\nus to generate images that are more faithful to the diffusion prior. In\naddition, we propose a method to keep the evolution of latent variables within\nthe range space of the encoder, by projection. This helps to reduce image\nartifacts, a major problem when using latent diffusion models instead of\npixel-based diffusion models. Our combined method, called P2L, outperforms both\nimage- and latent-diffusion model-based inverse problem solvers on a variety of\ntasks, such as super-resolution, deblurring, and inpainting.\n", "date": "2023-10-02", "categories": []}
{"id": "2309.16677", "title": "Mechanical Artifacts in Optical Projection Tomography: Classification\n  and Automatic Calibration", "abstract": "  Optical projection tomography (OPT) is a powerful tool for biomedical\nstudies. It achieves 3D visualization of mesoscopic biological samples with\nhigh spatial resolution using conventional tomographic-reconstruction\nalgorithms. However, various artifacts degrade the quality of the reconstructed\nimages due to experimental imperfections in the OPT instruments. While many\nefforts have been made to characterize and correct for these artifacts, they\nfocus on one specific type of artifacts. This work has two contributions.\nFirst, we systematically document a catalog of mechanical artifacts based on a\n3D description of the imaging system that uses a set of angular and\ntranslational parameters. Then, we introduce a calibration algorithm that\nrecovers the unknown system parameters fed into the final 3D iterative\nreconstruction algorithm for a distortion-free volumetric image. Simulations\nwith beads data and experimental results on a fluorescent textile fiber confirm\nthat our algorithm successfully removes miscalibration artifacts in the\nreconstruction.\n", "date": "2023-07-19", "categories": []}
{"id": "2309.16949", "title": "CrossZoom: Simultaneously Motion Deblurring and Event Super-Resolving", "abstract": "  Even though the collaboration between traditional and neuromorphic event\ncameras brings prosperity to frame-event based vision applications, the\nperformance is still confined by the resolution gap crossing two modalities in\nboth spatial and temporal domains. This paper is devoted to bridging the gap by\nincreasing the temporal resolution for images, i.e., motion deblurring, and the\nspatial resolution for events, i.e., event super-resolving, respectively. To\nthis end, we introduce CrossZoom, a novel unified neural Network (CZ-Net) to\njointly recover sharp latent sequences within the exposure period of a blurry\ninput and the corresponding High-Resolution (HR) events. Specifically, we\npresent a multi-scale blur-event fusion architecture that leverages the\nscale-variant properties and effectively fuses cross-modality information to\nachieve cross-enhancement. Attention-based adaptive enhancement and\ncross-interaction prediction modules are devised to alleviate the distortions\ninherent in Low-Resolution (LR) events and enhance the final results through\nthe prior blur-event complementary information. Furthermore, we propose a new\ndataset containing HR sharp-blurry images and the corresponding HR-LR event\nstreams to facilitate future research. Extensive qualitative and quantitative\nexperiments on synthetic and real-world datasets demonstrate the effectiveness\nand robustness of the proposed method. Codes and datasets are released at\nhttps://bestrivenzc.github.io/CZ-Net/.\n", "date": "2023-09-29", "categories": []}
{"id": "2309.16975", "title": "Perceptual Tone Mapping Model for High Dynamic Range Imaging", "abstract": "  One of the key challenges in tone mapping is to preserve the perceptual\nquality of high dynamic range (HDR) images when mapping them to standard\ndynamic range (SDR) displays. Traditional tone mapping operators (TMOs)\ncompress the luminance of HDR images without considering the surround and\ndisplay conditions emanating into suboptimal results. Current research\naddresses this challenge by incorporating perceptual color appearance\nattributes. In this work, we propose a TMO (TMOz) that leverages CIECAM16\nperceptual attributes, i.e., brightness, colorfulness, and hue. TMOz accounts\nfor the effects of both the surround and the display conditions to achieve more\noptimal colorfulness reproduction. The perceptual brightness is compressed, and\nthe perceptual color scales, i.e., colorfulness and hue are derived from HDR\nimages by employing CIECAM16 color adaptation equations. A psychophysical\nexperiment was conducted to automate the brightness compression parameter. The\nmodel employs fully automatic and adaptive approach, obviating the requirement\nfor manual parameter selection. TMOz was evaluated in terms of contrast,\ncolorfulness and overall image quality. The objective and subjective evaluation\nmethods revealed that the proposed model outperformed the state-of-the-art\nTMOs.\n", "date": "2023-09-29", "categories": []}
{"id": "2309.16372", "title": "Aperture Diffraction for Compact Snapshot Spectral Imaging", "abstract": "  We demonstrate a compact, cost-effective snapshot spectral imaging system\nnamed Aperture Diffraction Imaging Spectrometer (ADIS), which consists only of\nan imaging lens with an ultra-thin orthogonal aperture mask and a mosaic filter\nsensor, requiring no additional physical footprint compared to common RGB\ncameras. Then we introduce a new optical design that each point in the object\nspace is multiplexed to discrete encoding locations on the mosaic filter sensor\nby diffraction-based spatial-spectral projection engineering generated from the\northogonal mask. The orthogonal projection is uniformly accepted to obtain a\nweakly calibration-dependent data form to enhance modulation robustness.\nMeanwhile, the Cascade Shift-Shuffle Spectral Transformer (CSST) with strong\nperception of the diffraction degeneration is designed to solve a\nsparsity-constrained inverse problem, realizing the volume reconstruction from\n2D measurements with Large amount of aliasing. Our system is evaluated by\nelaborating the imaging optical theory and reconstruction algorithm with\ndemonstrating the experimental imaging under a single exposure. Ultimately, we\nachieve the sub-super-pixel spatial resolution and high spectral resolution\nimaging. The code will be available at: https://github.com/Krito-ex/CSST.\n", "date": "2023-09-27", "categories": []}
{"id": "2210.16060", "title": "Deep network series for large-scale high-dynamic range imaging", "abstract": "  We propose a new approach for large-scale high-dynamic range computational\nimaging. Deep Neural Networks (DNNs) trained end-to-end can solve linear\ninverse imaging problems almost instantaneously. While unfolded architectures\nprovide robustness to measurement setting variations, embedding large-scale\nmeasurement operators in DNN architectures is impractical. Alternative\nPlug-and-Play (PnP) approaches, where the denoising DNNs are blind to the\nmeasurement setting, have proven effective to address scalability and\nhigh-dynamic range challenges, but rely on highly iterative algorithms. We\npropose a residual DNN series approach, also interpretable as a learned version\nof matching pursuit, where the reconstructed image is a sum of residual images\nprogressively increasing the dynamic range, and estimated iteratively by DNNs\ntaking the back-projected data residual of the previous iteration as input. We\ndemonstrate on radio-astronomical imaging simulations that a series of only few\nterms provides a reconstruction quality competitive with PnP, at a fraction of\nthe cost.\n", "date": "2022-10-28", "categories": []}
{"id": "2309.12047", "title": "Self-Calibrating, Fully Differentiable NLOS Inverse Rendering", "abstract": "  Existing time-resolved non-line-of-sight (NLOS) imaging methods reconstruct\nhidden scenes by inverting the optical paths of indirect illumination measured\nat visible relay surfaces. These methods are prone to reconstruction artifacts\ndue to inversion ambiguities and capture noise, which are typically mitigated\nthrough the manual selection of filtering functions and parameters. We\nintroduce a fully-differentiable end-to-end NLOS inverse rendering pipeline\nthat self-calibrates the imaging parameters during the reconstruction of hidden\nscenes, using as input only the measured illumination while working both in the\ntime and frequency domains. Our pipeline extracts a geometric representation of\nthe hidden scene from NLOS volumetric intensities and estimates the\ntime-resolved illumination at the relay wall produced by such geometric\ninformation using differentiable transient rendering. We then use gradient\ndescent to optimize imaging parameters by minimizing the error between our\nsimulated time-resolved illumination and the measured illumination. Our\nend-to-end differentiable pipeline couples diffraction-based volumetric NLOS\nreconstruction with path-space light transport and a simple ray marching\ntechnique to extract detailed, dense sets of surface points and normals of\nhidden scenes. We demonstrate the robustness of our method to consistently\nreconstruct geometry and albedo, even under significant noise levels.\n", "date": "2023-09-21", "categories": []}
{"id": "2309.14755", "title": "Image Denoising via Style Disentanglement", "abstract": "  Image denoising is a fundamental task in low-level computer vision. While\nrecent deep learning-based image denoising methods have achieved impressive\nperformance, they are black-box models and the underlying denoising principle\nremains unclear. In this paper, we propose a novel approach to image denoising\nthat offers both clear denoising mechanism and good performance. We view noise\nas a type of image style and remove it by incorporating noise-free styles\nderived from clean images. To achieve this, we design novel losses and network\nmodules to extract noisy styles from noisy images and noise-free styles from\nclean images. The noise-free style induces low-response activations for noise\nfeatures and high-response activations for content features in the feature\nspace. This leads to the separation of clean contents from noise, effectively\ndenoising the image. Unlike disentanglement-based image editing tasks that edit\nsemantic-level attributes using styles, our main contribution lies in editing\npixel-level attributes through global noise-free styles. We conduct extensive\nexperiments on synthetic noise removal and real-world image denoising datasets\n(SIDD and DND), demonstrating the effectiveness of our method in terms of both\nPSNR and SSIM metrics. Moreover, we experimentally validate that our method\noffers good interpretability.\n", "date": "2023-09-26", "categories": []}
{"id": "2309.14954", "title": "Addressing preferred orientation in single-particle cryo-EM through\n  AI-generated auxiliary particles", "abstract": "  The single-particle cryo-EM field faces the persistent challenge of preferred\norientation, lacking general computational solutions. We introduce cryoPROS, an\nAI-based approach designed to address the above issue. By generating the\nauxiliary particles with a conditional deep generative model, cryoPROS\naddresses the intrinsic bias in orientation estimation for the observed\nparticles. We effectively employed cryoPROS in the cryo-EM single particle\nanalysis of the hemagglutinin trimer, showing the ability to restore the\nnear-atomic resolution structure on non-tilt data. Moreover, the enhanced\nversion named cryoPROS-MP significantly improves the resolution of the membrane\nprotein NaX using the no-tilted data that contains the effects of micelles.\nCompared to the classical approaches, cryoPROS does not need special\nexperimental or image acquisition techniques, providing a purely computational\nyet effective solution for the preferred orientation problem. Finally, we\nconduct extensive experiments that establish the low risk of model bias and the\nhigh robustness of cryoPROS.\n", "date": "2023-09-26", "categories": []}
{"id": "2309.14967", "title": "A novel approach for holographic 3D content generation without depth map", "abstract": "  In preparation for observing holographic 3D content, acquiring a set of RGB\ncolor and depth map images per scene is necessary to generate\ncomputer-generated holograms (CGHs) when using the fast Fourier transform (FFT)\nalgorithm. However, in real-world situations, these paired formats of RGB color\nand depth map images are not always fully available. We propose a deep\nlearning-based method to synthesize the volumetric digital holograms using only\nthe given RGB image, so that we can overcome environments where RGB color and\ndepth map images are partially provided. The proposed method uses only the\ninput of RGB image to estimate its depth map and then generate its CGH\nsequentially. Through experiments, we demonstrate that the volumetric hologram\ngenerated through our proposed model is more accurate than that of competitive\nmodels, under the situation that only RGB color data can be provided.\n", "date": "2023-09-26", "categories": []}
{"id": "2308.10510", "title": "Frequency Compensated Diffusion Model for Real-scene Dehazing", "abstract": "  Due to distribution shift, deep learning based methods for image dehazing\nsuffer from performance degradation when applied to real-world hazy images. In\nthis paper, we consider a dehazing framework based on conditional diffusion\nmodels for improved generalization to real haze. First, we find that optimizing\nthe training objective of diffusion models, i.e., Gaussian noise vectors, is\nnon-trivial. The spectral bias of deep networks hinders the higher frequency\nmodes in Gaussian vectors from being learned and hence impairs the\nreconstruction of image details. To tackle this issue, we design a network\nunit, named Frequency Compensation block (FCB), with a bank of filters that\njointly emphasize the mid-to-high frequencies of an input signal. We\ndemonstrate that diffusion models with FCB achieve significant gains in both\nperceptual and distortion metrics. Second, to further boost the generalization\nperformance, we propose a novel data synthesis pipeline, HazeAug, to augment\nhaze in terms of degree and diversity. Within the framework, a solid baseline\nfor blind dehazing is set up where models are trained on synthetic hazy-clean\npairs, and directly generalize to real data. Extensive evaluations show that\nthe proposed dehazing diffusion model significantly outperforms\nstate-of-the-art methods on real-world images. Our code is at\nhttps://github.com/W-Jilly/frequency-compensated-diffusion-model-pytorch.\n", "date": "2023-08-21", "categories": []}
{"id": "2309.11858", "title": "OSNet & MNetO: Two Types of General Reconstruction Architectures for\n  Linear Computed Tomography in Multi-Scenarios", "abstract": "  Recently, linear computed tomography (LCT) systems have actively attracted\nattention. To weaken projection truncation and image the region of interest\n(ROI) for LCT, the backprojection filtration (BPF) algorithm is an effective\nsolution. However, in BPF for LCT, it is difficult to achieve stable interior\nreconstruction, and for differentiated backprojection (DBP) images of LCT,\nmultiple rotation-finite inversion of Hilbert transform (Hilbert\nfiltering)-inverse rotation operations will blur the image. To satisfy multiple\nreconstruction scenarios for LCT, including interior ROI, complete object, and\nexterior region beyond field-of-view (FOV), and avoid the rotation operations\nof Hilbert filtering, we propose two types of reconstruction architectures. The\nfirst overlays multiple DBP images to obtain a complete DBP image, then uses a\nnetwork to learn the overlying Hilbert filtering function, referred to as the\nOverlay-Single Network (OSNet). The second uses multiple networks to train\ndifferent directional Hilbert filtering models for DBP images of multiple\nlinear scannings, respectively, and then overlays the reconstructed results,\ni.e., Multiple Networks Overlaying (MNetO). In two architectures, we introduce\na Swin Transformer (ST) block to the generator of pix2pixGAN to extract both\nlocal and global features from DBP images at the same time. We investigate two\narchitectures from different networks, FOV sizes, pixel sizes, number of\nprojections, geometric magnification, and processing time. Experimental results\nshow that two architectures can both recover images. OSNet outperforms BPF in\nvarious scenarios. For the different networks, ST-pix2pixGAN is superior to\npix2pixGAN and CycleGAN. MNetO exhibits a few artifacts due to the differences\namong the multiple models, but any one of its models is suitable for imaging\nthe exterior edge in a certain direction.\n", "date": "2023-09-21", "categories": []}
{"id": "2309.13571", "title": "Matrix Completion-Informed Deep Unfolded Equilibrium Models for\n  Self-Supervised k-Space Interpolation in MRI", "abstract": "  Recently, regularization model-driven deep learning (DL) has gained\nsignificant attention due to its ability to leverage the potent\nrepresentational capabilities of DL while retaining the theoretical guarantees\nof regularization models. However, most of these methods are tailored for\nsupervised learning scenarios that necessitate fully sampled labels, which can\npose challenges in practical MRI applications. To tackle this challenge, we\npropose a self-supervised DL approach for accelerated MRI that is theoretically\nguaranteed and does not rely on fully sampled labels. Specifically, we achieve\nneural network structure regularization by exploiting the inherent structural\nlow-rankness of the $k$-space data. Simultaneously, we constrain the network\nstructure to resemble a nonexpansive mapping, ensuring the network's\nconvergence to a fixed point. Thanks to this well-defined network structure,\nthis fixed point can completely reconstruct the missing $k$-space data based on\nmatrix completion theory, even in situations where full-sampled labels are\nunavailable. Experiments validate the effectiveness of our proposed method and\ndemonstrate its superiority over existing self-supervised approaches and\ntraditional regularization methods, achieving performance comparable to that of\nsupervised learning methods in certain scenarios.\n", "date": "2023-09-24", "categories": []}
{"id": "2309.13851", "title": "DISeR: Designing Imaging Systems with Reinforcement Learning", "abstract": "  Imaging systems consist of cameras to encode visual information about the\nworld and perception models to interpret this encoding. Cameras contain (1)\nillumination sources, (2) optical elements, and (3) sensors, while perception\nmodels use (4) algorithms. Directly searching over all combinations of these\nfour building blocks to design an imaging system is challenging due to the size\nof the search space. Moreover, cameras and perception models are often designed\nindependently, leading to sub-optimal task performance. In this paper, we\nformulate these four building blocks of imaging systems as a context-free\ngrammar (CFG), which can be automatically searched over with a learned camera\ndesigner to jointly optimize the imaging system with task-specific perception\nmodels. By transforming the CFG to a state-action space, we then show how the\ncamera designer can be implemented with reinforcement learning to intelligently\nsearch over the combinatorial space of possible imaging system configurations.\nWe demonstrate our approach on two tasks, depth estimation and camera rig\ndesign for autonomous vehicles, showing that our method yields rigs that\noutperform industry-wide standards. We believe that our proposed approach is an\nimportant step towards automating imaging system design.\n", "date": "2023-09-25", "categories": []}
{"id": "2308.10820", "title": "Pixel Adaptive Deep Unfolding Transformer for Hyperspectral Image\n  Reconstruction", "abstract": "  Hyperspectral Image (HSI) reconstruction has made gratifying progress with\nthe deep unfolding framework by formulating the problem into a data module and\na prior module. Nevertheless, existing methods still face the problem of\ninsufficient matching with HSI data. The issues lie in three aspects: 1) fixed\ngradient descent step in the data module while the degradation of HSI is\nagnostic in the pixel-level. 2) inadequate prior module for 3D HSI cube. 3)\nstage interaction ignoring the differences in features at different stages. To\naddress these issues, in this work, we propose a Pixel Adaptive Deep Unfolding\nTransformer (PADUT) for HSI reconstruction. In the data module, a pixel\nadaptive descent step is employed to focus on pixel-level agnostic degradation.\nIn the prior module, we introduce the Non-local Spectral Transformer (NST) to\nemphasize the 3D characteristics of HSI for recovering. Moreover, inspired by\nthe diverse expression of features in different stages and depths, the stage\ninteraction is improved by the Fast Fourier Transform (FFT). Experimental\nresults on both simulated and real scenes exhibit the superior performance of\nour method compared to state-of-the-art HSI reconstruction methods. The code is\nreleased at: https://github.com/MyuLi/PADUT.\n", "date": "2023-08-21", "categories": []}
{"id": "2305.07712", "title": "Poisson-Gaussian Holographic Phase Retrieval with Score-based Image\n  Prior", "abstract": "  Phase retrieval (PR) is a crucial problem in many imaging applications. This\nstudy focuses on resolving the holographic phase retrieval problem in\nsituations where the measurements are affected by a combination of Poisson and\nGaussian noise, which commonly occurs in optical imaging systems. To address\nthis problem, we propose a new algorithm called \"AWFS\" that uses the\naccelerated Wirtinger flow (AWF) with a score function as generative prior.\nSpecifically, we formulate the PR problem as an optimization problem that\nincorporates both data fidelity and regularization terms. We calculate the\ngradient of the log-likelihood function for PR and determine its corresponding\nLipschitz constant. Additionally, we introduce a generative prior in our\nregularization framework by using score matching to capture information about\nthe gradient of image prior distributions. We provide theoretical analysis that\nestablishes a critical-point convergence guarantee for the proposed algorithm.\nThe results of our simulation experiments on three different datasets show the\nfollowing: 1) By using the PG likelihood model, the proposed algorithm improves\nreconstruction compared to algorithms based solely on Gaussian or Poisson\nlikelihood. 2) The proposed score-based image prior method, performs better\nthan the method based on denoising diffusion probabilistic model (DDPM), as\nwell as plug-and-play alternating direction method of multipliers (PnP-ADMM)\nand regularization by denoising (RED).\n", "date": "2023-05-12", "categories": []}
{"id": "2309.11847", "title": "MEFLUT: Unsupervised 1D Lookup Tables for Multi-exposure Image Fusion", "abstract": "  In this paper, we introduce a new approach for high-quality multi-exposure\nimage fusion (MEF). We show that the fusion weights of an exposure can be\nencoded into a 1D lookup table (LUT), which takes pixel intensity value as\ninput and produces fusion weight as output. We learn one 1D LUT for each\nexposure, then all the pixels from different exposures can query 1D LUT of that\nexposure independently for high-quality and efficient fusion. Specifically, to\nlearn these 1D LUTs, we involve attention mechanism in various dimensions\nincluding frame, channel and spatial ones into the MEF task so as to bring us\nsignificant quality improvement over the state-of-the-art (SOTA). In addition,\nwe collect a new MEF dataset consisting of 960 samples, 155 of which are\nmanually tuned by professionals as ground-truth for evaluation. Our network is\ntrained by this dataset in an unsupervised manner. Extensive experiments are\nconducted to demonstrate the effectiveness of all the newly proposed\ncomponents, and results show that our approach outperforms the SOTA in our and\nanother representative dataset SICE, both qualitatively and quantitatively.\nMoreover, our 1D LUT approach takes less than 4ms to run a 4K image on a PC\nGPU. Given its high quality, efficiency and robustness, our method has been\nshipped into millions of Android mobiles across multiple brands world-wide.\nCode is available at: https://github.com/Hedlen/MEFLUT.\n", "date": "2023-09-21", "categories": []}
{"id": "2309.11122", "title": "Hyperspectral Benchmark: Bridging the Gap between HSI Applications\n  through Comprehensive Dataset and Pretraining", "abstract": "  Hyperspectral Imaging (HSI) serves as a non-destructive spatial spectroscopy\ntechnique with a multitude of potential applications. However, a recurring\nchallenge lies in the limited size of the target datasets, impeding exhaustive\narchitecture search. Consequently, when venturing into novel applications,\nreliance on established methodologies becomes commonplace, in the hope that\nthey exhibit favorable generalization characteristics. Regrettably, this\noptimism is often unfounded due to the fine-tuned nature of models tailored to\nspecific HSI contexts.\n  To address this predicament, this study introduces an innovative benchmark\ndataset encompassing three markedly distinct HSI applications: food inspection,\nremote sensing, and recycling. This comprehensive dataset affords a finer\nassessment of hyperspectral model capabilities. Moreover, this benchmark\nfacilitates an incisive examination of prevailing state-of-the-art techniques,\nconsequently fostering the evolution of superior methodologies.\n  Furthermore, the enhanced diversity inherent in the benchmark dataset\nunderpins the establishment of a pretraining pipeline for HSI. This pretraining\nregimen serves to enhance the stability of training processes for larger\nmodels. Additionally, a procedural framework is delineated, offering insights\ninto the handling of applications afflicted by limited target dataset sizes.\n", "date": "2023-09-20", "categories": []}
{"id": "2309.11421", "title": "CalibFPA: A Focal Plane Array Imaging System based on Online\n  Deep-Learning Calibration", "abstract": "  Compressive focal plane arrays (FPA) enable cost-effective high-resolution\n(HR) imaging by acquisition of several multiplexed measurements on a\nlow-resolution (LR) sensor. Multiplexed encoding of the visual scene is\ntypically performed via electronically controllable spatial light modulators\n(SLM). An HR image is then reconstructed from the encoded measurements by\nsolving an inverse problem that involves the forward model of the imaging\nsystem. To capture system non-idealities such as optical aberrations, a\nmainstream approach is to conduct an offline calibration scan to measure the\nsystem response for a point source at each spatial location on the imaging\ngrid. However, it is challenging to run calibration scans when using structured\nSLMs as they cannot encode individual grid locations. In this study, we propose\na novel compressive FPA system based on online deep-learning calibration of\nmultiplexed LR measurements (CalibFPA). We introduce a piezo-stage that\nlocomotes a pre-printed fixed coded aperture. A deep neural network is then\nleveraged to correct for the influences of system non-idealities in multiplexed\nmeasurements without the need for offline calibration scans. Finally, a deep\nplug-and-play algorithm is used to reconstruct images from corrected\nmeasurements. On simulated and experimental datasets, we demonstrate that\nCalibFPA outperforms state-of-the-art compressive FPA methods. We also report\nanalyses to validate the design elements in CalibFPA and assess computational\ncomplexity.\n", "date": "2023-09-20", "categories": []}
{"id": "2309.08838", "title": "AOSR-Net: All-in-One Sandstorm Removal Network", "abstract": "  Most existing sandstorm image enhancement methods are based on traditional\ntheory and prior knowledge, which often restrict their applicability in\nreal-world scenarios. In addition, these approaches often adopt a strategy of\ncolor correction followed by dust removal, which makes the algorithm structure\ntoo complex. To solve the issue, we introduce a novel image restoration model,\nnamed all-in-one sandstorm removal network (AOSR-Net). This model is developed\nbased on a re-formulated sandstorm scattering model, which directly establishes\nthe image mapping relationship by integrating intermediate parameters. Such\nintegration scheme effectively addresses the problems of over-enhancement and\nweak generalization in the field of sand dust image enhancement. Experimental\nresults on synthetic and real-world sandstorm images demonstrate the\nsuperiority of the proposed AOSR-Net over state-of-the-art (SOTA) algorithms.\n", "date": "2023-09-16", "categories": []}
{"id": "2309.08844", "title": "Emerging Approaches for THz Array Imaging: A Tutorial Review and\n  Software Tool", "abstract": "  Accelerated by the increasing attention drawn by 5G, 6G, and Internet of\nThings applications, communication and sensing technologies have rapidly\nevolved from millimeter-wave (mmWave) to terahertz (THz) in recent years.\nEnabled by significant advancements in electromagnetic (EM) hardware, mmWave\nand THz frequency regimes spanning 30 GHz to 300 GHz and 300 GHz to 3000 GHz,\nrespectively, can be employed for a host of applications. The main feature of\nTHz systems is high-bandwidth transmission, enabling ultra-high-resolution\nimaging and high-throughput communications; however, challenges in both the\nhardware and algorithmic arenas remain for the ubiquitous adoption of THz\ntechnology. Spectra comprising mmWave and THz frequencies are well-suited for\nsynthetic aperture radar (SAR) imaging at sub-millimeter resolutions for a wide\nspectrum of tasks like material characterization and nondestructive testing\n(NDT). This article provides a tutorial review of systems and algorithms for\nTHz SAR in the near-field with an emphasis on emerging algorithms that combine\nsignal processing and machine learning techniques. As part of this study, an\noverview of classical and data-driven THz SAR algorithms is provided, focusing\non object detection for security applications and SAR image super-resolution.\nWe also discuss relevant issues, challenges, and future research directions for\nemerging algorithms and THz SAR, including standardization of system and\nalgorithm benchmarking, adoption of state-of-the-art deep learning techniques,\nsignal processing-optimized machine learning, and hybrid data-driven signal\nprocessing algorithms...\n", "date": "2023-09-16", "categories": []}
{"id": "2309.09426", "title": "Joint Demosaicing and Denoising with Double Deep Image Priors", "abstract": "  Demosaicing and denoising of RAW images are crucial steps in the processing\npipeline of modern digital cameras. As only a third of the color information\nrequired to produce a digital image is captured by the camera sensor, the\nprocess of demosaicing is inherently ill-posed. The presence of noise further\nexacerbates this problem. Performing these two steps sequentially may distort\nthe content of the captured RAW images and accumulate errors from one step to\nanother. Recent deep neural-network-based approaches have shown the\neffectiveness of joint demosaicing and denoising to mitigate such challenges.\nHowever, these methods typically require a large number of training samples and\ndo not generalize well to different types and intensities of noise. In this\npaper, we propose a novel joint demosaicing and denoising method, dubbed\nJDD-DoubleDIP, which operates directly on a single RAW image without requiring\nany training data. We validate the effectiveness of our method on two popular\ndatasets -- Kodak and McMaster -- with various noises and noise intensities.\nThe experimental results show that our method consistently outperforms other\ncompared methods in terms of PSNR, SSIM, and qualitative visual perception.\n", "date": "2023-09-18", "categories": []}
{"id": "2309.09975", "title": "GEDepth: Ground Embedding for Monocular Depth Estimation", "abstract": "  Monocular depth estimation is an ill-posed problem as the same 2D image can\nbe projected from infinite 3D scenes. Although the leading algorithms in this\nfield have reported significant improvement, they are essentially geared to the\nparticular compound of pictorial observations and camera parameters (i.e.,\nintrinsics and extrinsics), strongly limiting their generalizability in\nreal-world scenarios. To cope with this challenge, this paper proposes a novel\nground embedding module to decouple camera parameters from pictorial cues, thus\npromoting the generalization capability. Given camera parameters, the proposed\nmodule generates the ground depth, which is stacked with the input image and\nreferenced in the final depth prediction. A ground attention is designed in the\nmodule to optimally combine ground depth with residual depth. Our ground\nembedding is highly flexible and lightweight, leading to a plug-in module that\nis amenable to be integrated into various depth estimation networks.\nExperiments reveal that our approach achieves the state-of-the-art results on\npopular benchmarks, and more importantly, renders significant generalization\nimprovement on a wide range of cross-domain tests.\n", "date": "2023-09-18", "categories": []}
{"id": "2208.09483", "title": "Blind Image Deblurring with Unknown Kernel Size and Substantial Noise", "abstract": "  Blind image deblurring (BID) has been extensively studied in computer vision\nand adjacent fields. Modern methods for BID can be grouped into two categories:\nsingle-instance methods that deal with individual instances using statistical\ninference and numerical optimization, and data-driven methods that train\ndeep-learning models to deblur future instances directly. Data-driven methods\ncan be free from the difficulty in deriving accurate blur models, but are\nfundamentally limited by the diversity and quality of the training data --\ncollecting sufficiently expressive and realistic training data is a standing\nchallenge. In this paper, we focus on single-instance methods that remain\ncompetitive and indispensable. However, most such methods do not prescribe how\nto deal with unknown kernel size and substantial noise, precluding practical\ndeployment. Indeed, we show that several state-of-the-art (SOTA)\nsingle-instance methods are unstable when the kernel size is overspecified,\nand/or the noise level is high. On the positive side, we propose a practical\nBID method that is stable against both, the first of its kind. Our method\nbuilds on the recent ideas of solving inverse problems by integrating the\nphysical models and structured deep neural networks, without extra training\ndata. We introduce several crucial modifications to achieve the desired\nstability. Extensive empirical tests on standard synthetic datasets, as well as\nreal-world NTIRE2020 and RealBlur datasets, show the superior effectiveness and\npracticality of our BID method compared to SOTA single-instance as well as\ndata-driven methods. The code of our method is available at:\n\\url{https://github.com/sun-umn/Blind-Image-Deblurring}.\n", "date": "2022-08-18", "categories": []}
{"id": "2309.08033", "title": "Depth Estimation from a Single Optical Encoded Image using a Learned\n  Colored-Coded Aperture", "abstract": "  Depth estimation from a single image of a conventional camera is a\nchallenging task since depth cues are lost during the acquisition process.\nState-of-the-art approaches improve the discrimination between different depths\nby introducing a binary-coded aperture (CA) in the lens aperture that generates\ndifferent coded blur patterns at different depths. Color-coded apertures (CCA)\ncan also produce color misalignment in the captured image which can be utilized\nto estimate disparity. Leveraging advances in deep learning, more recent works\nhave explored the data-driven design of a diffractive optical element (DOE) for\nencoding depth information through chromatic aberrations. However, compared\nwith binary CA or CCA, DOEs are more expensive to fabricate and require\nhigh-precision devices. Different from previous CCA-based approaches that\nemploy few basic colors, in this work we propose a CCA with a greater number of\ncolor filters and richer spectral information to optically encode relevant\ndepth information in a single snapshot. Furthermore, we propose to jointly\nlearn the color-coded aperture (CCA) pattern and a convolutional neural network\n(CNN) to retrieve depth information by using an end-to-end optimization\napproach. We demonstrate through different experiments on three different data\nsets that the designed color-encoding has the potential to remove depth\nambiguities and provides better depth estimates compared to state-of-the-art\napproaches. Additionally, we build a low-cost prototype of our CCA using a\nphotographic film and validate the proposed approach in real scenarios.\n", "date": "2023-09-14", "categories": []}
{"id": "2309.08596", "title": "Robust e-NeRF: NeRF from Sparse & Noisy Events under Non-Uniform Motion", "abstract": "  Event cameras offer many advantages over standard cameras due to their\ndistinctive principle of operation: low power, low latency, high temporal\nresolution and high dynamic range. Nonetheless, the success of many downstream\nvisual applications also hinges on an efficient and effective scene\nrepresentation, where Neural Radiance Field (NeRF) is seen as the leading\ncandidate. Such promise and potential of event cameras and NeRF inspired recent\nworks to investigate on the reconstruction of NeRF from moving event cameras.\nHowever, these works are mainly limited in terms of the dependence on dense and\nlow-noise event streams, as well as generalization to arbitrary contrast\nthreshold values and camera speed profiles. In this work, we propose Robust\ne-NeRF, a novel method to directly and robustly reconstruct NeRFs from moving\nevent cameras under various real-world conditions, especially from sparse and\nnoisy events generated under non-uniform motion. It consists of two key\ncomponents: a realistic event generation model that accounts for various\nintrinsic parameters (e.g. time-independent, asymmetric threshold and\nrefractory period) and non-idealities (e.g. pixel-to-pixel threshold\nvariation), as well as a complementary pair of normalized reconstruction losses\nthat can effectively generalize to arbitrary speed profiles and intrinsic\nparameter values without such prior knowledge. Experiments on real and novel\nrealistically simulated sequences verify our effectiveness. Our code, synthetic\ndataset and improved event simulator are public.\n", "date": "2023-09-15", "categories": []}
{"id": "2309.06724", "title": "Deep Nonparametric Convexified Filtering for Computational Photography,\n  Image Synthesis and Adversarial Defense", "abstract": "  We aim to provide a general framework of for computational photography that\nrecovers the real scene from imperfect images, via the Deep Nonparametric\nConvexified Filtering (DNCF). It is consists of a nonparametric deep network to\nresemble the physical equations behind the image formation, such as denoising,\nsuper-resolution, inpainting, and flash. DNCF has no parameterization dependent\non training data, therefore has a strong generalization and robustness to\nadversarial image manipulation. During inference, we also encourage the network\nparameters to be nonnegative and create a bi-convex function on the input and\nparameters, and this adapts to second-order optimization algorithms with\ninsufficient running time, having 10X acceleration over Deep Image Prior. With\nthese tools, we empirically verify its capability to defend image\nclassification deep networks against adversary attack algorithms in real-time.\n", "date": "2023-09-13", "categories": []}
{"id": "2309.06652", "title": "Event-Driven Imaging in Turbid Media: A Confluence of Optoelectronics\n  and Neuromorphic Computation", "abstract": "  In this paper a new optical-computational method is introduced to unveil\nimages of targets whose visibility is severely obscured by light scattering in\ndense, turbid media. The targets of interest are taken to be dynamic in that\ntheir optical properties are time-varying whether stationary in space or\nmoving. The scheme, to our knowledge the first of its kind, is human vision\ninspired whereby diffuse photons collected from the turbid medium are first\ntransformed to spike trains by a dynamic vision sensor as in the retina, and\nimage reconstruction is then performed by a neuromorphic computing approach\nmimicking the brain. We combine benchtop experimental data in both reflection\n(backscattering) and transmission geometries with support from physics-based\nsimulations to develop a neuromorphic computational model and then apply this\nfor image reconstruction of different MNIST characters and image sets by a\ndedicated deep spiking neural network algorithm. Image reconstruction is\nachieved under conditions of turbidity where an original image is\nunintelligible to the human eye or a digital video camera, yet clearly and\nquantifiable identifiable when using the new neuromorphic computational\napproach.\n", "date": "2023-09-13", "categories": []}
{"id": "2309.04752", "title": "Deep Video Restoration for Under-Display Camera", "abstract": "  Images or videos captured by the Under-Display Camera (UDC) suffer from\nsevere degradation, such as saturation degeneration and color shift. While\nrestoration for UDC has been a critical task, existing works of UDC restoration\nfocus only on images. UDC video restoration (UDC-VR) has not been explored in\nthe community. In this work, we first propose a GAN-based generation pipeline\nto simulate the realistic UDC degradation process. With the pipeline, we build\nthe first large-scale UDC video restoration dataset called PexelsUDC, which\nincludes two subsets named PexelsUDC-T and PexelsUDC-P corresponding to\ndifferent displays for UDC. Using the proposed dataset, we conduct extensive\nbenchmark studies on existing video restoration methods and observe their\nlimitations on the UDC-VR task. To this end, we propose a novel\ntransformer-based baseline method that adaptively enhances degraded videos. The\nkey components of the method are a spatial branch with local-aware\ntransformers, a temporal branch embedded temporal transformers, and a\nspatial-temporal fusion module. These components drive the model to fully\nexploit spatial and temporal information for UDC-VR. Extensive experiments show\nthat our method achieves state-of-the-art performance on PexelsUDC. The\nbenchmark and the baseline method are expected to promote the progress of\nUDC-VR in the community, which will be made public.\n", "date": "2023-09-09", "categories": []}
{"id": "2207.11336", "title": "First AI for deep super-resolution wide-field imaging in radio\n  astronomy: unveiling structure in ESO 137--006", "abstract": "  We introduce the first AI-based framework for deep, super-resolution,\nwide-field radio-interferometric imaging, and demonstrate it on observations of\nthe ESO~137-006 radio galaxy. The algorithmic framework to solve the inverse\nproblem for image reconstruction builds on a recent ``plug-and-play'' scheme\nwhereby a denoising operator is injected as an image regulariser in an\noptimisation algorithm, which alternates until convergence between denoising\nsteps and gradient-descent data-fidelity steps. We investigate handcrafted and\nlearned variants of high-resolution high-dynamic range denoisers. We propose a\nparallel algorithm implementation relying on automated decompositions of the\nimage into facets and the measurement operator into sparse low-dimensional\nblocks, enabling scalability to large data and image dimensions. We validate\nour framework for image formation at a wide field of view containing\nESO~137-006, from 19 gigabytes of MeerKAT data at 1053 and 1399 MHz. The\nrecovered maps exhibit significantly more resolution and dynamic range than\nCLEAN, revealing collimated synchrotron threads close to the galactic core.\n", "date": "2022-07-22", "categories": []}
{"id": "2302.14149", "title": "Scalable precision wide-field imaging in radio interferometry: II. AIRI\n  validated on ASKAP data", "abstract": "  Accompanying Part I, this sequel delineates a validation of the recently\nproposed AI for Regularisation in radio-interferometric Imaging (AIRI)\nalgorithm on observations from the Australian Square Kilometre Array Pathfinder\n(ASKAP). The monochromatic AIRI-ASKAP images showcased in this work are formed\nusing the same parallelised and automated imaging framework described in Part\nI: ``uSARA validated on ASKAP data''. Using a Plug-and-Play approach, AIRI\ndiffers from uSARA by substituting a trained denoising deep neural network\n(DNN) for the proximal operator in the regularisation step of the\nforward-backward algorithm during deconvolution. We build a trained shelf of\nDNN denoisers which target the estimated image-dynamic-ranges of our selected\ndata. Furthermore, we quantify variations of AIRI reconstructions when\nselecting the nearest DNN on the shelf versus using a universal DNN with the\nhighest dynamic range, opening the door to a more complete framework that not\nonly delivers image estimation but also quantifies epistemic model uncertainty.\nWe continue our comparative analysis of source structure, diffuse flux\nmeasurements, and spectral index maps of selected target sources as imaged by\nAIRI and the algorithms in Part I -- uSARA and WSClean. Overall we see an\nimprovement over uSARA and WSClean in the reconstruction of diffuse components\nin AIRI images. The scientific potential delivered by AIRI is evident in\nfurther imaging precision, more accurate spectral index maps, and a significant\nacceleration in deconvolution time, whereby AIRI is four times faster than its\nsub-iterative sparsity-based counterpart uSARA.\n", "date": "2023-02-27", "categories": []}
{"id": "2309.00066", "title": "SoDaCam: Software-defined Cameras via Single-Photon Imaging", "abstract": "  Reinterpretable cameras are defined by their post-processing capabilities\nthat exceed traditional imaging. We present \"SoDaCam\" that provides\nreinterpretable cameras at the granularity of photons, from photon-cubes\nacquired by single-photon devices. Photon-cubes represent the spatio-temporal\ndetections of photons as a sequence of binary frames, at frame-rates as high as\n100 kHz. We show that simple transformations of the photon-cube, or photon-cube\nprojections, provide the functionality of numerous imaging systems including:\nexposure bracketing, flutter shutter cameras, video compressive systems, event\ncameras, and even cameras that move during exposure. Our photon-cube\nprojections offer the flexibility of being software-defined constructs that are\nonly limited by what is computable, and shot-noise. We exploit this flexibility\nto provide new capabilities for the emulated cameras. As an added benefit, our\nprojections provide camera-dependent compression of photon-cubes, which we\ndemonstrate using an implementation of our projections on a novel compute\narchitecture that is designed for single-photon imaging.\n", "date": "2023-08-31", "categories": []}
{"id": "2309.04171", "title": "PRISTA-Net: Deep Iterative Shrinkage Thresholding Network for Coded\n  Diffraction Patterns Phase Retrieval", "abstract": "  The problem of phase retrieval (PR) involves recovering an unknown image from\nlimited amplitude measurement data and is a challenge nonlinear inverse problem\nin computational imaging and image processing. However, many of the PR methods\nare based on black-box network models that lack interpretability and\nplug-and-play (PnP) frameworks that are computationally complex and require\ncareful parameter tuning. To address this, we have developed PRISTA-Net, a deep\nunfolding network (DUN) based on the first-order iterative shrinkage\nthresholding algorithm (ISTA). This network utilizes a learnable nonlinear\ntransformation to address the proximal-point mapping sub-problem associated\nwith the sparse priors, and an attention mechanism to focus on phase\ninformation containing image edges, textures, and structures. Additionally, the\nfast Fourier transform (FFT) is used to learn global features to enhance local\ninformation, and the designed logarithmic-based loss function leads to\nsignificant improvements when the noise level is low. All parameters in the\nproposed PRISTA-Net framework, including the nonlinear transformation,\nthreshold parameters, and step size, are learned end-to-end instead of being\nmanually set. This method combines the interpretability of traditional methods\nwith the fast inference ability of deep learning and is able to handle noise at\neach iteration during the unfolding stage, thus improving recovery quality.\nExperiments on Coded Diffraction Patterns (CDPs) measurements demonstrate that\nour approach outperforms the existing state-of-the-art methods in terms of\nqualitative and quantitative evaluations. Our source codes are available at\n\\emph{https://github.com/liuaxou/PRISTA-Net}.\n", "date": "2023-09-08", "categories": []}
{"id": "2309.04366", "title": "CNN Injected Transformer for Image Exposure Correction", "abstract": "  Capturing images with incorrect exposure settings fails to deliver a\nsatisfactory visual experience. Only when the exposure is properly set, can the\ncolor and details of the images be appropriately preserved. Previous exposure\ncorrection methods based on convolutions often produce exposure deviation in\nimages as a consequence of the restricted receptive field of convolutional\nkernels. This issue arises because convolutions are not capable of capturing\nlong-range dependencies in images accurately. To overcome this challenge, we\ncan apply the Transformer to address the exposure correction problem,\nleveraging its capability in modeling long-range dependencies to capture global\nrepresentation. However, solely relying on the window-based Transformer leads\nto visually disturbing blocking artifacts due to the application of\nself-attention in small patches. In this paper, we propose a CNN Injected\nTransformer (CIT) to harness the individual strengths of CNN and Transformer\nsimultaneously. Specifically, we construct the CIT by utilizing a window-based\nTransformer to exploit the long-range interactions among different regions in\nthe entire image. Within each CIT block, we incorporate a channel attention\nblock (CAB) and a half-instance normalization block (HINB) to assist the\nwindow-based self-attention to acquire the global statistics and refine local\nfeatures. In addition to the hybrid architecture design for exposure\ncorrection, we apply a set of carefully formulated loss functions to improve\nthe spatial coherence and rectify potential color deviations. Extensive\nexperiments demonstrate that our image exposure correction method outperforms\nstate-of-the-art approaches in terms of both quantitative and qualitative\nmetrics.\n", "date": "2023-09-08", "categories": []}
{"id": "2207.03367", "title": "Joint Super-Resolution and Inverse Tone-Mapping: A Feature Decomposition\n  Aggregation Network and A New Benchmark", "abstract": "  Joint Super-Resolution and Inverse Tone-Mapping (joint SR-ITM) aims to\nincrease the resolution and dynamic range of low-resolution and standard\ndynamic range images. Recent networks mainly resort to image decomposition\ntechniques with complex multi-branch architectures. However, the fixed\ndecomposition techniques would largely restricts their power on versatile\nimages. To exploit the potential power of decomposition mechanism, in this\npaper, we generalize it from the image domain to the broader feature domain. To\nthis end, we propose a lightweight Feature Decomposition Aggregation Network\n(FDAN). In particular, we design a Feature Decomposition Block (FDB) to achieve\nlearnable separation of detail and base feature maps, and develop a\nHierarchical Feature Decomposition Group by cascading FDBs for powerful\nmulti-level feature decomposition. Moreover, to better evaluate the comparison\nmethods, we collect a large-scale dataset for joint SR-ITM, i.e., SRITM-4K,\nwhich provides versatile scenarios for robust model training and evaluation.\nExperimental results on two benchmark datasets demonstrate that our FDAN is\nefficient and outperforms state-of-the-art methods on joint SR-ITM. The code of\nour FDAN and the SRITM-4K dataset are available at\nhttps://github.com/CS-GangXu/FDAN.\n", "date": "2022-07-07", "categories": []}
{"id": "2309.03216", "title": "A Multisensor Hyperspectral Benchmark Dataset For Unmixing of Intimate\n  Mixtures", "abstract": "  Optical hyperspectral cameras capture the spectral reflectance of materials.\nSince many materials behave as heterogeneous intimate mixtures with which each\nphoton interacts differently, the relationship between spectral reflectance and\nmaterial composition is very complex. Quantitative validation of spectral\nunmixing algorithms requires high-quality ground truth fractional abundance\ndata, which are very difficult to obtain. In this work, we generated a\ncomprehensive laboratory ground truth dataset of intimately mixed mineral\npowders. For this, five clay powders (Kaolin, Roof clay, Red clay, mixed clay,\nand Calcium hydroxide) were mixed homogeneously to prepare 325 samples of 60\nbinary, 150 ternary, 100 quaternary, and 15 quinary mixtures. Thirteen\ndifferent hyperspectral sensors have been used to acquire the reflectance\nspectra of these mixtures in the visible, near, short, mid, and long-wavelength\ninfrared regions (350-15385) nm. {\\color{black} Overlaps in wavelength regions\ndue to the operational ranges of each sensor} and variations in acquisition\nconditions {\\color{black} resulted in} a large amount of spectral variability.\nGround truth composition is given by construction, but to verify that the\ngenerated samples are sufficiently homogeneous, XRD and XRF elemental analysis\nis performed. We believe these data will be beneficial for validating advanced\nmethods for nonlinear unmixing and material composition estimation, including\nstudying spectral variability and training supervised unmixing approaches. The\ndatasets can be downloaded from the following link:\nhttps://github.com/VisionlabUA/Multisensor_datasets.\n", "date": "2023-08-30", "categories": []}
{"id": "2309.03811", "title": "Panoramas from Photons", "abstract": "  Scene reconstruction in the presence of high-speed motion and low\nillumination is important in many applications such as augmented and virtual\nreality, drone navigation, and autonomous robotics. Traditional motion\nestimation techniques fail in such conditions, suffering from too much blur in\nthe presence of high-speed motion and strong noise in low-light conditions.\nSingle-photon cameras have recently emerged as a promising technology capable\nof capturing hundreds of thousands of photon frames per second thanks to their\nhigh speed and extreme sensitivity. Unfortunately, traditional computer vision\ntechniques are not well suited for dealing with the binary-valued photon data\ncaptured by these cameras because these are corrupted by extreme Poisson noise.\nHere we present a method capable of estimating extreme scene motion under\nchallenging conditions, such as low light or high dynamic range, from a\nsequence of high-speed image frames such as those captured by a single-photon\ncamera. Our method relies on iteratively improving a motion estimate by\ngrouping and aggregating frames after-the-fact, in a stratified manner. We\ndemonstrate the creation of high-quality panoramas under fast motion and\nextremely low light, and super-resolution results using a custom single-photon\ncamera prototype. For code and supplemental material see our\n$\\href{https://wisionlab.com/project/panoramas-from-photons/}{\\text{project\nwebpage}}$.\n", "date": "2023-09-07", "categories": []}
{"id": "2309.03900", "title": "Learning Continuous Exposure Value Representations for Single-Image HDR\n  Reconstruction", "abstract": "  Deep learning is commonly used to reconstruct HDR images from LDR images. LDR\nstack-based methods are used for single-image HDR reconstruction, generating an\nHDR image from a deep learning-generated LDR stack. However, current methods\ngenerate the stack with predetermined exposure values (EVs), which may limit\nthe quality of HDR reconstruction. To address this, we propose the continuous\nexposure value representation (CEVR), which uses an implicit function to\ngenerate LDR images with arbitrary EVs, including those unseen during training.\nOur approach generates a continuous stack with more images containing diverse\nEVs, significantly improving HDR reconstruction. We use a cycle training\nstrategy to supervise the model in generating continuous EV LDR images without\ncorresponding ground truths. Our CEVR model outperforms existing methods, as\ndemonstrated by experimental results.\n", "date": "2023-09-07", "categories": []}
{"id": "2304.06322", "title": "Learning-based Spatial and Angular Information Separation for Light\n  Field Compression", "abstract": "  Light fields are a type of image data that capture both spatial and angular\nscene information by recording light rays emitted by a scene from different\norientations. In this context, spatial information is defined as features that\nremain static regardless of perspectives, while angular information refers to\nfeatures that vary between viewpoints. We propose a novel neural network that,\nby design, can separate angular and spatial information of a light field. The\nnetwork represents spatial information using spatial kernels shared among all\nSub-Aperture Images (SAIs), and angular information using sets of angular\nkernels for each SAI. To further improve the representation capability of the\nnetwork without increasing parameter number, we also introduce angular kernel\nallocation and kernel tensor decomposition mechanisms. Extensive experiments\ndemonstrate the benefits of information separation: when applied to the\ncompression task, our network outperforms other state-of-the-art methods by a\nlarge margin. And angular information can be easily transferred to other scenes\nfor rendering dense views, showing the successful separation and the potential\nuse case for the view synthesis task. We plan to release the code upon\nacceptance of the paper to encourage further research on this topic.\n", "date": "2023-04-13", "categories": []}
{"id": "2309.03185", "title": "Bayes' Rays: Uncertainty Quantification for Neural Radiance Fields", "abstract": "  Neural Radiance Fields (NeRFs) have shown promise in applications like view\nsynthesis and depth estimation, but learning from multiview images faces\ninherent uncertainties. Current methods to quantify them are either heuristic\nor computationally demanding. We introduce BayesRays, a post-hoc framework to\nevaluate uncertainty in any pre-trained NeRF without modifying the training\nprocess. Our method establishes a volumetric uncertainty field using spatial\nperturbations and a Bayesian Laplace approximation. We derive our algorithm\nstatistically and show its superior performance in key metrics and\napplications. Additional results available at: https://bayesrays.github.io.\n", "date": "2023-09-06", "categories": []}
{"id": "2308.15942", "title": "Stage-by-stage Wavelet Optimization Refinement Diffusion Model for\n  Sparse-View CT Reconstruction", "abstract": "  Diffusion models have emerged as potential tools to tackle the challenge of\nsparse-view CT reconstruction, displaying superior performance compared to\nconventional methods. Nevertheless, these prevailing diffusion models\npredominantly focus on the sinogram or image domains, which can lead to\ninstability during model training, potentially culminating in convergence\ntowards local minimal solutions. The wavelet trans-form serves to disentangle\nimage contents and features into distinct frequency-component bands at varying\nscales, adeptly capturing diverse directional structures. Employing the Wavelet\ntransform as a guiding sparsity prior significantly enhances the robustness of\ndiffusion models. In this study, we present an innovative approach named the\nStage-by-stage Wavelet Optimization Refinement Diffusion (SWORD) model for\nsparse-view CT reconstruction. Specifically, we establish a unified\nmathematical model integrating low-frequency and high-frequency generative\nmodels, achieving the solution with optimization procedure. Furthermore, we\nperform the low-frequency and high-frequency generative models on wavelet's\ndecomposed components rather than sinogram or image domains, ensuring the\nstability of model training. Our method rooted in established optimization\ntheory, comprising three distinct stages, including low-frequency generation,\nhigh-frequency refinement and domain transform. Our experimental results\ndemonstrate that the proposed method outperforms existing state-of-the-art\nmethods both quantitatively and qualitatively.\n", "date": "2023-08-30", "categories": []}
{"id": "2309.01624", "title": "AGG-Net: Attention Guided Gated-convolutional Network for Depth Image\n  Completion", "abstract": "  Recently, stereo vision based on lightweight RGBD cameras has been widely\nused in various fields. However, limited by the imaging principles, the\ncommonly used RGB-D cameras based on TOF, structured light, or binocular vision\nacquire some invalid data inevitably, such as weak reflection, boundary\nshadows, and artifacts, which may bring adverse impacts to the follow-up work.\nIn this paper, we propose a new model for depth image completion based on the\nAttention Guided Gated-convolutional Network (AGG-Net), through which more\naccurate and reliable depth images can be obtained from the raw depth maps and\nthe corresponding RGB images. Our model employs a UNet-like architecture which\nconsists of two parallel branches of depth and color features. In the encoding\nstage, an Attention Guided Gated-Convolution (AG-GConv) module is proposed to\nrealize the fusion of depth and color features at different scales, which can\neffectively reduce the negative impacts of invalid depth data on the\nreconstruction. In the decoding stage, an Attention Guided Skip Connection\n(AG-SC) module is presented to avoid introducing too many depth-irrelevant\nfeatures to the reconstruction. The experimental results demonstrate that our\nmethod outperforms the state-of-the-art methods on the popular benchmarks\nNYU-Depth V2, DIML, and SUN RGB-D.\n", "date": "2023-09-04", "categories": []}
{"id": "2303.06681", "title": "Learning Deep Intensity Field for Extremely Sparse-View CBCT\n  Reconstruction", "abstract": "  Sparse-view cone-beam CT (CBCT) reconstruction is an important direction to\nreduce radiation dose and benefit clinical applications. Previous voxel-based\ngeneration methods represent the CT as discrete voxels, resulting in high\nmemory requirements and limited spatial resolution due to the use of 3D\ndecoders. In this paper, we formulate the CT volume as a continuous intensity\nfield and develop a novel DIF-Net to perform high-quality CBCT reconstruction\nfrom extremely sparse (fewer than 10) projection views at an ultrafast speed.\nThe intensity field of a CT can be regarded as a continuous function of 3D\nspatial points. Therefore, the reconstruction can be reformulated as regressing\nthe intensity value of an arbitrary 3D point from given sparse projections.\nSpecifically, for a point, DIF-Net extracts its view-specific features from\ndifferent 2D projection views. These features are subsequently aggregated by a\nfusion module for intensity estimation. Notably, thousands of points can be\nprocessed in parallel to improve efficiency during training and testing. In\npractice, we collect a knee CBCT dataset to train and evaluate DIF-Net.\nExtensive experiments show that our approach can reconstruct CBCT with high\nimage quality and high spatial resolution from extremely sparse views within\n1.6 seconds, significantly outperforming state-of-the-art methods. Our code\nwill be available at https://github.com/xmed-lab/DIF-Net.\n", "date": "2023-03-12", "categories": []}
{"id": "2307.12751", "title": "ICF-SRSR: Invertible scale-Conditional Function for Self-Supervised\n  Real-world Single Image Super-Resolution", "abstract": "  Single image super-resolution (SISR) is a challenging ill-posed problem that\naims to up-sample a given low-resolution (LR) image to a high-resolution (HR)\ncounterpart. Due to the difficulty in obtaining real LR-HR training pairs,\nrecent approaches are trained on simulated LR images degraded by simplified\ndown-sampling operators, e.g., bicubic. Such an approach can be problematic in\npractice because of the large gap between the synthesized and real-world LR\nimages. To alleviate the issue, we propose a novel Invertible scale-Conditional\nFunction (ICF), which can scale an input image and then restore the original\ninput with different scale conditions. By leveraging the proposed ICF, we\nconstruct a novel self-supervised SISR framework (ICF-SRSR) to handle the\nreal-world SR task without using any paired/unpaired training data.\nFurthermore, our ICF-SRSR can generate realistic and feasible LR-HR pairs,\nwhich can make existing supervised SISR networks more robust. Extensive\nexperiments demonstrate the effectiveness of the proposed method in handling\nSISR in a fully self-supervised manner. Our ICF-SRSR demonstrates superior\nperformance compared to the existing methods trained on synthetic paired images\nin real-world scenarios and exhibits comparable performance compared to\nstate-of-the-art supervised/unsupervised methods on public benchmark datasets.\n", "date": "2023-07-24", "categories": []}
{"id": "2308.15918", "title": "Physics-Informed DeepMRI: Bridging the Gap from Heat Diffusion to\n  k-Space Interpolation", "abstract": "  In the field of parallel imaging (PI), alongside image-domain regularization\nmethods, substantial research has been dedicated to exploring $k$-space\ninterpolation. However, the interpretability of these methods remains an\nunresolved issue. Furthermore, these approaches currently face acceleration\nlimitations that are comparable to those experienced by image-domain methods.\nIn order to enhance interpretability and overcome the acceleration limitations,\nthis paper introduces an interpretable framework that unifies both $k$-space\ninterpolation techniques and image-domain methods, grounded in the physical\nprinciples of heat diffusion equations. Building upon this foundational\nframework, a novel $k$-space interpolation method is proposed. Specifically, we\nmodel the process of high-frequency information attenuation in $k$-space as a\nheat diffusion equation, while the effort to reconstruct high-frequency\ninformation from low-frequency regions can be conceptualized as a reverse heat\nequation. However, solving the reverse heat equation poses a challenging\ninverse problem. To tackle this challenge, we modify the heat equation to align\nwith the principles of magnetic resonance PI physics and employ the score-based\ngenerative method to precisely execute the modified reverse heat diffusion.\nFinally, experimental validation conducted on publicly available datasets\ndemonstrates the superiority of the proposed approach over traditional\n$k$-space interpolation methods, deep learning-based $k$-space interpolation\nmethods, and conventional diffusion models in terms of reconstruction accuracy,\nparticularly in high-frequency regions.\n", "date": "2023-08-30", "categories": []}
{"id": "2304.11751", "title": "Score-Based Diffusion Models as Principled Priors for Inverse Imaging", "abstract": "  Priors are essential for reconstructing images from noisy and/or incomplete\nmeasurements. The choice of the prior determines both the quality and\nuncertainty of recovered images. We propose turning score-based diffusion\nmodels into principled image priors (\"score-based priors\") for analyzing a\nposterior of images given measurements. Previously, probabilistic priors were\nlimited to handcrafted regularizers and simple distributions. In this work, we\nempirically validate the theoretically-proven probability function of a\nscore-based diffusion model. We show how to sample from resulting posteriors by\nusing this probability function for variational inference. Our results,\nincluding experiments on denoising, deblurring, and interferometric imaging,\nsuggest that score-based priors enable principled inference with a\nsophisticated, data-driven image prior.\n", "date": "2023-04-23", "categories": []}
{"id": "2305.09121", "title": "A Conditional Denoising Diffusion Probabilistic Model for Radio\n  Interferometric Image Reconstruction", "abstract": "  In radio astronomy, signals from radio telescopes are transformed into images\nof observed celestial objects, or sources. However, these images, called dirty\nimages, contain real sources as well as artifacts due to signal sparsity and\nother factors. Therefore, radio interferometric image reconstruction is\nperformed on dirty images, aiming to produce clean images in which artifacts\nare reduced and real sources are recovered. So far, existing methods have\nlimited success on recovering faint sources, preserving detailed structures,\nand eliminating artifacts. In this paper, we present VIC-DDPM, a Visibility and\nImage Conditioned Denoising Diffusion Probabilistic Model. Our main idea is to\nuse both the original visibility data in the spectral domain and dirty images\nin the spatial domain to guide the image generation process with DDPM. This\nway, we can leverage DDPM to generate fine details and eliminate noise, while\nutilizing visibility data to separate signals from noise and retaining spatial\ninformation in dirty images. We have conducted experiments in comparison with\nboth traditional methods and recent deep learning based approaches. Our results\nshow that our method significantly improves the resulting images by reducing\nartifacts, preserving fine details, and recovering dim sources. This\nadvancement further facilitates radio astronomical data analysis tasks on\ncelestial phenomena.\n", "date": "2023-05-16", "categories": []}
{"id": "2308.02559", "title": "DLSIA: Deep Learning for Scientific Image Analysis", "abstract": "  We introduce DLSIA (Deep Learning for Scientific Image Analysis), a\nPython-based machine learning library that empowers scientists and researchers\nacross diverse scientific domains with a range of customizable convolutional\nneural network (CNN) architectures for a wide variety of tasks in image\nanalysis to be used in downstream data processing, or for\nexperiment-in-the-loop computing scenarios. DLSIA features easy-to-use\narchitectures such as autoencoders, tunable U-Nets, and parameter-lean\nmixed-scale dense networks (MSDNets). Additionally, we introduce sparse\nmixed-scale networks (SMSNets), generated using random graphs and sparse\nconnections. As experimental data continues to grow in scale and complexity,\nDLSIA provides accessible CNN construction and abstracts CNN complexities,\nallowing scientists to tailor their machine learning approaches, accelerate\ndiscoveries, foster interdisciplinary collaboration, and advance research in\nscientific image analysis.\n", "date": "2023-08-02", "categories": []}
{"id": "2308.02952", "title": "Multispectral Quantitative Phase Imaging Using a Diffractive Optical\n  Network", "abstract": "  As a label-free imaging technique, quantitative phase imaging (QPI) provides\noptical path length information of transparent specimens for various\napplications in biology, materials science, and engineering. Multispectral QPI\nmeasures quantitative phase information across multiple spectral bands,\npermitting the examination of wavelength-specific phase and dispersion\ncharacteristics of samples. Here, we present the design of a diffractive\nprocessor that can all-optically perform multispectral quantitative phase\nimaging of transparent phase-only objects in a snapshot. Our design utilizes\nspatially engineered diffractive layers, optimized through deep learning, to\nencode the phase profile of the input object at a predetermined set of\nwavelengths into spatial intensity variations at the output plane, allowing\nmultispectral QPI using a monochrome focal plane array. Through numerical\nsimulations, we demonstrate diffractive multispectral processors to\nsimultaneously perform quantitative phase imaging at 9 and 16 target spectral\nbands in the visible spectrum. These diffractive multispectral processors\nmaintain uniform performance across all the wavelength channels, revealing a\ndecent QPI performance at each target wavelength. The generalization of these\ndiffractive processor designs is validated through numerical tests on unseen\nobjects, including thin Pap smear images. Due to its all-optical processing\ncapability using passive dielectric diffractive materials, this diffractive\nmultispectral QPI processor offers a compact and power-efficient solution for\nhigh-throughput quantitative phase microscopy and spectroscopy. This framework\ncan operate at different parts of the electromagnetic spectrum and be used for\na wide range of phase imaging and sensing applications.\n", "date": "2023-08-05", "categories": []}
{"id": "2308.14144", "title": "Learning end-to-end inversion of circular Radon transforms in the\n  partial radial setup", "abstract": "  We present a deep learning-based computational algorithm for inversion of\ncircular Radon transforms in the partial radial setup, arising in photoacoustic\ntomography. We first demonstrate that the truncated singular value\ndecomposition-based method, which is the only traditional algorithm available\nto solve this problem, leads to severe artifacts which renders the\nreconstructed field as unusable. With the objective of overcoming this\ncomputational bottleneck, we train a ResBlock based U-Net to recover the\ninferred field that directly operates on the measured data. Numerical results\nwith augmented Shepp-Logan phantoms, in the presence of noisy full and limited\nview data, demonstrate the superiority of the proposed algorithm.\n", "date": "2023-08-27", "categories": []}
{"id": "2304.00015", "title": "DRIP: Deep Regularizers for Inverse Problems", "abstract": "  In this paper we consider inverse problems that are mathematically ill-posed.\nThat is, given some (noisy) data, there is more than one solution that\napproximately fits the data. In recent years, deep neural techniques that find\nthe most appropriate solution, in the sense that it contains a-priori\ninformation, were developed. However, they suffer from several shortcomings.\nFirst, most techniques cannot guarantee that the solution fits the data at\ninference. Second, while the derivation of the techniques is inspired by the\nexistence of a valid scalar regularization function, such techniques do not in\npractice rely on such a function, and therefore veer away from classical\nvariational techniques. In this work we introduce a new family of neural\nregularizers for the solution of inverse problems. These regularizers are based\non a variational formulation and are guaranteed to fit the data. We demonstrate\ntheir use on a number of highly ill-posed problems, from image deblurring to\nlimited angle tomography.\n", "date": "2023-03-30", "categories": []}
{"id": "2308.13473", "title": "Unlocking the Performance of Proximity Sensors by Utilizing Transient\n  Histograms", "abstract": "  We provide methods which recover planar scene geometry by utilizing the\ntransient histograms captured by a class of close-range time-of-flight (ToF)\ndistance sensor. A transient histogram is a one dimensional temporal waveform\nwhich encodes the arrival time of photons incident on the ToF sensor.\nTypically, a sensor processes the transient histogram using a proprietary\nalgorithm to produce distance estimates, which are commonly used in several\nrobotics applications. Our methods utilize the transient histogram directly to\nenable recovery of planar geometry more accurately than is possible using only\nproprietary distance estimates, and consistent recovery of the albedo of the\nplanar surface, which is not possible with proprietary distance estimates\nalone. This is accomplished via a differentiable rendering pipeline, which\nsimulates the transient imaging process, allowing direct optimization of scene\ngeometry to match observations. To validate our methods, we capture 3,800\nmeasurements of eight planar surfaces from a wide range of viewpoints, and show\nthat our method outperforms the proprietary-distance-estimate baseline by an\norder of magnitude in most scenarios. We demonstrate a simple robotics\napplication which uses our method to sense the distance to and slope of a\nplanar surface from a sensor mounted on the end effector of a robot arm.\n", "date": "2023-08-25", "categories": []}
{"id": "2301.06496", "title": "Efficient data transport over multimode light-pipes with Megapixel\n  images using differentiable ray tracing and Machine-learning", "abstract": "  Retrieving images transmitted through multi-mode fibers is of growing\ninterest, thanks to their ability to confine and transport light efficiently in\na compact system. Here, we demonstrate machine-learning-based decoding of\nlarge-scale digital images (pages), maximizing page capacity for optical\nstorage applications. Using a millimeter-sized square cross-section waveguide,\nwe image an 8-bit spatial light modulator, presenting data as a matrix of\nsymbols. Normally, decoders will incur a prohibitive O(n^2) computational\nscaling to decode n symbols in spatially scrambled data. However, by combining\na digital twin of the setup with a U-Net, we can retrieve up to 66 kB using\nefficient convolutional operations only. We compare trainable ray-tracing-based\nwith eigenmode-based twins and show the former to be superior thanks to its\nability to overcome the simulation-to-experiment gap by adjusting to optical\nimperfections. We train the pipeline end-to-end using a differentiable\nmutual-information estimator based on the von-Mises distribution, generally\napplicable to phase-coding channels.\n", "date": "2023-01-16", "categories": []}
{"id": "2304.05669", "title": "Factorized Inverse Path Tracing for Efficient and Accurate\n  Material-Lighting Estimation", "abstract": "  Inverse path tracing has recently been applied to joint material and lighting\nestimation, given geometry and multi-view HDR observations of an indoor scene.\nHowever, it has two major limitations: path tracing is expensive to compute,\nand ambiguities exist between reflection and emission. Our Factorized Inverse\nPath Tracing (FIPT) addresses these challenges by using a factored light\ntransport formulation and finds emitters driven by rendering errors. Our\nalgorithm enables accurate material and lighting optimization faster than\nprevious work, and is more effective at resolving ambiguities. The exhaustive\nexperiments on synthetic scenes show that our method (1) outperforms\nstate-of-the-art indoor inverse rendering and relighting methods particularly\nin the presence of complex illumination effects; (2) speeds up inverse path\ntracing optimization to less than an hour. We further demonstrate robustness to\nnoisy inputs through material and lighting estimates that allow plausible\nrelighting in a real scene. The source code is available at:\nhttps://github.com/lwwu2/fipt\n", "date": "2023-04-12", "categories": []}
{"id": "2304.13207", "title": "EverLight: Indoor-Outdoor Editable HDR Lighting Estimation", "abstract": "  Because of the diversity in lighting environments, existing illumination\nestimation techniques have been designed explicitly on indoor or outdoor\nenvironments. Methods have focused specifically on capturing accurate energy\n(e.g., through parametric lighting models), which emphasizes shading and strong\ncast shadows; or producing plausible texture (e.g., with GANs), which\nprioritizes plausible reflections. Approaches which provide editable lighting\ncapabilities have been proposed, but these tend to be with simplified lighting\nmodels, offering limited realism. In this work, we propose to bridge the gap\nbetween these recent trends in the literature, and propose a method which\ncombines a parametric light model with 360{\\deg} panoramas, ready to use as\nHDRI in rendering engines. We leverage recent advances in GAN-based LDR\npanorama extrapolation from a regular image, which we extend to HDR using\nparametric spherical gaussians. To achieve this, we introduce a novel lighting\nco-modulation method that injects lighting-related features throughout the\ngenerator, tightly coupling the original or edited scene illumination within\nthe panorama generation process. In our representation, users can easily edit\nlight direction, intensity, number, etc. to impact shading while providing\nrich, complex reflections while seamlessly blending with the edits.\nFurthermore, our method encompasses indoor and outdoor environments,\ndemonstrating state-of-the-art results even when compared to domain-specific\nmethods.\n", "date": "2023-04-26", "categories": []}
{"id": "2308.10910", "title": "Federated Pseudo Modality Generation for Incomplete Multi-Modal MRI\n  Reconstruction", "abstract": "  While multi-modal learning has been widely used for MRI reconstruction, it\nrelies on paired multi-modal data which is difficult to acquire in real\nclinical scenarios. Especially in the federated setting, the common situation\nis that several medical institutions only have single-modal data, termed the\nmodality missing issue. Therefore, it is infeasible to deploy a standard\nfederated learning framework in such conditions. In this paper, we propose a\nnovel communication-efficient federated learning framework, namely Fed-PMG, to\naddress the missing modality challenge in federated multi-modal MRI\nreconstruction. Specifically, we utilize a pseudo modality generation mechanism\nto recover the missing modality for each single-modal client by sharing the\ndistribution information of the amplitude spectrum in frequency space. However,\nthe step of sharing the original amplitude spectrum leads to heavy\ncommunication costs. To reduce the communication cost, we introduce a\nclustering scheme to project the set of amplitude spectrum into finite cluster\ncentroids, and share them among the clients. With such an elaborate design, our\napproach can effectively complete the missing modality within an acceptable\ncommunication cost. Extensive experiments demonstrate that our proposed method\ncan attain similar performance with the ideal scenario, i.e., all clients have\nthe full set of modalities. The source code will be released.\n", "date": "2023-08-20", "categories": []}
{"id": "2308.11116", "title": "LAN-HDR: Luminance-based Alignment Network for High Dynamic Range Video\n  Reconstruction", "abstract": "  As demands for high-quality videos continue to rise, high-resolution and\nhigh-dynamic range (HDR) imaging techniques are drawing attention. To generate\nan HDR video from low dynamic range (LDR) images, one of the critical steps is\nthe motion compensation between LDR frames, for which most existing works\nemployed the optical flow algorithm. However, these methods suffer from flow\nestimation errors when saturation or complicated motions exist. In this paper,\nwe propose an end-to-end HDR video composition framework, which aligns LDR\nframes in the feature space and then merges aligned features into an HDR frame,\nwithout relying on pixel-domain optical flow. Specifically, we propose a\nluminance-based alignment network for HDR (LAN-HDR) consisting of an alignment\nmodule and a hallucination module. The alignment module aligns a frame to the\nadjacent reference by evaluating luminance-based attention, excluding color\ninformation. The hallucination module generates sharp details, especially for\nwashed-out areas due to saturation. The aligned and hallucinated features are\nthen blended adaptively to complement each other. Finally, we merge the\nfeatures to generate a final HDR frame. In training, we adopt a temporal loss,\nin addition to frame reconstruction losses, to enhance temporal consistency and\nthus reduce flickering. Extensive experiments demonstrate that our method\nperforms better or comparable to state-of-the-art methods on several\nbenchmarks.\n", "date": "2023-08-22", "categories": []}
{"id": "2308.09865", "title": "A Theory of Topological Derivatives for Inverse Rendering of Geometry", "abstract": "  We introduce a theoretical framework for differentiable surface evolution\nthat allows discrete topology changes through the use of topological\nderivatives for variational optimization of image functionals. While prior\nmethods for inverse rendering of geometry rely on silhouette gradients for\ntopology changes, such signals are sparse. In contrast, our theory derives\ntopological derivatives that relate the introduction of vanishing holes and\nphases to changes in image intensity. As a result, we enable differentiable\nshape perturbations in the form of hole or phase nucleation. We validate the\nproposed theory with optimization of closed curves in 2D and surfaces in 3D to\nlend insights into limitations of current methods and enable improved\napplications such as image vectorization, vector-graphics generation from text\nprompts, single-image reconstruction of shape ambigrams and multi-view 3D\nreconstruction.\n", "date": "2023-08-19", "categories": []}
{"id": "2301.08433", "title": "Unsupervised Light Field Depth Estimation via Multi-view Feature\n  Matching with Occlusion Prediction", "abstract": "  Depth estimation from light field (LF) images is a fundamental step for\nnumerous applications. Recently, learning-based methods have achieved higher\naccuracy and efficiency than the traditional methods. However, it is costly to\nobtain sufficient depth labels for supervised training. In this paper, we\npropose an unsupervised framework to estimate depth from LF images. First, we\ndesign a disparity estimation network (DispNet) with a coarse-to-fine structure\nto predict disparity maps from different view combinations. It explicitly\nperforms multi-view feature matching to learn the correspondences effectively.\nAs occlusions may cause the violation of photo-consistency, we introduce an\nocclusion prediction network (OccNet) to predict the occlusion maps, which are\nused as the element-wise weights of photometric loss to solve the occlusion\nissue and assist the disparity learning. With the disparity maps estimated by\nmultiple input combinations, we then propose a disparity fusion strategy based\non the estimated errors with effective occlusion handling to obtain the final\ndisparity map with higher accuracy. Experimental results demonstrate that our\nmethod achieves superior performance on both the dense and sparse LF images,\nand also shows better robustness and generalization on the real-world LF images\ncompared to the other methods.\n", "date": "2023-01-20", "categories": []}
{"id": "2305.12986", "title": "Sparsity and Coefficient Permutation Based Two-Domain AMP for Image\n  Block Compressed Sensing", "abstract": "  The learned denoising-based approximate message passing (LDAMP) algorithm has\nattracted great attention for image compressed sensing (CS) tasks. However, it\nhas two issues: first, its global measurement model severely restricts its\napplicability to high-dimensional images, and its block-based measurement\nmethod exhibits obvious block artifacts; second, the denoiser in the LDAMP is\ntoo simple, and existing denoisers have limited ability in detail recovery. In\nthis paper, to overcome the issues and develop a high-performance LDAMP method\nfor image block compressed sensing (BCS), we propose a novel sparsity and\ncoefficient permutation-based AMP (SCP-AMP) method consisting of the\nblock-based sampling and the two-domain reconstruction modules. In the sampling\nmodule, SCP-AMP adopts a discrete cosine transform (DCT) based sparsity\nstrategy to reduce the impact of the high-frequency coefficient on the\nreconstruction, followed by a coefficient permutation strategy to avoid block\nartifacts. In the reconstruction module, a two-domain AMP method with DCT\ndomain noise correction and pixel domain denoising is proposed for iterative\nreconstruction. Regarding the denoiser, we proposed a multi-level deep\nattention network (MDANet) to enhance the texture details by employing\nmulti-level features and multiple attention mechanisms. Extensive experiments\ndemonstrated that the proposed SCP-AMP method achieved better reconstruction\naccuracy than other state-of-the-art BCS algorithms in terms of both visual\nperception and objective metrics.\n", "date": "2023-05-22", "categories": []}
{"id": "2308.08715", "title": "V-FUSE: Volumetric Depth Map Fusion with Long-Range Constraints", "abstract": "  We introduce a learning-based depth map fusion framework that accepts a set\nof depth and confidence maps generated by a Multi-View Stereo (MVS) algorithm\nas input and improves them. This is accomplished by integrating volumetric\nvisibility constraints that encode long-range surface relationships across\ndifferent views into an end-to-end trainable architecture. We also introduce a\ndepth search window estimation sub-network trained jointly with the larger\nfusion sub-network to reduce the depth hypothesis search space along each ray.\nOur method learns to model depth consensus and violations of visibility\nconstraints directly from the data; effectively removing the necessity of\nfine-tuning fusion parameters. Extensive experiments on MVS datasets show\nsubstantial improvements in the accuracy of the output fused depth and\nconfidence maps.\n", "date": "2023-08-17", "categories": []}
{"id": "2308.08830", "title": "ICoNIK: Generating Respiratory-Resolved Abdominal MR Reconstructions\n  Using Neural Implicit Representations in k-Space", "abstract": "  Motion-resolved reconstruction for abdominal magnetic resonance imaging (MRI)\nremains a challenge due to the trade-off between residual motion blurring\ncaused by discretized motion states and undersampling artefacts. In this work,\nwe propose to generate blurring-free motion-resolved abdominal reconstructions\nby learning a neural implicit representation directly in k-space (NIK). Using\nmeasured sampling points and a data-derived respiratory navigator signal, we\ntrain a network to generate continuous signal values. To aid the regularization\nof sparsely sampled regions, we introduce an additional informed correction\nlayer (ICo), which leverages information from neighboring regions to correct\nNIK's prediction. Our proposed generative reconstruction methods, NIK and\nICoNIK, outperform standard motion-resolved reconstruction techniques and\nprovide a promising solution to address motion artefacts in abdominal MRI.\n", "date": "2023-08-17", "categories": []}
{"id": "2308.09467", "title": "Quantitative Susceptibility Mapping through Model-based Deep Image Prior\n  (MoDIP)", "abstract": "  The data-driven approach of supervised learning methods has limited\napplicability in solving dipole inversion in Quantitative Susceptibility\nMapping (QSM) with varying scan parameters across different objects. To address\nthis generalization issue in supervised QSM methods, we propose a novel\ntraining-free model-based unsupervised method called MoDIP (Model-based Deep\nImage Prior). MoDIP comprises a small, untrained network and a Data Fidelity\nOptimization (DFO) module. The network converges to an interim state, acting as\nan implicit prior for image regularization, while the optimization process\nenforces the physical model of QSM dipole inversion. Experimental results\ndemonstrate MoDIP's excellent generalizability in solving QSM dipole inversion\nacross different scan parameters. It exhibits robustness against pathological\nbrain QSM, achieving over 32% accuracy improvement than supervised deep\nlearning and traditional iterative methods. It is also 33% more computationally\nefficient and runs 4 times faster than conventional DIP-based approaches,\nenabling 3D high-resolution image reconstruction in under 4.5 minutes.\n", "date": "2023-08-18", "categories": []}
{"id": "2308.08094", "title": "Snapshot High Dynamic Range Imaging with a Polarization Camera", "abstract": "  High dynamic range (HDR) images are important for a range of tasks, from\nnavigation to consumer photography. Accordingly, a host of specialized HDR\nsensors have been developed, the most successful of which are based on\ncapturing variable per-pixel exposures. In essence, these methods capture an\nentire exposure bracket sequence at once in a single shot. This paper presents\na straightforward but highly effective approach for turning an off-the-shelf\npolarization camera into a high-performance HDR camera. By placing a linear\npolarizer in front of the polarization camera, we are able to simultaneously\ncapture four images with varied exposures, which are determined by the\norientation of the polarizer. We develop an outlier-robust and self-calibrating\nalgorithm to reconstruct an HDR image (at a single polarity) from these\nmeasurements. Finally, we demonstrate the efficacy of our approach with\nextensive real-world experiments.\n", "date": "2023-08-16", "categories": []}
{"id": "2303.18125", "title": "Towards Nonlinear-Motion-Aware and Occlusion-Robust Rolling Shutter\n  Correction", "abstract": "  This paper addresses the problem of rolling shutter correction in complex\nnonlinear and dynamic scenes with extreme occlusion. Existing methods suffer\nfrom two main drawbacks. Firstly, they face challenges in estimating the\naccurate correction field due to the uniform velocity assumption, leading to\nsignificant image correction errors under complex motion. Secondly, the drastic\nocclusion in dynamic scenes prevents current solutions from achieving better\nimage quality because of the inherent difficulties in aligning and aggregating\nmultiple frames. To tackle these challenges, we model the curvilinear\ntrajectory of pixels analytically and propose a geometry-based Quadratic\nRolling Shutter (QRS) motion solver, which precisely estimates the high-order\ncorrection field of individual pixels. Besides, to reconstruct high-quality\nocclusion frames in dynamic scenes, we present a 3D video architecture that\neffectively Aligns and Aggregates multi-frame context, namely, RSA2-Net. We\nevaluate our method across a broad range of cameras and video sequences,\ndemonstrating its significant superiority. Specifically, our method surpasses\nthe state-of-the-art by +4.98, +0.77, and +4.33 of PSNR on Carla-RS, Fastec-RS,\nand BS-RSC datasets, respectively. Code is available at\nhttps://github.com/DelinQu/qrsc.\n", "date": "2023-03-31", "categories": []}
{"id": "2207.02399", "title": "Learning Apparent Diffusion Coefficient Maps from Accelerated Radial\n  k-Space Diffusion-Weighted MRI in Mice using a Deep CNN-Transformer Model", "abstract": "  Purpose: To accelerate radially sampled diffusion weighted spin-echo\n(Rad-DW-SE) acquisition method for generating high quality apparent diffusion\ncoefficient (ADC) maps. Methods: A deep learning method was developed to\ngenerate accurate ADC maps from accelerated DWI data acquired with the\nRad-DW-SE method. The deep learning method integrates convolutional neural\nnetworks (CNNs) with vision transformers to generate high quality ADC maps from\naccelerated DWI data, regularized by a monoexponential ADC model fitting term.\nA model was trained on DWI data of 147 mice and evaluated on DWI data of 36\nmice, with acceleration factors of 4x and 8x compared to the original\nacquisition parameters. We have made our code publicly available at GitHub:\nhttps://github.com/ymli39/DeepADC-Net-Learning-Apparent-Diffusion-Coefficient-Maps,\nand our dataset can be downloaded at\nhttps://pennpancreaticcancerimagingresource.github.io/data.html. Results:\nAblation studies and experimental results have demonstrated that the proposed\ndeep learning model generates higher quality ADC maps from accelerated DWI data\nthan alternative deep learning methods under comparison when their performance\nis quantified in whole images as well as in regions of interest, including\ntumors, kidneys, and muscles. Conclusions: The deep learning method with\nintegrated CNNs and transformers provides an effective means to accurately\ncompute ADC maps from accelerated DWI data acquired with the Rad-DW-SE method.\n", "date": "2022-07-06", "categories": []}
{"id": "2305.01643", "title": "Neural LiDAR Fields for Novel View Synthesis", "abstract": "  We present Neural Fields for LiDAR (NFL), a method to optimise a neural field\nscene representation from LiDAR measurements, with the goal of synthesizing\nrealistic LiDAR scans from novel viewpoints. NFL combines the rendering power\nof neural fields with a detailed, physically motivated model of the LiDAR\nsensing process, thus enabling it to accurately reproduce key sensor behaviors\nlike beam divergence, secondary returns, and ray dropping. We evaluate NFL on\nsynthetic and real LiDAR scans and show that it outperforms explicit\nreconstruct-then-simulate methods as well as other NeRF-style methods on LiDAR\nnovel view synthesis task. Moreover, we show that the improved realism of the\nsynthesized views narrows the domain gap to real scans and translates to better\nregistration and semantic segmentation performance.\n", "date": "2023-05-02", "categories": []}
{"id": "2305.09533", "title": "NightHazeFormer: Single Nighttime Haze Removal Using Prior Query\n  Transformer", "abstract": "  Nighttime image dehazing is a challenging task due to the presence of\nmultiple types of adverse degrading effects including glow, haze, blurry,\nnoise, color distortion, and so on. However, most previous studies mainly focus\non daytime image dehazing or partial degradations presented in nighttime hazy\nscenes, which may lead to unsatisfactory restoration results. In this paper, we\npropose an end-to-end transformer-based framework for nighttime haze removal,\ncalled NightHazeFormer. Our proposed approach consists of two stages:\nsupervised pre-training and semi-supervised fine-tuning. During the\npre-training stage, we introduce two powerful priors into the transformer\ndecoder to generate the non-learnable prior queries, which guide the model to\nextract specific degradations. For the fine-tuning, we combine the generated\npseudo ground truths with input real-world nighttime hazy images as paired\nimages and feed into the synthetic domain to fine-tune the pre-trained model.\nThis semi-supervised fine-tuning paradigm helps improve the generalization to\nreal domain. In addition, we also propose a large-scale synthetic dataset\ncalled UNREAL-NH, to simulate the real-world nighttime haze scenarios\ncomprehensively. Extensive experiments on several synthetic and real-world\ndatasets demonstrate the superiority of our NightHazeFormer over\nstate-of-the-art nighttime haze removal methods in terms of both visually and\nquantitatively.\n", "date": "2023-05-16", "categories": []}
{"id": "2308.02968", "title": "Robust estimation of exposure ratios in multi-exposure image stacks", "abstract": "  Merging multi-exposure image stacks into a high dynamic range (HDR) image\nrequires knowledge of accurate exposure times. When exposure times are\ninaccurate, for example, when they are extracted from a camera's EXIF metadata,\nthe reconstructed HDR images reveal banding artifacts at smooth gradients. To\nremedy this, we propose to estimate exposure ratios directly from the input\nimages. We derive the exposure time estimation as an optimization problem, in\nwhich pixels are selected from pairs of exposures to minimize estimation error\ncaused by camera noise. When pixel values are represented in the logarithmic\ndomain, the problem can be solved efficiently using a linear solver. We\ndemonstrate that the estimation can be easily made robust to pixel misalignment\ncaused by camera or object motion by collecting pixels from multiple spatial\ntiles. The proposed automatic exposure estimation and alignment eliminates\nbanding artifacts in popular datasets and is essential for applications that\nrequire physically accurate reconstructions, such as measuring the modulation\ntransfer function of a display. The code for the method is available.\n", "date": "2023-08-05", "categories": []}
{"id": "2308.06998", "title": "Mutual Information-driven Triple Interaction Network for Efficient Image\n  Dehazing", "abstract": "  Multi-stage architectures have exhibited efficacy in image dehazing, which\nusually decomposes a challenging task into multiple more tractable sub-tasks\nand progressively estimates latent hazy-free images. Despite the remarkable\nprogress, existing methods still suffer from the following shortcomings: (1)\nlimited exploration of frequency domain information; (2) insufficient\ninformation interaction; (3) severe feature redundancy. To remedy these issues,\nwe propose a novel Mutual Information-driven Triple interaction Network\n(MITNet) based on spatial-frequency dual domain information and two-stage\narchitecture. To be specific, the first stage, named amplitude-guided haze\nremoval, aims to recover the amplitude spectrum of the hazy images for haze\nremoval. And the second stage, named phase-guided structure refined, devotes to\nlearning the transformation and refinement of the phase spectrum. To facilitate\nthe information exchange between two stages, an Adaptive Triple Interaction\nModule (ATIM) is developed to simultaneously aggregate cross-domain,\ncross-scale, and cross-stage features, where the fused features are further\nused to generate content-adaptive dynamic filters so that applying them to\nenhance global context representation. In addition, we impose the mutual\ninformation minimization constraint on paired scale encoder and decoder\nfeatures from both stages. Such an operation can effectively reduce information\nredundancy and enhance cross-stage feature complementarity. Extensive\nexperiments on multiple public datasets exhibit that our MITNet performs\nsuperior performance with lower model complexity.The code and models are\navailable at https://github.com/it-hao/MITNet.\n", "date": "2023-08-14", "categories": []}
{"id": "2308.07228", "title": "RestoreFormer++: Towards Real-World Blind Face Restoration from\n  Undegraded Key-Value Pairs", "abstract": "  Blind face restoration aims at recovering high-quality face images from those\nwith unknown degradations. Current algorithms mainly introduce priors to\ncomplement high-quality details and achieve impressive progress. However, most\nof these algorithms ignore abundant contextual information in the face and its\ninterplay with the priors, leading to sub-optimal performance. Moreover, they\npay less attention to the gap between the synthetic and real-world scenarios,\nlimiting the robustness and generalization to real-world applications. In this\nwork, we propose RestoreFormer++, which on the one hand introduces\nfully-spatial attention mechanisms to model the contextual information and the\ninterplay with the priors, and on the other hand, explores an extending\ndegrading model to help generate more realistic degraded face images to\nalleviate the synthetic-to-real-world gap. Compared with current algorithms,\nRestoreFormer++ has several crucial benefits. First, instead of using a\nmulti-head self-attention mechanism like the traditional visual transformer, we\nintroduce multi-head cross-attention over multi-scale features to fully explore\nspatial interactions between corrupted information and high-quality priors. In\nthis way, it can facilitate RestoreFormer++ to restore face images with higher\nrealness and fidelity. Second, in contrast to the recognition-oriented\ndictionary, we learn a reconstruction-oriented dictionary as priors, which\ncontains more diverse high-quality facial details and better accords with the\nrestoration target. Third, we introduce an extending degrading model that\ncontains more realistic degraded scenarios for training data synthesizing, and\nthus helps to enhance the robustness and generalization of our RestoreFormer++\nmodel. Extensive experiments show that RestoreFormer++ outperforms\nstate-of-the-art algorithms on both synthetic and real-world datasets.\n", "date": "2023-08-14", "categories": []}
{"id": "2209.05483", "title": "Self-Supervised Coordinate Projection Network for Sparse-View Computed\n  Tomography", "abstract": "  In the present work, we propose a Self-supervised COordinate Projection\nnEtwork (SCOPE) to reconstruct the artifacts-free CT image from a single SV\nsinogram by solving the inverse tomography imaging problem. Compared with\nrecent related works that solve similar problems using implicit neural\nrepresentation network (INR), our essential contribution is an effective and\nsimple re-projection strategy that pushes the tomography image reconstruction\nquality over supervised deep learning CT reconstruction works. The proposed\nstrategy is inspired by the simple relationship between linear algebra and\ninverse problems. To solve the under-determined linear equation system, we\nfirst introduce INR to constrain the solution space via image continuity prior\nand achieve a rough solution. And secondly, we propose to generate a dense view\nsinogram that improves the rank of the linear equation system and produces a\nmore stable CT image solution space. Our experiment results demonstrate that\nthe re-projection strategy significantly improves the image reconstruction\nquality (+3 dB for PSNR at least). Besides, we integrate the recent hash\nencoding into our SCOPE model, which greatly accelerates the model training.\nFinally, we evaluate SCOPE in parallel and fan X-ray beam SVCT reconstruction\ntasks. Experimental results indicate that the proposed SCOPE model outperforms\ntwo latest INR-based methods and two well-popular supervised DL methods\nquantitatively and qualitatively.\n", "date": "2022-09-12", "categories": []}
{"id": "2301.08365", "title": "On Retrospective k-space Subsampling schemes For Deep MRI Reconstruction", "abstract": "  Acquiring fully-sampled MRI $k$-space data is time-consuming, and collecting\naccelerated data can reduce the acquisition time. Employing 2D\nCartesian-rectilinear subsampling schemes is a conventional approach for\naccelerated acquisitions; however, this often results in imprecise\nreconstructions, even with the use of Deep Learning (DL), especially at high\nacceleration factors. Non-rectilinear or non-Cartesian trajectories can be\nimplemented in MRI scanners as alternative subsampling options. This work\ninvestigates the impact of the $k$-space subsampling scheme on the quality of\nreconstructed accelerated MRI measurements produced by trained DL models. The\nRecurrent Variational Network (RecurrentVarNet) was used as the DL-based\nMRI-reconstruction architecture. Cartesian, fully-sampled multi-coil $k$-space\nmeasurements from three datasets were retrospectively subsampled with different\naccelerations using eight distinct subsampling schemes: four\nCartesian-rectilinear, two Cartesian non-rectilinear, and two non-Cartesian.\nExperiments were conducted in two frameworks: scheme-specific, where a distinct\nmodel was trained and evaluated for each dataset-subsampling scheme pair, and\nmulti-scheme, where for each dataset a single model was trained on data\nrandomly subsampled by any of the eight schemes and evaluated on data\nsubsampled by all schemes. In both frameworks, RecurrentVarNets trained and\nevaluated on non-rectilinearly subsampled data demonstrated superior\nperformance, particularly for high accelerations. In the multi-scheme setting,\nreconstruction performance on rectilinearly subsampled data improved when\ncompared to the scheme-specific experiments. Our findings demonstrate the\npotential for using DL-based methods, trained on non-rectilinearly subsampled\nmeasurements, to optimize scan time and image quality.\n", "date": "2023-01-20", "categories": []}
{"id": "2308.05365", "title": "TriDo-Former: A Triple-Domain Transformer for Direct PET Reconstruction\n  from Low-Dose Sinograms", "abstract": "  To obtain high-quality positron emission tomography (PET) images while\nminimizing radiation exposure, various methods have been proposed for\nreconstructing standard-dose PET (SPET) images from low-dose PET (LPET)\nsinograms directly. However, current methods often neglect boundaries during\nsinogram-to-image reconstruction, resulting in high-frequency distortion in the\nfrequency domain and diminished or fuzzy edges in the reconstructed images.\nFurthermore, the convolutional architectures, which are commonly used, lack the\nability to model long-range non-local interactions, potentially leading to\ninaccurate representations of global structures. To alleviate these problems,\nwe propose a transformer-based model that unites triple domains of sinogram,\nimage, and frequency for direct PET reconstruction, namely TriDo-Former.\nSpecifically, the TriDo-Former consists of two cascaded networks, i.e., a\nsinogram enhancement transformer (SE-Former) for denoising the input LPET\nsinograms and a spatial-spectral reconstruction transformer (SSR-Former) for\nreconstructing SPET images from the denoised sinograms. Different from the\nvanilla transformer that splits an image into 2D patches, based specifically on\nthe PET imaging mechanism, our SE-Former divides the sinogram into 1D\nprojection view angles to maintain its inner-structure while denoising,\npreventing the noise in the sinogram from prorogating into the image domain.\nMoreover, to mitigate high-frequency distortion and improve reconstruction\ndetails, we integrate global frequency parsers (GFPs) into SSR-Former. The GFP\nserves as a learnable frequency filter that globally adjusts the frequency\ncomponents in the frequency domain, enforcing the network to restore\nhigh-frequency details resembling real SPET images. Validations on a clinical\ndataset demonstrate that our TriDo-Former outperforms the state-of-the-art\nmethods qualitatively and quantitatively.\n", "date": "2023-08-10", "categories": []}
{"id": "2308.05447", "title": "A Generalized Physical-knowledge-guided Dynamic Model for Underwater\n  Image Enhancement", "abstract": "  Underwater images often suffer from color distortion and low contrast\nresulting in various image types, due to the scattering and absorption of light\nby water. While it is difficult to obtain high-quality paired training samples\nwith a generalized model. To tackle these challenges, we design a Generalized\nUnderwater image enhancement method via a Physical-knowledge-guided Dynamic\nModel (short for GUPDM), consisting of three parts: Atmosphere-based Dynamic\nStructure (ADS), Transmission-guided Dynamic Structure (TDS), and Prior-based\nMulti-scale Structure (PMS). In particular, to cover complex underwater scenes,\nthis study changes the global atmosphere light and the transmission to simulate\nvarious underwater image types (e.g., the underwater image color ranging from\nyellow to blue) through the formation model. We then design ADS and TDS that\nuse dynamic convolutions to adaptively extract prior information from\nunderwater images and generate parameters for PMS. These two modules enable the\nnetwork to select appropriate parameters for various water types adaptively.\nBesides, the multi-scale feature extraction module in PMS uses convolution\nblocks with different kernel sizes and obtains weights for each feature map via\nchannel attention block and fuses them to boost the receptive field of the\nnetwork. The source code will be available at\n\\href{https://github.com/shiningZZ/GUPDM}{https://github.com/shiningZZ/GUPDM}.\n", "date": "2023-08-10", "categories": []}
{"id": "2308.05449", "title": "Transforming Breast Cancer Diagnosis: Towards Real-Time Ultrasound to\n  Mammogram Conversion for Cost-Effective Diagnosis", "abstract": "  Ultrasound (US) imaging is better suited for intraoperative settings because\nit is real-time and more portable than other imaging techniques, such as\nmammography. However, US images are characterized by lower spatial resolution\nnoise-like artifacts. This research aims to address these limitations by\nproviding surgeons with mammogram-like image quality in real-time from noisy US\nimages. Unlike previous approaches for improving US image quality that aim to\nreduce artifacts by treating them as (speckle noise), we recognize their value\nas informative wave interference pattern (WIP). To achieve this, we utilize the\nStride software to numerically solve the forward model, generating ultrasound\nimages from mammograms images by solving wave-equations. Additionally, we\nleverage the power of domain adaptation to enhance the realism of the simulated\nultrasound images. Then, we utilize generative adversarial networks (GANs) to\ntackle the inverse problem of generating mammogram-quality images from\nultrasound images. The resultant images have considerably more discernible\ndetails than the original US images.\n", "date": "2023-08-10", "categories": []}
{"id": "2308.05745", "title": "Iterative Reweighted Least Squares Networks With Convergence Guarantees\n  for Solving Inverse Imaging Problems", "abstract": "  In this work we present a novel optimization strategy for image\nreconstruction tasks under analysis-based image regularization, which promotes\nsparse and/or low-rank solutions in some learned transform domain. We\nparameterize such regularizers using potential functions that correspond to\nweighted extensions of the $\\ell_p^p$-vector and $\\mathcal{S}_p^p$\nSchatten-matrix quasi-norms with $0 < p \\le 1$. Our proposed minimization\nstrategy extends the Iteratively Reweighted Least Squares (IRLS) method,\ntypically used for synthesis-based $\\ell_p$ and $\\mathcal{S}_p$ norm and\nanalysis-based $\\ell_1$ and nuclear norm regularization. We prove that under\nmild conditions our minimization algorithm converges linearly to a stationary\npoint, and we provide an upper bound for its convergence rate. Further, to\nselect the parameters of the regularizers that deliver the best results for the\nproblem at hand, we propose to learn them from training data by formulating the\nsupervised learning process as a stochastic bilevel optimization problem. We\nshow that thanks to the convergence guarantees of our proposed minimization\nstrategy, such optimization can be successfully performed with a\nmemory-efficient implicit back-propagation scheme. We implement our learned\nIRLS variants as recurrent networks and assess their performance on the\nchallenging image reconstruction tasks of non-blind deblurring,\nsuper-resolution and demosaicking. The comparisons against other existing\nlearned reconstruction approaches demonstrate that our overall method is very\ncompetitive and in many cases outperforms existing unrolled networks, whose\nnumber of parameters is orders of magnitude higher than in our case.\n", "date": "2023-08-10", "categories": []}
{"id": "2302.06436", "title": "Geometric Constraints Enable Self-Supervised Sinogram Inpainting in\n  Sparse-View Tomography", "abstract": "  The diagnostic quality of computed tomography (CT) scans is usually\nrestricted by the induced patient dose, scan speed, and image quality.\nSparse-angle tomographic scans reduce radiation exposure and accelerate data\nacquisition, but suffer from image artifacts and noise. Existing image\nprocessing algorithms can restore CT reconstruction quality but often require\nlarge training data sets or can not be used for truncated objects. This work\npresents a self-supervised projection inpainting method that allows optimizing\nmissing projective views via gradient-based optimization. By reconstructing\nindependent stacks of projection data, a self-supervised loss is calculated in\nthe CT image domain and used to directly optimize projection image intensities\nto match the missing tomographic views constrained by the projection geometry.\nOur experiments on real X-ray microscope (XRM) tomographic mouse tibia bone\nscans show that our method improves reconstructions by 3.1-7.4%/7.7-17.6% in\nterms of PSNR/SSIM with respect to the interpolation baseline. Our approach is\napplicable as a flexible self-supervised projection inpainting tool for\ntomographic applications.\n", "date": "2023-02-13", "categories": []}
{"id": "2308.04682", "title": "Score Priors Guided Deep Variational Inference for Unsupervised\n  Real-World Single Image Denoising", "abstract": "  Real-world single image denoising is crucial and practical in computer\nvision. Bayesian inversions combined with score priors now have proven\neffective for single image denoising but are limited to white Gaussian noise.\nMoreover, applying existing score-based methods for real-world denoising\nrequires not only the explicit train of score priors on the target domain but\nalso the careful design of sampling procedures for posterior inference, which\nis complicated and impractical. To address these limitations, we propose a\nscore priors-guided deep variational inference, namely ScoreDVI, for practical\nreal-world denoising. By considering the deep variational image posterior with\na Gaussian form, score priors are extracted based on easily accessible minimum\nMSE Non-$i.i.d$ Gaussian denoisers and variational samples, which in turn\nfacilitate optimizing the variational image posterior. Such a procedure\nadaptively applies cheap score priors to denoising. Additionally, we exploit a\nNon-$i.i.d$ Gaussian mixture model and variational noise posterior to model the\nreal-world noise. This scheme also enables the pixel-wise fusion of multiple\nimage priors and variational image posteriors. Besides, we develop a\nnoise-aware prior assignment strategy that dynamically adjusts the weight of\nimage priors in the optimization. Our method outperforms other single\nimage-based real-world denoising methods and achieves comparable performance to\ndataset-based unsupervised methods.\n", "date": "2023-08-09", "categories": []}
{"id": "2308.04821", "title": "HyperCoil-Recon: A Hypernetwork-based Adaptive Coil Configuration Task\n  Switching Network for MRI Reconstruction", "abstract": "  Parallel imaging, a fast MRI technique, involves dynamic adjustments based on\nthe configuration i.e. number, positioning, and sensitivity of the coils with\nrespect to the anatomy under study. Conventional deep learning-based image\nreconstruction models have to be trained or fine-tuned for each configuration,\nposing a barrier to clinical translation, given the lack of computational\nresources and machine learning expertise for clinicians to train models at\ndeployment. Joint training on diverse datasets learns a single weight set that\nmight underfit to deviated configurations. We propose, HyperCoil-Recon, a\nhypernetwork-based coil configuration task-switching network for multi-coil MRI\nreconstruction that encodes varying configurations of the numbers of coils in a\nmulti-tasking perspective, posing each configuration as a task. The\nhypernetworks infer and embed task-specific weights into the reconstruction\nnetwork, 1) effectively utilizing the contextual knowledge of common and\nvarying image features among the various fields-of-view of the coils, and 2)\nenabling generality to unseen configurations at test time. Experiments reveal\nthat our approach 1) adapts on the fly to various unseen configurations up to\n32 coils when trained on lower numbers (i.e. 7 to 11) of randomly varying\ncoils, and to 120 deviated unseen configurations when trained on 18\nconfigurations in a single model, 2) matches the performance of coil\nconfiguration-specific models, and 3) outperforms configuration-invariant\nmodels with improvement margins of around 1 dB / 0.03 and 0.3 dB / 0.02 in PSNR\n/ SSIM for knee and brain data. Our code is available at\nhttps://github.com/sriprabhar/HyperCoil-Recon\n", "date": "2023-08-09", "categories": []}
{"id": "2304.11141", "title": "H2TF for Hyperspectral Image Denoising: Where Hierarchical Nonlinear\n  Transform Meets Hierarchical Matrix Factorization", "abstract": "  Recently, tensor singular value decomposition (t-SVD) has emerged as a\npromising tool for hyperspectral image (HSI) processing. In the t-SVD, there\nare two key building blocks: (i) the low-rank enhanced transform and (ii) the\naccompanying low-rank characterization of transformed frontal slices. Previous\nt-SVD methods mainly focus on the developments of (i), while neglecting the\nother important aspect, i.e., the exact characterization of transformed frontal\nslices. In this letter, we exploit the potentiality in both building blocks by\nleveraging the \\underline{\\bf H}ierarchical nonlinear transform and the\n\\underline{\\bf H}ierarchical matrix factorization to establish a new\n\\underline{\\bf T}ensor \\underline{\\bf F}actorization (termed as H2TF). Compared\nto shallow counter partners, e.g., low-rank matrix factorization or its convex\nsurrogates, H2TF can better capture complex structures of transformed frontal\nslices due to its hierarchical modeling abilities. We then suggest the\nH2TF-based HSI denoising model and develop an alternating direction method of\nmultipliers-based algorithm to address the resultant model. Extensive\nexperiments validate the superiority of our method over state-of-the-art HSI\ndenoising methods.\n", "date": "2023-04-21", "categories": []}
{"id": "2308.02797", "title": "Thin On-Sensor Nanophotonic Array Cameras", "abstract": "  Today's commodity camera systems rely on compound optics to map light\noriginating from the scene to positions on the sensor where it gets recorded as\nan image. To record images without optical aberrations, i.e., deviations from\nGauss' linear model of optics, typical lens systems introduce increasingly\ncomplex stacks of optical elements which are responsible for the height of\nexisting commodity cameras. In this work, we investigate flat nanophotonic\ncomputational cameras as an alternative that employs an array of skewed\nlenslets and a learned reconstruction approach. The optical array is embedded\non a metasurface that, at 700~nm height, is flat and sits on the sensor cover\nglass at 2.5~mm focal distance from the sensor. To tackle the highly chromatic\nresponse of a metasurface and design the array over the entire sensor, we\npropose a differentiable optimization method that continuously samples over the\nvisible spectrum and factorizes the optical modulation for different incident\nfields into individual lenses. We reconstruct a megapixel image from our flat\nimager with a learned probabilistic reconstruction method that employs a\ngenerative diffusion model to sample an implicit prior. To tackle\nscene-dependent aberrations in broadband, we propose a method for acquiring\npaired captured training data in varying illumination conditions. We assess the\nproposed flat camera design in simulation and with an experimental prototype,\nvalidating that the method is capable of recovering images from diverse scenes\nin broadband with a single nanophotonic layer.\n", "date": "2023-08-05", "categories": []}
{"id": "2308.04163", "title": "Under-Display Camera Image Restoration with Scattering Effect", "abstract": "  The under-display camera (UDC) provides consumers with a full-screen visual\nexperience without any obstruction due to notches or punched holes. However,\nthe semi-transparent nature of the display inevitably introduces the severe\ndegradation into UDC images. In this work, we address the UDC image restoration\nproblem with the specific consideration of the scattering effect caused by the\ndisplay. We explicitly model the scattering effect by treating the display as a\npiece of homogeneous scattering medium. With the physical model of the\nscattering effect, we improve the image formation pipeline for the image\nsynthesis to construct a realistic UDC dataset with ground truths. To suppress\nthe scattering effect for the eventual UDC image recovery, a two-branch\nrestoration network is designed. More specifically, the scattering branch\nleverages global modeling capabilities of the channel-wise self-attention to\nestimate parameters of the scattering effect from degraded images. While the\nimage branch exploits the local representation advantage of CNN to recover\nclear scenes, implicitly guided by the scattering branch. Extensive experiments\nare conducted on both real-world and synthesized data, demonstrating the\nsuperiority of the proposed method over the state-of-the-art UDC restoration\ntechniques. The source code and dataset are available at\n\\url{https://github.com/NamecantbeNULL/SRUDC}.\n", "date": "2023-08-08", "categories": []}
{"id": "2308.04207", "title": "Robust retrieval of material chemical states in X-ray microspectroscopy", "abstract": "  X-ray microspectroscopic techniques are essential for studying morphological\nand chemical changes in materials, providing high-resolution structural and\nspectroscopic information. However, its practical data analysis for reliably\nretrieving the chemical states remains a major obstacle to accelerating the\nfundamental understanding of materials in many research fields. In this work,\nwe propose a novel data formulation model for X-ray microspectroscopy and\ndevelop a dedicated unmixing framework to solve this problem, which is robust\nto noise and spectral variability. Moreover, this framework is not limited to\nthe analysis of two-state material chemistry, making it an effective\nalternative to conventional and widely-used methods. In addition, an\nalternative directional multiplier method with provable convergence is applied\nto obtain the solution efficiently. Our framework can accurately identify and\ncharacterize chemical states in complex and heterogeneous samples, even under\nchallenging conditions such as low signal-to-noise ratios and overlapping\nspectral features. Extensive experimental results on simulated and real\ndatasets demonstrate its effectiveness and reliability.\n", "date": "2023-08-08", "categories": []}
{"id": "2207.11297", "title": "Accelerated and Quantitative 3D Semisolid MT/CEST Imaging using a\n  Generative Adversarial Network (GAN-CEST)", "abstract": "  Purpose: To substantially shorten the acquisition time required for\nquantitative 3D chemical exchange saturation transfer (CEST) and semisolid\nmagnetization transfer (MT) imaging and allow for rapid chemical exchange\nparameter map reconstruction. Methods: Three-dimensional CEST and MT magnetic\nresonance fingerprinting (MRF) datasets of L-arginine phantoms, whole-brains,\nand calf muscles from healthy volunteers, cancer patients, and cardiac patients\nwere acquired using 3T clinical scanners at 3 different sites, using 3\ndifferent scanner models and coils. A generative adversarial network supervised\nframework (GAN-CEST) was then designed and trained to learn the mapping from a\nreduced input data space to the quantitative exchange parameter space, while\npreserving perceptual and quantitative content. Results: The GAN-CEST 3D\nacquisition time was 42-52 seconds, 70% shorter than CEST-MRF. The quantitative\nreconstruction of the entire brain took 0.8 seconds. An excellent agreement was\nobserved between the ground truth and GAN-based L-arginine concentration and pH\nvalues (Pearson's r > 0.97, NRMSE < 1.5%). GAN-CEST images from a brain-tumor\nsubject yielded a semi-solid volume fraction and exchange rate NRMSE of\n3.8$\\pm$1.3% and 4.6$\\pm$1.3%, respectively, and SSIM of 96.3$\\pm$1.6% and\n95.0$\\pm$2.4%, respectively. The mapping of the calf-muscle exchange parameters\nin a cardiac patient, yielded NRMSE < 7% and SSIM > 94% for the semi-solid\nexchange parameters. In regions with large susceptibility artifacts, GAN-CEST\nhas demonstrated improved performance and reduced noise compared to MRF.\nConclusion: GAN-CEST can substantially reduce the acquisition time for\nquantitative semisolid MT/CEST mapping, while retaining performance even when\nfacing pathologies and scanner models that were not available during training.\n", "date": "2022-07-22", "categories": []}
{"id": "2209.08288", "title": "Self-supervised learning of hologram reconstruction using physics\n  consistency", "abstract": "  The past decade has witnessed transformative applications of deep learning in\nvarious computational imaging, sensing and microscopy tasks. Due to the\nsupervised learning schemes employed, these methods mostly depend on\nlarge-scale, diverse, and labeled training data. The acquisition and\npreparation of such training image datasets are often laborious and costly,\nalso leading to biased estimation and limited generalization to new sample\ntypes. Here, we report a self-supervised learning model, termed GedankenNet,\nthat eliminates the need for labeled or experimental training data, and\ndemonstrate its effectiveness and superior generalization on hologram\nreconstruction tasks. Without prior knowledge about the sample types to be\nimaged, the self-supervised learning model was trained using a\nphysics-consistency loss and artificial random images that are synthetically\ngenerated without any experiments or resemblance to real-world samples. After\nits self-supervised training, GedankenNet successfully generalized to\nexperimental holograms of various unseen biological samples, reconstructing the\nphase and amplitude images of different types of objects using experimentally\nacquired test holograms. Without access to experimental data or knowledge of\nreal samples of interest or their spatial features, GedankenNet's\nself-supervised learning achieved complex-valued image reconstructions that are\nconsistent with the Maxwell's equations, and its output inference and object\nsolutions accurately represent the wave propagation in free-space. GedankenNet\nframework also exhibits resilience to random, unknown perturbations in the\nphysical forward model, including changes in the hologram distances, pixel size\nand illumination wavelength. This self-supervised learning of image\nreconstruction tasks creates new opportunities for various inverse problems in\nholography, microscopy and computational imaging fields.\n", "date": "2022-09-17", "categories": []}
{"id": "2211.10103", "title": "Let's Enhance: A Deep Learning Approach to Extreme Deblurring of Text\n  Images", "abstract": "  This work presents a novel deep-learning-based pipeline for the inverse\nproblem of image deblurring, leveraging augmentation and pre-training with\nsynthetic data. Our results build on our winning submission to the recent\nHelsinki Deblur Challenge 2021, whose goal was to explore the limits of\nstate-of-the-art deblurring algorithms in a real-world data setting. The task\nof the challenge was to deblur out-of-focus images of random text, thereby in a\ndownstream task, maximizing an optical-character-recognition-based score\nfunction. A key step of our solution is the data-driven estimation of the\nphysical forward model describing the blur process. This enables a stream of\nsynthetic data, generating pairs of ground-truth and blurry images on-the-fly,\nwhich is used for an extensive augmentation of the small amount of challenge\ndata provided. The actual deblurring pipeline consists of an approximate\ninversion of the radial lens distortion (determined by the estimated forward\nmodel) and a U-Net architecture, which is trained end-to-end. Our algorithm was\nthe only one passing the hardest challenge level, achieving over $70\\%$\ncharacter recognition accuracy. Our findings are well in line with the paradigm\nof data-centric machine learning, and we demonstrate its effectiveness in the\ncontext of inverse problems. Apart from a detailed presentation of our\nmethodology, we also analyze the importance of several design choices in a\nseries of ablation studies. The code of our challenge submission is available\nunder https://github.com/theophil-trippe/HDC_TUBerlin_version_1.\n", "date": "2022-11-18", "categories": []}
{"id": "2211.14751", "title": "Estimating Reflectance Layer from A Single Image: Integrating\n  Reflectance Guidance and Shadow/Specular Aware Learning", "abstract": "  Estimating the reflectance layer from a single image is a challenging task.\nIt becomes more challenging when the input image contains shadows or specular\nhighlights, which often render an inaccurate estimate of the reflectance layer.\nTherefore, we propose a two-stage learning method, including reflectance\nguidance and a Shadow/Specular-Aware (S-Aware) network to tackle the problem.\nIn the first stage, an initial reflectance layer free from shadows and\nspecularities is obtained with the constraint of novel losses that are guided\nby prior-based shadow-free and specular-free images. To further enforce the\nreflectance layer to be independent of shadows and specularities in the\nsecond-stage refinement, we introduce an S-Aware network that distinguishes the\nreflectance image from the input image. Our network employs a classifier to\ncategorize shadow/shadow-free, specular/specular-free classes, enabling the\nactivation features to function as attention maps that focus on shadow/specular\nregions. Our quantitative and qualitative evaluations show that our method\noutperforms the state-of-the-art methods in the reflectance layer estimation\nthat is free from shadows and specularities. Code is at:\n\\url{https://github.com/jinyeying/S-Aware-network}.\n", "date": "2022-11-27", "categories": []}
{"id": "2308.02714", "title": "Exploring the Effect of Sparse Recovery on the Quality of Image\n  Superresolution", "abstract": "  Dictionary learning can be used for image superresolution by learning a pair\nof coupled dictionaries of image patches from high-resolution and\nlow-resolution image pairs such that the corresponding pairs share the same\nsparse vector when represented by the coupled dictionaries. These dictionaries\nthen can be used to to reconstruct the corresponding high-resolution patches\nfrom low-resolution input images based on sparse recovery. The idea is to\nrecover the shared sparse vector using the low-resolution dictionary and then\nmultiply it by the high-resolution dictionary to recover the corresponding\nhigh-resolution image patch. In this work, we study the effect of the sparse\nrecovery algorithm that we use on the quality of the reconstructed images. We\noffer empirical experiments to search for the best sparse recovery algorithm\nthat can be used for this purpose.\n", "date": "2023-08-04", "categories": []}
{"id": "2308.02794", "title": "Unfolding Once is Enough: A Deployment-Friendly Transformer Unit for\n  Super-Resolution", "abstract": "  Recent years have witnessed a few attempts of vision transformers for single\nimage super-resolution (SISR). Since the high resolution of intermediate\nfeatures in SISR models increases memory and computational requirements,\nefficient SISR transformers are more favored. Based on some popular transformer\nbackbone, many methods have explored reasonable schemes to reduce the\ncomputational complexity of the self-attention module while achieving\nimpressive performance. However, these methods only focus on the performance on\nthe training platform (e.g., Pytorch/Tensorflow) without further optimization\nfor the deployment platform (e.g., TensorRT). Therefore, they inevitably\ncontain some redundant operators, posing challenges for subsequent deployment\nin real-world applications. In this paper, we propose a deployment-friendly\ntransformer unit, namely UFONE (i.e., UnFolding ONce is Enough), to alleviate\nthese problems. In each UFONE, we introduce an Inner-patch Transformer Layer\n(ITL) to efficiently reconstruct the local structural information from patches\nand a Spatial-Aware Layer (SAL) to exploit the long-range dependencies between\npatches. Based on UFONE, we propose a Deployment-friendly Inner-patch\nTransformer Network (DITN) for the SISR task, which can achieve favorable\nperformance with low latency and memory usage on both training and deployment\nplatforms. Furthermore, to further boost the deployment efficiency of the\nproposed DITN on TensorRT, we also provide an efficient substitution for layer\nnormalization and propose a fusion optimization strategy for specific\noperators. Extensive experiments show that our models can achieve competitive\nresults in terms of qualitative and quantitative performance with high\ndeployment efficiency. Code is available at\n\\url{https://github.com/yongliuy/DITN}.\n", "date": "2023-08-05", "categories": []}
{"id": "2308.03018", "title": "Recurrent Spike-based Image Restoration under General Illumination", "abstract": "  Spike camera is a new type of bio-inspired vision sensor that records light\nintensity in the form of a spike array with high temporal resolution (20,000\nHz). This new paradigm of vision sensor offers significant advantages for many\nvision tasks such as high speed image reconstruction. However, existing\nspike-based approaches typically assume that the scenes are with sufficient\nlight intensity, which is usually unavailable in many real-world scenarios such\nas rainy days or dusk scenes. To unlock more spike-based application scenarios,\nwe propose a Recurrent Spike-based Image Restoration (RSIR) network, which is\nthe first work towards restoring clear images from spike arrays under general\nillumination. Specifically, to accurately describe the noise distribution under\ndifferent illuminations, we build a physical-based spike noise model according\nto the sampling process of the spike camera. Based on the noise model, we\ndesign our RSIR network which consists of an adaptive spike transformation\nmodule, a recurrent temporal feature fusion module, and a frequency-based spike\ndenoising module. Our RSIR can process the spike array in a recursive manner to\nensure that the spike temporal information is well utilized. In the training\nprocess, we generate the simulated spike data based on our noise model to train\nour network. Extensive experiments on real-world datasets with different\nilluminations demonstrate the effectiveness of the proposed network. The code\nand dataset are released at https://github.com/BIT-Vision/RSIR.\n", "date": "2023-08-06", "categories": []}
{"id": "2308.03608", "title": "Recurrent Self-Supervised Video Denoising with Denser Receptive Field", "abstract": "  Self-supervised video denoising has seen decent progress through the use of\nblind spot networks. However, under their blind spot constraints, previous\nself-supervised video denoising methods suffer from significant information\nloss and texture destruction in either the whole reference frame or neighbor\nframes, due to their inadequate consideration of the receptive field. Moreover,\nthe limited number of available neighbor frames in previous methods leads to\nthe discarding of distant temporal information. Nonetheless, simply adopting\nexisting recurrent frameworks does not work, since they easily break the\nconstraints on the receptive field imposed by self-supervision. In this paper,\nwe propose RDRF for self-supervised video denoising, which not only fully\nexploits both the reference and neighbor frames with a denser receptive field,\nbut also better leverages the temporal information from both local and distant\nneighbor features. First, towards a comprehensive utilization of information\nfrom both reference and neighbor frames, RDRF realizes a denser receptive field\nby taking more neighbor pixels along the spatial and temporal dimensions.\nSecond, it features a self-supervised recurrent video denoising framework,\nwhich concurrently integrates distant and near-neighbor temporal features. This\nenables long-term bidirectional information aggregation, while mitigating error\naccumulation in the plain recurrent framework. Our method exhibits superior\nperformance on both synthetic and real video denoising datasets. Codes will be\navailable at https://github.com/Wang-XIaoDingdd/RDRF.\n", "date": "2023-08-07", "categories": []}
{"id": "2308.01379", "title": "Computational Long Exposure Mobile Photography", "abstract": "  Long exposure photography produces stunning imagery, representing moving\nelements in a scene with motion-blur. It is generally employed in two\nmodalities, producing either a foreground or a background blur effect.\nForeground blur images are traditionally captured on a tripod-mounted camera\nand portray blurred moving foreground elements, such as silky water or light\ntrails, over a perfectly sharp background landscape. Background blur images,\nalso called panning photography, are captured while the camera is tracking a\nmoving subject, to produce an image of a sharp subject over a background\nblurred by relative motion. Both techniques are notoriously challenging and\nrequire additional equipment and advanced skills. In this paper, we describe a\ncomputational burst photography system that operates in a hand-held smartphone\ncamera app, and achieves these effects fully automatically, at the tap of the\nshutter button. Our approach first detects and segments the salient subject. We\ntrack the scene motion over multiple frames and align the images in order to\npreserve desired sharpness and to produce aesthetically pleasing motion\nstreaks. We capture an under-exposed burst and select the subset of input\nframes that will produce blur trails of controlled length, regardless of scene\nor camera motion velocity. We predict inter-frame motion and synthesize\nmotion-blur to fill the temporal gaps between the input frames. Finally, we\ncomposite the blurred image with the sharp regular exposure to protect the\nsharpness of faces or areas of the scene that are barely moving, and produce a\nfinal high resolution and high dynamic range (HDR) photograph. Our system\ndemocratizes a capability previously reserved to professionals, and makes this\ncreative style accessible to most casual photographers.\n  More information and supplementary material can be found on our project\nwebpage: https://motion-mode.github.io/\n", "date": "2023-08-02", "categories": []}
{"id": "2308.01541", "title": "DMDC: Dynamic-mask-based dual camera design for snapshot Hyperspectral\n  Imaging", "abstract": "  Deep learning methods are developing rapidly in coded aperture snapshot\nspectral imaging (CASSI). The number of parameters and FLOPs of existing\nstate-of-the-art methods (SOTA) continues to increase, but the reconstruction\naccuracy improves slowly. Current methods still face two problems: 1) The\nperformance of the spatial light modulator (SLM) is not fully developed due to\nthe limitation of fixed Mask coding. 2) The single input limits the network\nperformance. In this paper we present a dynamic-mask-based dual camera system,\nwhich consists of an RGB camera and a CASSI system running in parallel. First,\nthe system learns the spatial feature distribution of the scene based on the\nRGB images, then instructs the SLM to encode each scene, and finally sends both\nRGB and CASSI images to the network for reconstruction. We further designed the\nDMDC-net, which consists of two separate networks, a small-scale CNN-based\ndynamic mask network for dynamic adjustment of the mask and a multimodal\nreconstruction network for reconstruction using RGB and CASSI measurements.\nExtensive experiments on multiple datasets show that our method achieves more\nthan 9 dB improvement in PSNR over the SOTA.\n(https://github.com/caizeyu1992/DMDC)\n", "date": "2023-08-03", "categories": []}
{"id": "2308.01594", "title": "Reference-Free Isotropic 3D EM Reconstruction using Diffusion Models", "abstract": "  Electron microscopy (EM) images exhibit anisotropic axial resolution due to\nthe characteristics inherent to the imaging modality, presenting challenges in\nanalysis and downstream tasks.In this paper, we propose a diffusion-model-based\nframework that overcomes the limitations of requiring reference data or prior\nknowledge about the degradation process. Our approach utilizes 2D diffusion\nmodels to consistently reconstruct 3D volumes and is well-suited for highly\ndownsampled data. Extensive experiments conducted on two public datasets\ndemonstrate the robustness and superiority of leveraging the generative prior\ncompared to supervised learning methods. Additionally, we demonstrate our\nmethod's feasibility for self-supervised reconstruction, which can restore a\nsingle anisotropic volume without any training data.\n", "date": "2023-08-03", "categories": []}
{"id": "2303.12280", "title": "NLOS-NeuS: Non-line-of-sight Neural Implicit Surface", "abstract": "  Non-line-of-sight (NLOS) imaging is conducted to infer invisible scenes from\nindirect light on visible objects. The neural transient field (NeTF) was\nproposed for representing scenes as neural radiance fields in NLOS scenes. We\npropose NLOS neural implicit surface (NLOS-NeuS), which extends the NeTF to\nneural implicit surfaces with a signed distance function (SDF) for\nreconstructing three-dimensional surfaces in NLOS scenes. We introduce two\nconstraints as loss functions for correctly learning an SDF to avoid non-zero\nlevel-set surfaces. We also introduce a lower bound constraint of an SDF based\non the geometry of the first-returning photons. The experimental results\nindicate that these constraints are essential for learning a correct SDF in\nNLOS scenes. Compared with previous methods with discretized representation,\nNLOS-NeuS with the neural continuous representation enables us to reconstruct\nsmooth surfaces while preserving fine details in NLOS scenes. To the best of\nour knowledge, this is the first study on neural implicit surfaces with volume\nrendering in NLOS scenes.\n", "date": "2023-03-22", "categories": []}
{"id": "2306.13909", "title": "Deep learning-based deconvolution for interferometric radio transient\n  reconstruction", "abstract": "  Radio astronomy is currently thriving with new large ground-based radio\ntelescopes coming online in preparation for the upcoming Square Kilometre Array\n(SKA). Facilities like LOFAR, MeerKAT/SKA, ASKAP/SKA, and the future SKA-LOW\nbring tremendous sensitivity in time and frequency, improved angular\nresolution, and also high-rate data streams that need to be processed. They\nenable advanced studies of radio transients, volatile by nature, that can be\ndetected or missed in the data. These transients are markers of high-energy\naccelerations of electrons and manifest in a wide range of temporal scales.\nUsually studied with dynamic spectroscopy of time series analysis, there is a\nmotivation to search for such sources in large interferometric datasets. This\nrequires efficient and robust signal reconstruction algorithms. To correctly\naccount for the temporal dependency of the data, we improve the classical image\ndeconvolution inverse problem by adding the temporal dependency in the\nreconstruction problem. Then, we introduce two novel neural network\narchitectures that can do both spatial and temporal modeling of the data and\nthe instrumental response. Then, we simulate representative time-dependent\nimage cubes of point source distributions and realistic telescope pointings of\nMeerKAT to generate toy models to build the training, validation, and test\ndatasets. Finally, based on the test data, we evaluate the source profile\nreconstruction performance of the proposed methods and classical image\ndeconvolution algorithm CLEAN applied frame-by-frame. In the presence of\nincreasing noise level in data frame, the proposed methods display a high level\nof robustness compared to frame-by-frame imaging with CLEAN. The deconvolved\nimage cubes bring a factor of 3 improvement in fidelity of the recovered\ntemporal profiles and a factor of 2 improvement in background denoising.\n", "date": "2023-06-24", "categories": []}
{"id": "2307.06607", "title": "Image Denoising and the Generative Accumulation of Photons", "abstract": "  We present a fresh perspective on shot noise corrupted images and noise\nremoval. By viewing image formation as the sequential accumulation of photons\non a detector grid, we show that a network trained to predict where the next\nphoton could arrive is in fact solving the minimum mean square error (MMSE)\ndenoising task. This new perspective allows us to make three contributions: We\npresent a new strategy for self-supervised denoising, We present a new method\nfor sampling from the posterior of possible solutions by iteratively sampling\nand adding small numbers of photons to the image. We derive a full generative\nmodel by starting this process from an empty canvas. We call this approach\ngenerative accumulation of photons (GAP). We evaluate our method quantitatively\nand qualitatively on 4 new fluorescence microscopy datasets, which will be made\navailable to the community. We find that it outperforms supervised,\nself-supervised and unsupervised baselines or performs on-par.\n", "date": "2023-07-13", "categories": []}
{"id": "2308.00591", "title": "Visibility Enhancement for Low-light Hazy Scenarios", "abstract": "  Low-light hazy scenes commonly appear at dusk and early morning. The visual\nenhancement for low-light hazy images is an ill-posed problem. Even though\nnumerous methods have been proposed for image dehazing and low-light\nenhancement respectively, simply integrating them cannot deliver pleasing\nresults for this particular task. In this paper, we present a novel method to\nenhance visibility for low-light hazy scenarios. To handle this challenging\ntask, we propose two key techniques, namely cross-consistency\ndehazing-enhancement framework and physically based simulation for low-light\nhazy dataset. Specifically, the framework is designed for enhancing visibility\nof the input image via fully utilizing the clues from different sub-tasks. The\nsimulation is designed for generating the dataset with ground-truths by the\nproposed low-light hazy imaging model. The extensive experimental results show\nthat the proposed method outperforms the SOTA solutions on different metrics\nincluding SSIM (9.19%) and PSNR(5.03%). In addition, we conduct a user study on\nreal images to demonstrate the effectiveness and necessity of the proposed\nmethod by human visual perception.\n", "date": "2023-08-01", "categories": []}
{"id": "2307.16242", "title": "SR-R$^2$KAC: Improving Single Image Defocus Deblurring", "abstract": "  We propose an efficient deep learning method for single image defocus\ndeblurring (SIDD) by further exploring inverse kernel properties. Although the\ncurrent inverse kernel method, i.e., kernel-sharing parallel atrous convolution\n(KPAC), can address spatially varying defocus blurs, it has difficulty in\nhandling large blurs of this kind. To tackle this issue, we propose a Residual\nand Recursive Kernel-sharing Atrous Convolution (R$^2$KAC). R$^2$KAC builds on\na significant observation of inverse kernels, that is, successive use of\ninverse-kernel-based deconvolutions with fixed size helps remove unexpected\nlarge blurs but produces ringing artifacts. Specifically, on top of\nkernel-sharing atrous convolutions used to simulate multi-scale inverse\nkernels, R$^2$KAC applies atrous convolutions recursively to simulate a large\ninverse kernel. Specifically, on top of kernel-sharing atrous convolutions,\nR$^2$KAC stacks atrous convolutions recursively to simulate a large inverse\nkernel. To further alleviate the contingent effect of recursive stacking, i.e.,\nringing artifacts, we add identity shortcuts between atrous convolutions to\nsimulate residual deconvolutions. Lastly, a scale recurrent module is embedded\nin the R$^2$KAC network, leading to SR-R$^2$KAC, so that multi-scale\ninformation from coarse to fine is exploited to progressively remove the\nspatially varying defocus blurs. Extensive experimental results show that our\nmethod achieves the state-of-the-art performance.\n", "date": "2023-07-30", "categories": []}
{"id": "2307.16783", "title": "From Generation to Suppression: Towards Effective Irregular Glow Removal\n  for Nighttime Visibility Enhancement", "abstract": "  Most existing Low-Light Image Enhancement (LLIE) methods are primarily\ndesigned to improve brightness in dark regions, which suffer from severe\ndegradation in nighttime images. However, these methods have limited\nexploration in another major visibility damage, the glow effects in real night\nscenes. Glow effects are inevitable in the presence of artificial light sources\nand cause further diffused blurring when directly enhanced. To settle this\nissue, we innovatively consider the glow suppression task as learning physical\nglow generation via multiple scattering estimation according to the Atmospheric\nPoint Spread Function (APSF). In response to the challenges posed by uneven\nglow intensity and varying source shapes, an APSF-based Nighttime Imaging Model\nwith Near-field Light Sources (NIM-NLS) is specifically derived to design a\nscalable Light-aware Blind Deconvolution Network (LBDN). The glow-suppressed\nresult is then brightened via a Retinex-based Enhancement Module (REM).\nRemarkably, the proposed glow suppression method is based on zero-shot learning\nand does not rely on any paired or unpaired training data. Empirical\nevaluations demonstrate the effectiveness of the proposed method in both glow\nsuppression and low-light enhancement tasks.\n", "date": "2023-07-31", "categories": []}
{"id": "2212.03235", "title": "Complex-valued Retrievals From Noisy Images Using Diffusion Models", "abstract": "  In diverse microscopy modalities, sensors measure only real-valued\nintensities. Additionally, the sensor readouts are affected by\nPoissonian-distributed photon noise. Traditional restoration algorithms\ntypically aim to minimize the mean squared error (MSE) between the original and\nrecovered images. This often leads to blurry outcomes with poor perceptual\nquality. Recently, deep diffusion models (DDMs) have proven to be highly\ncapable of sampling images from the a-posteriori probability of the sought\nvariables, resulting in visually pleasing high-quality images. These models\nhave mostly been suggested for real-valued images suffering from Gaussian\nnoise. In this study, we generalize annealed Langevin Dynamics, a type of DDM,\nto tackle the fundamental challenges in optical imaging of complex-valued\nobjects (and real images) affected by Poisson noise. We apply our algorithm to\nvarious optical scenarios, such as Fourier Ptychography, Phase Retrieval, and\nPoisson denoising. Our algorithm is evaluated on simulations and biological\nempirical data.\n", "date": "2022-12-06", "categories": []}
{"id": "2305.05077", "title": "Atmospheric Turbulence Correction via Variational Deep Diffusion", "abstract": "  Atmospheric Turbulence (AT) correction is a challenging restoration task as\nit consists of two distortions: geometric distortion and spatially variant\nblur. Diffusion models have shown impressive accomplishments in photo-realistic\nimage synthesis and beyond. In this paper, we propose a novel deep conditional\ndiffusion model under a variational inference framework to solve the AT\ncorrection problem. We use this framework to improve performance by learning\nlatent prior information from the input and degradation processes. We use the\nlearned information to further condition the diffusion model. Experiments are\nconducted in a comprehensive synthetic AT dataset. We show that the proposed\nframework achieves good quantitative and qualitative results.\n", "date": "2023-05-08", "categories": []}
{"id": "2307.14363", "title": "Unsupervised reconstruction of accelerated cardiac cine MRI using Neural\n  Fields", "abstract": "  Cardiac cine MRI is the gold standard for cardiac functional assessment, but\nthe inherently slow acquisition process creates the necessity of reconstruction\napproaches for accelerated undersampled acquisitions. Several regularization\napproaches that exploit spatial-temporal redundancy have been proposed to\nreconstruct undersampled cardiac cine MRI. More recently, methods based on\nsupervised deep learning have been also proposed to further accelerate\nacquisition and reconstruction. However, these techniques rely on usually large\ndataset for training, which are not always available. In this work, we propose\nan unsupervised approach based on implicit neural field representations for\ncardiac cine MRI (so called NF-cMRI). The proposed method was evaluated in\nin-vivo undersampled golden-angle radial multi-coil acquisitions for\nundersampling factors of 26x and 52x, achieving good image quality, and\ncomparable spatial and improved temporal depiction than a state-of-the-art\nreconstruction technique.\n", "date": "2023-07-24", "categories": []}
{"id": "2307.14579", "title": "Neural Representation-Based Method for Metal-induced Artifact Reduction\n  in Dental CBCT Imaging", "abstract": "  This study introduces a novel reconstruction method for dental cone-beam\ncomputed tomography (CBCT), focusing on effectively reducing metal-induced\nartifacts commonly encountered in the presence of prevalent metallic implants.\nDespite significant progress in metal artifact reduction techniques, challenges\npersist owing to the intricate physical interactions between polychromatic\nX-ray beams and metal objects, which are further compounded by the additional\neffects associated with metal-tooth interactions and factors specific to the\ndental CBCT data environment. To overcome these limitations, we propose an\nimplicit neural network that generates two distinct and informative tomographic\nimages. One image represents the monochromatic attenuation distribution at a\nspecific energy level, whereas the other captures the nonlinear beam-hardening\nfactor resulting from the polychromatic nature of X-ray beams. In contrast to\nexisting CT reconstruction techniques, the proposed method relies exclusively\non the Beer--Lambert law, effectively preventing the generation of\nmetal-induced artifacts during the backprojection process commonly implemented\nin conventional methods. Extensive experimental evaluations demonstrate that\nthe proposed method effectively reduces metal artifacts while providing\nhigh-quality image reconstructions, thus emphasizing the significance of the\nsecond image in capturing the nonlinear beam-hardening factor.\n", "date": "2023-07-27", "categories": []}
{"id": "2206.02070", "title": "Priors in Deep Image Restoration and Enhancement: A Survey", "abstract": "  Image restoration and enhancement is a process of improving the image quality\nby removing degradations, such as noise, blur, and resolution degradation. Deep\nlearning (DL) has recently been applied to image restoration and enhancement.\nDue to its ill-posed property, plenty of works have been explored priors to\nfacilitate training deep neural networks (DNNs). However, the importance of\npriors has not been systematically studied and analyzed by far in the research\ncommunity. Therefore, this paper serves as the first study that provides a\ncomprehensive overview of recent advancements in priors for deep image\nrestoration and enhancement. Our work covers five primary contents: (1) A\ntheoretical analysis of priors for deep image restoration and enhancement; (2)\nA hierarchical and structural taxonomy of priors commonly used in the DL-based\nmethods; (3) An insightful discussion on each prior regarding its principle,\npotential, and applications; (4) A summary of crucial problems by highlighting\nthe potential future directions, especially adopting the large-scale foundation\nmodels as prior, to spark more research in the community; (5) An open-source\nrepository that provides a taxonomy of all mentioned works and code links.\n", "date": "2022-06-04", "categories": []}
{"id": "2307.14341", "title": "Virtual Mirrors: Non-Line-of-Sight Imaging Beyond the Third Bounce", "abstract": "  Non-line-of-sight (NLOS) imaging methods are capable of reconstructing\ncomplex scenes that are not visible to an observer using indirect illumination.\nHowever, they assume only third-bounce illumination, so they are currently\nlimited to single-corner configurations, and present limited visibility when\nimaging surfaces at certain orientations. To reason about and tackle these\nlimitations, we make the key observation that planar diffuse surfaces behave\nspecularly at wavelengths used in the computational wave-based NLOS imaging\ndomain. We call such surfaces virtual mirrors. We leverage this observation to\nexpand the capabilities of NLOS imaging using illumination beyond the third\nbounce, addressing two problems: imaging single-corner objects at limited\nvisibility angles, and imaging objects hidden behind two corners. To image\nobjects at limited visibility angles, we first analyze the reflections of the\nknown illuminated point on surfaces of the scene as an estimator of the\nposition and orientation of objects with limited visibility. We then image\nthose limited visibility objects by computationally building secondary\napertures at other surfaces that observe the target object from a direct\nvisibility perspective. Beyond single-corner NLOS imaging, we exploit the\nspecular behavior of virtual mirrors to image objects hidden behind a second\ncorner by imaging the space behind such virtual mirrors, where the mirror image\nof objects hidden around two corners is formed. No specular surfaces were\ninvolved in the making of this paper.\n", "date": "2023-07-26", "categories": []}
{"id": "2306.12078", "title": "Accelerating Multiframe Blind Deconvolution via Deep Learning", "abstract": "  Ground-based solar image restoration is a computationally expensive procedure\nthat involves nonlinear optimization techniques. The presence of atmospheric\nturbulence produces perturbations in individual images that make it necessary\nto apply blind deconvolution techniques. These techniques rely on the\nobservation of many short exposure frames that are used to simultaneously infer\nthe instantaneous state of the atmosphere and the unperturbed object. We have\nrecently explored the use of machine learning to accelerate this process, with\npromising results. We build upon this previous work to propose several\ninteresting improvements that lead to better models. As well, we propose a new\nmethod to accelerate the restoration based on algorithm unrolling. In this\nmethod, the image restoration problem is solved with a gradient descent method\nthat is unrolled and accelerated aided by a few small neural networks. The role\nof the neural networks is to correct the estimation of the solution at each\niterative step. The model is trained to perform the optimization in a small\nfixed number of steps with a curated dataset. Our findings demonstrate that\nboth methods significantly reduce the restoration time compared to the standard\noptimization procedure. Furthermore, we showcase that these models can be\ntrained in an unsupervised manner using observed images from three different\ninstruments. Remarkably, they also exhibit robust generalization capabilities\nwhen applied to new datasets. To foster further research and collaboration, we\nopenly provide the trained models, along with the corresponding training and\nevaluation code, as well as the training dataset, to the scientific community.\n", "date": "2023-06-21", "categories": []}
{"id": "2307.02625", "title": "Retinex-based Image Denoising / Contrast Enhancement using Gradient\n  Graph Laplacian Regularizer", "abstract": "  Images captured in poorly lit conditions are often corrupted by acquisition\nnoise. Leveraging recent advances in graph-based regularization, we propose a\nfast Retinex-based restoration scheme that denoises and contrast-enhances an\nimage. Specifically, by Retinex theory we first assume that each image pixel is\na multiplication of its reflectance and illumination components. We next assume\nthat the reflectance and illumination components are piecewise constant (PWC)\nand continuous piecewise planar (PWP) signals, which can be recovered via graph\nLaplacian regularizer (GLR) and gradient graph Laplacian regularizer (GGLR)\nrespectively. We formulate quadratic objectives regularized by GLR and GGLR,\nwhich are minimized alternately until convergence by solving linear systems --\nwith improved condition numbers via proposed preconditioners -- via conjugate\ngradient (CG) efficiently. Experimental results show that our algorithm\nachieves competitive visual image quality while reducing computation complexity\nnoticeably.\n", "date": "2023-07-05", "categories": []}
{"id": "2307.06207", "title": "Local Conditional Neural Fields for Versatile and Generalizable\n  Large-Scale Reconstructions in Computational Imaging", "abstract": "  Deep learning has transformed computational imaging, but traditional\npixel-based representations limit their ability to capture continuous,\nmultiscale details of objects. Here we introduce a novel Local Conditional\nNeural Fields (LCNF) framework, leveraging a continuous implicit neural\nrepresentation to address this limitation. LCNF enables flexible object\nrepresentation and facilitates the reconstruction of multiscale information. We\ndemonstrate the capabilities of LCNF in solving the highly ill-posed inverse\nproblem in Fourier ptychographic microscopy (FPM) with multiplexed\nmeasurements, achieving robust, scalable, and generalizable large-scale phase\nretrieval. Unlike traditional neural fields frameworks, LCNF incorporates a\nlocal conditional representation that promotes model generalization, learning\nmultiscale information, and efficient processing of large-scale imaging data.\nBy combining an encoder and a decoder conditioned on a learned latent vector,\nLCNF achieves versatile continuous-domain super-resolution image\nreconstruction. We demonstrate accurate reconstruction of wide field-of-view,\nhigh-resolution phase images using only a few multiplexed measurements. LCNF\nrobustly captures the continuous object priors and eliminates various phase\nartifacts, even when it is trained on imperfect datasets. The framework\nexhibits strong generalization, reconstructing diverse objects even with\nlimited training data. Furthermore, LCNF can be trained on a physics simulator\nusing natural images and successfully applied to experimental measurements on\nbiological samples. Our results highlight the potential of LCNF for solving\nlarge-scale inverse problems in computational imaging, with broad applicability\nin various deep-learning-based techniques.\n", "date": "2023-07-12", "categories": []}
{"id": "2307.12145", "title": "A Vision for Cleaner Rivers: Harnessing Snapshot Hyperspectral Imaging\n  to Detect Macro-Plastic Litter", "abstract": "  Plastic waste entering the riverine harms local ecosystems leading to\nnegative ecological and economic impacts. Large parcels of plastic waste are\ntransported from inland to oceans leading to a global scale problem of floating\ndebris fields. In this context, efficient and automatized monitoring of\nmismanaged plastic waste is paramount. To address this problem, we analyze the\nfeasibility of macro-plastic litter detection using computational imaging\napproaches in river-like scenarios. We enable near-real-time tracking of\npartially submerged plastics by using snapshot Visible-Shortwave Infrared\nhyperspectral imaging. Our experiments indicate that imaging strategies\nassociated with machine learning classification approaches can lead to high\ndetection accuracy even in challenging scenarios, especially when leveraging\nhyperspectral data and nonlinear classifiers. All code, data, and models are\navailable online:\nhttps://github.com/RIVeR-Lab/hyperspectral_macro_plastic_detection.\n", "date": "2023-07-22", "categories": []}
{"id": "2307.12656", "title": "A Theoretically Guaranteed Quaternion Weighted Schatten p-norm\n  Minimization Method for Color Image Restoration", "abstract": "  Inspired by the fact that the matrix formulated by nonlocal similar patches\nin a natural image is of low rank, the rank approximation issue have been\nextensively investigated over the past decades, among which weighted nuclear\nnorm minimization (WNNM) and weighted Schatten $p$-norm minimization (WSNM) are\ntwo prevailing methods have shown great superiority in various image\nrestoration (IR) problems. Due to the physical characteristic of color images,\ncolor image restoration (CIR) is often a much more difficult task than its\ngrayscale image counterpart. However, when applied to CIR, the traditional\nWNNM/WSNM method only processes three color channels individually and fails to\nconsider their cross-channel correlations. Very recently, a quaternion-based\nWNNM approach (QWNNM) has been developed to mitigate this issue, which is\ncapable of representing the color image as a whole in the quaternion domain and\npreserving the inherent correlation among the three color channels. Despite its\nempirical success, unfortunately, the convergence behavior of QWNNM has not\nbeen strictly studied yet. In this paper, on the one side, we extend the WSNM\ninto quaternion domain and correspondingly propose a novel quaternion-based\nWSNM model (QWSNM) for tackling the CIR problems. Extensive experiments on two\nrepresentative CIR tasks, including color image denoising and deblurring,\ndemonstrate that the proposed QWSNM method performs favorably against many\nstate-of-the-art alternatives, in both quantitative and qualitative\nevaluations. On the other side, more importantly, we preliminarily provide a\ntheoretical convergence analysis, that is, by modifying the quaternion\nalternating direction method of multipliers (QADMM) through a simple\ncontinuation strategy, we theoretically prove that both the solution sequences\ngenerated by the QWNNM and QWSNM have fixed-point convergence guarantees.\n", "date": "2023-07-24", "categories": []}
{"id": "2306.07308", "title": "Self-Supervised Hyperspectral Inpainting with the Optimisation inspired\n  Deep Neural Network Prior", "abstract": "  Hyperspectral Image (HSI)s cover hundreds or thousands of narrow spectral\nbands, conveying a wealth of spatial and spectral information. However, due to\nthe instrumental errors and the atmospheric changes, the HSI obtained in\npractice are often contaminated by noise and dead pixels(lines), resulting in\nmissing information that may severely compromise the subsequent applications.\nWe introduce here a novel HSI missing pixel prediction algorithm, called Low\nRank and Sparsity Constraint Plug-and-Play (LRS-PnP). It is shown that LRS-PnP\nis able to predict missing pixels and bands even when all spectral bands of the\nimage are missing. The proposed LRS-PnP algorithm is further extended to a\nself-supervised model by combining the LRS-PnP with the Deep Image Prior (DIP),\ncalled LRS-PnP-DIP. In a series of experiments with real data, It is shown that\nthe LRS-PnP-DIP either achieves state-of-the-art inpainting performance\ncompared to other learning-based methods, or outperforms them.\n", "date": "2023-06-12", "categories": []}
{"id": "2307.11122", "title": "Diffusion Models for Probabilistic Deconvolution of Galaxy Images", "abstract": "  Telescopes capture images with a particular point spread function (PSF).\nInferring what an image would have looked like with a much sharper PSF, a\nproblem known as PSF deconvolution, is ill-posed because PSF convolution is not\nan invertible transformation. Deep generative models are appealing for PSF\ndeconvolution because they can infer a posterior distribution over candidate\nimages that, if convolved with the PSF, could have generated the observation.\nHowever, classical deep generative models such as VAEs and GANs often provide\ninadequate sample diversity. As an alternative, we propose a classifier-free\nconditional diffusion model for PSF deconvolution of galaxy images. We\ndemonstrate that this diffusion model captures a greater diversity of possible\ndeconvolutions compared to a conditional VAE.\n", "date": "2023-07-20", "categories": []}
{"id": "2307.10667", "title": "Efficient Unified Demosaicing for Bayer and Non-Bayer Patterned Image\n  Sensors", "abstract": "  As the physical size of recent CMOS image sensors (CIS) gets smaller, the\nlatest mobile cameras are adopting unique non-Bayer color filter array (CFA)\npatterns (e.g., Quad, Nona, QxQ), which consist of homogeneous color units with\nadjacent pixels. These non-Bayer sensors are superior to conventional Bayer CFA\nthanks to their changeable pixel-bin sizes for different light conditions but\nmay introduce visual artifacts during demosaicing due to their inherent pixel\npattern structures and sensor hardware characteristics. Previous demosaicing\nmethods have primarily focused on Bayer CFA, necessitating distinct\nreconstruction methods for non-Bayer patterned CIS with various CFA modes under\ndifferent lighting conditions. In this work, we propose an efficient unified\ndemosaicing method that can be applied to both conventional Bayer RAW and\nvarious non-Bayer CFAs' RAW data in different operation modes. Our Knowledge\nLearning-based demosaicing model for Adaptive Patterns, namely KLAP, utilizes\nCFA-adaptive filters for only 1% key filters in the network for each CFA, but\nstill manages to effectively demosaic all the CFAs, yielding comparable\nperformance to the large-scale models. Furthermore, by employing meta-learning\nduring inference (KLAP-M), our model is able to eliminate unknown\nsensor-generic artifacts in real RAW data, effectively bridging the gap between\nsynthetic images and real sensor RAW. Our KLAP and KLAP-M methods achieved\nstate-of-the-art demosaicing performance in both synthetic and real RAW data of\nBayer and non-Bayer CFAs.\n", "date": "2023-07-20", "categories": []}
{"id": "2306.16197", "title": "Multi-IMU with Online Self-Consistency for Freehand 3D Ultrasound\n  Reconstruction", "abstract": "  Ultrasound (US) imaging is a popular tool in clinical diagnosis, offering\nsafety, repeatability, and real-time capabilities. Freehand 3D US is a\ntechnique that provides a deeper understanding of scanned regions without\nincreasing complexity. However, estimating elevation displacement and\naccumulation error remains challenging, making it difficult to infer the\nrelative position using images alone. The addition of external lightweight\nsensors has been proposed to enhance reconstruction performance without adding\ncomplexity, which has been shown to be beneficial. We propose a novel online\nself-consistency network (OSCNet) using multiple inertial measurement units\n(IMUs) to improve reconstruction performance. OSCNet utilizes a modal-level\nself-supervised strategy to fuse multiple IMU information and reduce\ndifferences between reconstruction results obtained from each IMU data.\nAdditionally, a sequence-level self-consistency strategy is proposed to improve\nthe hierarchical consistency of prediction results among the scanning sequence\nand its sub-sequences. Experiments on large-scale arm and carotid datasets with\nmultiple scanning tactics demonstrate that our OSCNet outperforms previous\nmethods, achieving state-of-the-art reconstruction performance.\n", "date": "2023-06-28", "categories": []}
{"id": "2307.09818", "title": "Deep unrolling Shrinkage Network for Dynamic MR imaging", "abstract": "  Deep unrolling networks that utilize sparsity priors have achieved great\nsuccess in dynamic magnetic resonance (MR) imaging. The convolutional neural\nnetwork (CNN) is usually utilized to extract the transformed domain, and then\nthe soft thresholding (ST) operator is applied to the CNN-transformed data to\nenforce the sparsity priors. However, the ST operator is usually constrained to\nbe the same across all channels of the CNN-transformed data. In this paper, we\npropose a novel operator, called soft thresholding with channel attention\n(AST), that learns the threshold for each channel. In particular, we put\nforward a novel deep unrolling shrinkage network (DUS-Net) by unrolling the\nalternating direction method of multipliers (ADMM) for optimizing the\ntransformed $l_1$ norm dynamic MR reconstruction model. Experimental results on\nan open-access dynamic cine MR dataset demonstrate that the proposed DUS-Net\noutperforms the state-of-the-art methods. The source code is available at\n\\url{https://github.com/yhao-z/DUS-Net}.\n", "date": "2023-07-19", "categories": []}
{"id": "2307.09841", "title": "Compressive Image Scanning Microscope", "abstract": "  We present a novel approach to implement compressive sensing in laser\nscanning microscopes (LSM), specifically in image scanning microscopy (ISM),\nusing a single-photon avalanche diode (SPAD) array detector. Our method\naddresses two significant limitations in applying compressive sensing to LSM:\nthe time to compute the sampling matrix and the quality of reconstructed\nimages. We employ a fixed sampling strategy, skipping alternate rows and\ncolumns during data acquisition, which reduces the number of points scanned by\na factor of four and eliminates the need to compute different sampling\nmatrices. By exploiting the parallel images generated by the SPAD array, we\nimprove the quality of the reconstructed compressive-ISM images compared to\nstandard compressive confocal LSM images. Our results demonstrate the\neffectiveness of our approach in producing higher-quality images with reduced\ndata acquisition time and potential benefits in reducing photobleaching.\n", "date": "2023-07-19", "categories": []}
{"id": "2211.16855", "title": "ATASI-Net: An Efficient Sparse Reconstruction Network for Tomographic\n  SAR Imaging with Adaptive Threshold", "abstract": "  Tomographic SAR technique has attracted remarkable interest for its ability\nof three-dimensional resolving along the elevation direction via a stack of SAR\nimages collected from different cross-track angles. The emerged compressed\nsensing (CS)-based algorithms have been introduced into TomoSAR considering its\nsuper-resolution ability with limited samples. However, the conventional\nCS-based methods suffer from several drawbacks, including weak noise\nresistance, high computational complexity, and complex parameter fine-tuning.\nAiming at efficient TomoSAR imaging, this paper proposes a novel efficient\nsparse unfolding network based on the analytic learned iterative shrinkage\nthresholding algorithm (ALISTA) architecture with adaptive threshold, named\nAdaptive Threshold ALISTA-based Sparse Imaging Network (ATASI-Net). The weight\nmatrix in each layer of ATASI-Net is pre-computed as the solution of an\noff-line optimization problem, leaving only two scalar parameters to be learned\nfrom data, which significantly simplifies the training stage. In addition,\nadaptive threshold is introduced for each azimuth-range pixel, enabling the\nthreshold shrinkage to be not only layer-varied but also element-wise.\nMoreover, the final learned thresholds can be visualized and combined with the\nSAR image semantics for mutual feedback. Finally, extensive experiments on\nsimulated and real data are carried out to demonstrate the effectiveness and\nefficiency of the proposed method.\n", "date": "2022-11-30", "categories": []}
{"id": "2306.15548", "title": "Geometric Ultrasound Localization Microscopy", "abstract": "  Contrast-Enhanced Ultra-Sound (CEUS) has become a viable method for\nnon-invasive, dynamic visualization in medical diagnostics, yet Ultrasound\nLocalization Microscopy (ULM) has enabled a revolutionary breakthrough by\noffering ten times higher resolution. To date, Delay-And-Sum (DAS) beamformers\nare used to render ULM frames, ultimately determining the image resolution\ncapability. To take full advantage of ULM, this study questions whether\nbeamforming is the most effective processing step for ULM, suggesting an\nalternative approach that relies solely on Time-Difference-of-Arrival (TDoA)\ninformation. To this end, a novel geometric framework for micro bubble\nlocalization via ellipse intersections is proposed to overcome existing\nbeamforming limitations. We present a benchmark comparison based on a public\ndataset for which our geometric ULM outperforms existing baseline methods in\nterms of accuracy and robustness while only utilizing a portion of the\navailable transducer data.\n", "date": "2023-06-27", "categories": []}
{"id": "2307.08950", "title": "Deep Physics-Guided Unrolling Generalization for Compressed Sensing", "abstract": "  By absorbing the merits of both the model- and data-driven methods, deep\nphysics-engaged learning scheme achieves high-accuracy and interpretable image\nreconstruction. It has attracted growing attention and become the mainstream\nfor inverse imaging tasks. Focusing on the image compressed sensing (CS)\nproblem, we find the intrinsic defect of this emerging paradigm, widely\nimplemented by deep algorithm-unrolled networks, in which more plain iterations\ninvolving real physics will bring enormous computation cost and long inference\ntime, hindering their practical application. A novel deep\n$\\textbf{P}$hysics-guided un$\\textbf{R}$olled recovery $\\textbf{L}$earning\n($\\textbf{PRL}$) framework is proposed by generalizing the traditional\niterative recovery model from image domain (ID) to the high-dimensional feature\ndomain (FD). A compact multiscale unrolling architecture is then developed to\nenhance the network capacity and keep real-time inference speeds. Taking two\ndifferent perspectives of optimization and range-nullspace decomposition,\ninstead of building an algorithm-specific unrolled network, we provide two\nimplementations: $\\textbf{PRL-PGD}$ and $\\textbf{PRL-RND}$. Experiments exhibit\nthe significant performance and efficiency leading of PRL networks over other\nstate-of-the-art methods with a large potential for further improvement and\nreal application to other inverse imaging problems or optimization models.\n", "date": "2023-07-18", "categories": []}
{"id": "2307.09441", "title": "Convergent regularization in inverse problems and linear plug-and-play\n  denoisers", "abstract": "  Plug-and-play (PnP) denoising is a popular iterative framework for solving\nimaging inverse problems using off-the-shelf image denoisers. Their empirical\nsuccess has motivated a line of research that seeks to understand the\nconvergence of PnP iterates under various assumptions on the denoiser. While a\nsignificant amount of research has gone into establishing the convergence of\nthe PnP iteration for different regularity conditions on the denoisers, not\nmuch is known about the asymptotic properties of the converged solution as the\nnoise level in the measurement tends to zero, i.e., whether PnP methods are\nprovably convergent regularization schemes under reasonable assumptions on the\ndenoiser. This paper serves two purposes: first, we provide an overview of the\nclassical regularization theory in inverse problems and survey a few notable\nrecent data-driven methods that are provably convergent regularization schemes.\nWe then continue to discuss PnP algorithms and their established convergence\nguarantees. Subsequently, we consider PnP algorithms with linear denoisers and\npropose a novel spectral filtering technique to control the strength of\nregularization arising from the denoiser. Further, by relating the implicit\nregularization of the denoiser to an explicit regularization functional, we\nrigorously show that PnP with linear denoisers leads to a convergent\nregularization scheme. More specifically, we prove that in the limit as the\nnoise vanishes, the PnP reconstruction converges to the minimizer of a\nregularization potential subject to the solution satisfying the noiseless\noperator equation. The theoretical analysis is corroborated by numerical\nexperiments for the classical inverse problem of tomographic image\nreconstruction.\n", "date": "2023-07-18", "categories": []}
{"id": "2307.07291", "title": "Sampling-Priors-Augmented Deep Unfolding Network for Robust Video\n  Compressive Sensing", "abstract": "  Video Compressed Sensing (VCS) aims to reconstruct multiple frames from one\nsingle captured measurement, thus achieving high-speed scene recording with a\nlow-frame-rate sensor. Although there have been impressive advances in VCS\nrecently, those state-of-the-art (SOTA) methods also significantly increase\nmodel complexity and suffer from poor generality and robustness, which means\nthat those networks need to be retrained to accommodate the new system. Such\nlimitations hinder the real-time imaging and practical deployment of models. In\nthis work, we propose a Sampling-Priors-Augmented Deep Unfolding Network\n(SPA-DUN) for efficient and robust VCS reconstruction. Under the\noptimization-inspired deep unfolding framework, a lightweight and efficient\nU-net is exploited to downsize the model while improving overall performance.\nMoreover, the prior knowledge from the sampling model is utilized to\ndynamically modulate the network features to enable single SPA-DUN to handle\narbitrary sampling settings, augmenting interpretability and generality.\nExtensive experiments on both simulation and real datasets demonstrate that\nSPA-DUN is not only applicable for various sampling settings with one single\nmodel but also achieves SOTA performance with incredible efficiency.\n", "date": "2023-07-14", "categories": []}
{"id": "2304.11549", "title": "Spectral Sensitivity Estimation Without a Camera", "abstract": "  A number of problems in computer vision and related fields would be mitigated\nif camera spectral sensitivities were known. As consumer cameras are not\ndesigned for high-precision visual tasks, manufacturers do not disclose\nspectral sensitivities. Their estimation requires a costly optical setup, which\ntriggered researchers to come up with numerous indirect methods that aim to\nlower cost and complexity by using color targets. However, the use of color\ntargets gives rise to new complications that make the estimation more\ndifficult, and consequently, there currently exists no simple, low-cost, robust\ngo-to method for spectral sensitivity estimation. Furthermore, even if not\nlimited by hardware or cost, researchers frequently work with imagery from\nmultiple cameras that they do not have in their possession. To provide a\npractical solution to this problem, we propose a framework for spectral\nsensitivity estimation that not only does not require any hardware, but also\ndoes not require physical access to the camera itself. Similar to other work,\nwe formulate an optimization problem that minimizes a two-term objective\nfunction: a camera-specific term from a system of equations, and a universal\nterm that bounds the solution space. Different than other work, we use publicly\navailable high-quality calibration data to construct both terms. We use the\ncolorimetric mapping matrices provided by the Adobe DNG Converter to formulate\nthe camera-specific system of equations, and constrain the solutions using an\nautoencoder trained on a database of ground-truth curves. On average, we\nachieve reconstruction errors as low as those that can arise due to\nmanufacturing imperfections between two copies of the same camera. We provide\nour code and predicted sensitivities for 1,000+ cameras, and discuss which\ntasks can become trivial when camera responses are available.\n", "date": "2023-04-23", "categories": []}
{"id": "2305.19767", "title": "Analytical reconstructions of full-scan multiple source-translation\n  computed tomography under large field of views", "abstract": "  This paper is to investigate the high-quality analytical reconstructions of\nmultiple source-translation computed tomography (mSTCT) under an extended field\nof view (FOV). Under the larger FOVs, the previously proposed backprojection\nfiltration (BPF) algorithms for mSTCT, including D-BPF and S-BPF (their\ndifferences are different derivate directions along the detector and source,\nrespectively), make some errors and artifacts in the reconstructed images due\nto a backprojection weighting factor and the half-scan mode, which deviates\nfrom the intention of mSTCT imaging. In this paper, to achieve reconstruction\nwith as little error as possible under the extremely extended FOV, we combine\nthe full-scan mSTCT (F-mSTCT) geometry with the previous BPF algorithms to\nstudy the performance and derive a suitable redundancy-weighted function for\nF-mSTCT. The experimental results indicate FS-BPF can get high-quality, stable\nimages under the extremely extended FOV of imaging a large object, though it\nrequires more projections than FD-BPF. Finally, for different practical\nrequirements in extending FOV imaging, we give suggestions on algorithm\nselection.\n", "date": "2023-05-31", "categories": []}
{"id": "2307.06277", "title": "Stochastic Light Field Holography", "abstract": "  The Visual Turing Test is the ultimate goal to evaluate the realism of\nholographic displays. Previous studies have focused on addressing challenges\nsuch as limited \\'etendue and image quality over a large focal volume, but they\nhave not investigated the effect of pupil sampling on the viewing experience in\nfull 3D holograms. In this work, we tackle this problem with a novel hologram\ngeneration algorithm motivated by matching the projection operators of\nincoherent Light Field and coherent Wigner Function light transport. To this\nend, we supervise hologram computation using synthesized photographs, which are\nrendered on-the-fly using Light Field refocusing from stochastically sampled\npupil states during optimization. The proposed method produces holograms with\ncorrect parallax and focus cues, which are important for passing the Visual\nTuring Test. We validate that our approach compares favorably to\nstate-of-the-art CGH algorithms that use Light Field and Focal Stack\nsupervision. Our experiments demonstrate that our algorithm significantly\nimproves the realism of the viewing experience for a variety of different pupil\nstates.\n", "date": "2023-07-12", "categories": []}
{"id": "2307.04924", "title": "Count-Free Single-Photon 3D Imaging with Race Logic", "abstract": "  Single-photon cameras (SPCs) have emerged as a promising technology for\nhigh-resolution 3D imaging. A single-photon 3D camera determines the round-trip\ntime of a laser pulse by capturing the arrival of individual photons at each\ncamera pixel. Constructing photon-timestamp histograms is a fundamental\noperation for a single-photon 3D camera. However, in-pixel histogram processing\nis computationally expensive and requires large amount of memory per pixel.\nDigitizing and transferring photon timestamps to an off-sensor histogramming\nmodule is bandwidth and power hungry. Here we present an online approach for\ndistance estimation without explicitly storing photon counts. The two key\ningredients of our approach are (a) processing photon streams using race logic,\nwhich maintains photon data in the time-delay domain, and (b) constructing\ncount-free equi-depth histograms. Equi-depth histograms are a succinct\nrepresentation for ``peaky'' distributions, such as those obtained by an SPC\npixel from a laser pulse reflected by a surface. Our approach uses a binner\nelement that converges on the median (or, more generally, to another quantile)\nof a distribution. We cascade multiple binners to form an equi-depth\nhistogrammer that produces multi-bin histograms. Our evaluation shows that this\nmethod can provide an order of magnitude reduction in bandwidth and power\nconsumption while maintaining similar distance reconstruction accuracy as\nconventional processing methods.\n", "date": "2023-07-10", "categories": []}
{"id": "2307.04946", "title": "DDGM: Solving inverse problems by Diffusive Denoising of Gradient-based\n  Minimization", "abstract": "  Inverse problems generally require a regularizer or prior for a good\nsolution. A recent trend is to train a convolutional net to denoise images, and\nuse this net as a prior when solving the inverse problem. Several proposals\ndepend on a singular value decomposition of the forward operator, and several\nothers backpropagate through the denoising net at runtime. Here we propose a\nsimpler approach that combines the traditional gradient-based minimization of\nreconstruction error with denoising. Noise is also added at each step, so the\niterative dynamics resembles a Langevin or diffusion process. Both the level of\nadded noise and the size of the denoising step decay exponentially with time.\nWe apply our method to the problem of tomographic reconstruction from electron\nmicrographs acquired at multiple tilt angles. With empirical studies using\nsimulated tilt views, we find parameter settings for our method that produce\ngood results. We show that high accuracy can be achieved with as few as 50\ndenoising steps. We also compare with DDRM and DPS, more complex diffusion\nmethods of the kinds mentioned above. These methods are less accurate (as\nmeasured by MSE and SSIM) for our tomography problem, even after the generation\nhyperparameters are optimized. Finally we extend our method to reconstruction\nof arbitrary-sized images and show results on 128 $\\times$ 1568 pixel images\n", "date": "2023-07-11", "categories": []}
{"id": "2307.05087", "title": "SAR-NeRF: Neural Radiance Fields for Synthetic Aperture Radar Multi-View\n  Representation", "abstract": "  SAR images are highly sensitive to observation configurations, and they\nexhibit significant variations across different viewing angles, making it\nchallenging to represent and learn their anisotropic features. As a result,\ndeep learning methods often generalize poorly across different view angles.\nInspired by the concept of neural radiance fields (NeRF), this study combines\nSAR imaging mechanisms with neural networks to propose a novel NeRF model for\nSAR image generation. Following the mapping and projection pinciples, a set of\nSAR images is modeled implicitly as a function of attenuation coefficients and\nscattering intensities in the 3D imaging space through a differentiable\nrendering equation. SAR-NeRF is then constructed to learn the distribution of\nattenuation coefficients and scattering intensities of voxels, where the\nvectorized form of 3D voxel SAR rendering equation and the sampling\nrelationship between the 3D space voxels and the 2D view ray grids are\nanalytically derived. Through quantitative experiments on various datasets, we\nthoroughly assess the multi-view representation and generalization capabilities\nof SAR-NeRF. Additionally, it is found that SAR-NeRF augumented dataset can\nsignificantly improve SAR target classification performance under few-shot\nlearning setup, where a 10-type classification accuracy of 91.6\\% can be\nachieved by using only 12 images per class.\n", "date": "2023-07-11", "categories": []}
{"id": "2307.05270", "title": "APRF: Anti-Aliasing Projection Representation Field for Inverse Problem\n  in Imaging", "abstract": "  Sparse-view Computed Tomography (SVCT) reconstruction is an ill-posed inverse\nproblem in imaging that aims to acquire high-quality CT images based on\nsparsely-sampled measurements. Recent works use Implicit Neural Representations\n(INRs) to build the coordinate-based mapping between sinograms and CT images.\nHowever, these methods have not considered the correlation between adjacent\nprojection views, resulting in aliasing artifacts on SV sinograms. To address\nthis issue, we propose a self-supervised SVCT reconstruction method --\nAnti-Aliasing Projection Representation Field (APRF), which can build the\ncontinuous representation between adjacent projection views via the spatial\nconstraints. Specifically, APRF only needs SV sinograms for training, which\nfirst employs a line-segment sampling module to estimate the distribution of\nprojection views in a local region, and then synthesizes the corresponding\nsinogram values using center-based line integral module. After training APRF on\na single SV sinogram itself, it can synthesize the corresponding dense-view\n(DV) sinogram with consistent continuity. High-quality CT images can be\nobtained by applying re-projection techniques on the predicted DV sinograms.\nExtensive experiments on CT images demonstrate that APRF outperforms\nstate-of-the-art methods, yielding more accurate details and fewer artifacts.\nOur code will be publicly available soon.\n", "date": "2023-07-11", "categories": []}
{"id": "2307.04484", "title": "Invertible Low-Dimensional Modelling of X-ray Absorption Spectra for\n  Potential Applications in Spectral X-ray Imaging", "abstract": "  X-ray interaction with matter is an energy-dependent process that is\ncontingent on the atomic structure of the constituent material elements. The\nmost advanced models to capture this relationship currently rely on Monte Carlo\n(MC) simulations. Whilst these very accurate models, in many problems in\nspectral X-ray imaging, such as data compression, noise removal, spectral\nestimation, and the quantitative measurement of material compositions, these\nmodels are of limited use, as these applications typically require the\nefficient inversion of the model, that is, they require the estimation of the\nbest model parameters for a given spectral measurement. Current models that can\nbe easily inverted however typically only work when modelling spectra in\nregions away from their K-edges, so they have limited utility when modelling a\nwider range of materials. In this paper, we thus propose a novel, non-linear\nmodel that combines a deep neural network autoencoder with an optimal linear\nmodel based on the Singular Value Decomposition (SVD). We compare our new\nmethod to other alternative linear and non-linear approaches, a sparse model\nand an alternative deep learning model. We demonstrate the advantages of our\nmethod over traditional models, especially when modelling X-ray absorption\nspectra that contain K-edges in the energy range of interest.\n", "date": "2023-07-10", "categories": []}
{"id": "2307.03413", "title": "Unsupervised Hyperspectral and Multispectral Images Fusion Based on the\n  Cycle Consistency", "abstract": "  Hyperspectral images (HSI) with abundant spectral information reflected\nmaterials property usually perform low spatial resolution due to the hardware\nlimits. Meanwhile, multispectral images (MSI), e.g., RGB images, have a high\nspatial resolution but deficient spectral signatures. Hyperspectral and\nmultispectral image fusion can be cost-effective and efficient for acquiring\nboth high spatial resolution and high spectral resolution images. Many of the\nconventional HSI and MSI fusion algorithms rely on known spatial degradation\nparameters, i.e., point spread function, spectral degradation parameters,\nspectral response function, or both of them. Another class of deep\nlearning-based models relies on the ground truth of high spatial resolution HSI\nand needs large amounts of paired training images when working in a supervised\nmanner. Both of these models are limited in practical fusion scenarios. In this\npaper, we propose an unsupervised HSI and MSI fusion model based on the cycle\nconsistency, called CycFusion. The CycFusion learns the domain transformation\nbetween low spatial resolution HSI (LrHSI) and high spatial resolution MSI\n(HrMSI), and the desired high spatial resolution HSI (HrHSI) are considered to\nbe intermediate feature maps in the transformation networks. The CycFusion can\nbe trained with the objective functions of marginal matching in single\ntransform and cycle consistency in double transforms. Moreover, the estimated\nPSF and SRF are embedded in the model as the pre-training weights, which\nfurther enhances the practicality of our proposed model. Experiments conducted\non several datasets show that our proposed model outperforms all compared\nunsupervised fusion methods. The codes of this paper will be available at this\naddress: https: //github.com/shuaikaishi/CycFusion for reproducibility.\n", "date": "2023-07-07", "categories": []}
{"id": "2307.02875", "title": "Reference-based Motion Blur Removal: Learning to Utilize Sharpness in\n  the Reference Image", "abstract": "  Despite the recent advancement in the study of removing motion blur in an\nimage, it is still hard to deal with strong blurs. While there are limits in\nremoving blurs from a single image, it has more potential to use multiple\nimages, e.g., using an additional image as a reference to deblur a blurry\nimage. A typical setting is deburring an image using a nearby sharp image(s) in\na video sequence, as in the studies of video deblurring. This paper proposes a\nbetter method to use the information present in a reference image. The method\ndoes not need a strong assumption on the reference image. We can utilize an\nalternative shot of the identical scene, just like in video deblurring, or we\ncan even employ a distinct image from another scene. Our method first matches\nlocal patches of the target and reference images and then fuses their features\nto estimate a sharp image. We employ a patch-based feature matching strategy to\nsolve the difficult problem of matching the blurry image with the sharp\nreference. Our method can be integrated into pre-existing networks designed for\nsingle image deblurring. The experimental results show the effectiveness of the\nproposed method.\n", "date": "2023-07-06", "categories": []}
{"id": "2307.01990", "title": "Unsupervised Spectral Demosaicing with Lightweight Spectral Attention\n  Networks", "abstract": "  This paper presents a deep learning-based spectral demosaicing technique\ntrained in an unsupervised manner. Many existing deep learning-based techniques\nrelying on supervised learning with synthetic images, often underperform on\nreal-world images especially when the number of spectral bands increases.\nAccording to the characteristics of the spectral mosaic image, this paper\nproposes a mosaic loss function, the corresponding model structure, a\ntransformation strategy, and an early stopping strategy, which form a complete\nunsupervised spectral demosaicing framework. A challenge in real-world spectral\ndemosaicing is inconsistency between the model parameters and the computational\nresources of the imager. We reduce the complexity and parameters of the\nspectral attention module by dividing the spectral attention tensor into\nspectral attention matrices in the spatial dimension and spectral attention\nvector in the channel dimension, which is more suitable for unsupervised\nframework. This paper also presents Mosaic25, a real 25-band hyperspectral\nmosaic image dataset of various objects, illuminations, and materials for\nbenchmarking. Extensive experiments on synthetic and real-world datasets\ndemonstrate that the proposed method outperforms conventional unsupervised\nmethods in terms of spatial distortion suppression, spectral fidelity,\nrobustness, and computational cost.\n", "date": "2023-07-05", "categories": []}
{"id": "2307.00269", "title": "AE-RED: A Hyperspectral Unmixing Framework Powered by Deep Autoencoder\n  and Regularization by Denoising", "abstract": "  Spectral unmixing has been extensively studied with a variety of methods and\nused in many applications. Recently, data-driven techniques with deep learning\nmethods have obtained great attention to spectral unmixing for its superior\nlearning ability to automatically learn the structure information. In\nparticular, autoencoder based architectures are elaborately designed to solve\nblind unmixing and model complex nonlinear mixtures. Nevertheless, these\nmethods perform unmixing task as blackboxes and lack of interpretability. On\nthe other hand, conventional unmixing methods carefully design the regularizer\nto add explicit information, in which algorithms such as plug-and-play (PnP)\nstrategies utilize off-the-shelf denoisers to plug powerful priors. In this\npaper, we propose a generic unmixing framework to integrate the autoencoder\nnetwork with regularization by denoising (RED), named AE-RED. More specially,\nwe decompose the unmixing optimized problem into two subproblems. The first one\nis solved using deep autoencoders to implicitly regularize the estimates and\nmodel the mixture mechanism. The second one leverages the denoiser to bring in\nthe explicit information. In this way, both the characteristics of the deep\nautoencoder based unmixing methods and priors provided by denoisers are merged\ninto our well-designed framework to enhance the unmixing performance.\nExperiment results on both synthetic and real data sets show the superiority of\nour proposed framework compared with state-of-the-art unmixing approaches.\n", "date": "2023-07-01", "categories": []}
{"id": "2307.00619", "title": "Solving Linear Inverse Problems Provably via Posterior Sampling with\n  Latent Diffusion Models", "abstract": "  We present the first framework to solve linear inverse problems leveraging\npre-trained latent diffusion models. Previously proposed algorithms (such as\nDPS and DDRM) only apply to pixel-space diffusion models. We theoretically\nanalyze our algorithm showing provable sample recovery in a linear model\nsetting. The algorithmic insight obtained from our analysis extends to more\ngeneral settings often considered in practice. Experimentally, we outperform\npreviously proposed posterior sampling algorithms in a wide variety of problems\nincluding random inpainting, block inpainting, denoising, deblurring,\ndestriping, and super-resolution.\n", "date": "2023-07-02", "categories": []}
{"id": "2307.00821", "title": "Unveiling the Potential of Spike Streams for Foreground Occlusion\n  Removal from Densely Continuous Views", "abstract": "  The extraction of a clean background image by removing foreground occlusion\nholds immense practical significance, but it also presents several challenges.\nPresently, the majority of de-occlusion research focuses on addressing this\nissue through the extraction and synthesis of discrete images from calibrated\ncamera arrays. Nonetheless, the restoration quality tends to suffer when faced\nwith dense occlusions or high-speed motions due to limited perspectives and\nmotion blur. To successfully remove dense foreground occlusion, an effective\nmulti-view visual information integration approach is required. Introducing the\nspike camera as a novel type of neuromorphic sensor offers promising\ncapabilities with its ultra-high temporal resolution and high dynamic range. In\nthis paper, we propose an innovative solution for tackling the de-occlusion\nproblem through continuous multi-view imaging using only one spike camera\nwithout any prior knowledge of camera intrinsic parameters and camera poses. By\nrapidly moving the spike camera, we continually capture the dense stream of\nspikes from the occluded scene. To process the spikes, we build a novel model\n\\textbf{SpkOccNet}, in which we integrate information of spikes from continuous\nviewpoints within multi-windows, and propose a novel cross-view mutual\nattention mechanism for effective fusion and refinement. In addition, we\ncontribute the first real-world spike-based dataset \\textbf{S-OCC} for\nocclusion removal. The experimental results demonstrate that our proposed model\nefficiently removes dense occlusions in diverse scenes while exhibiting strong\ngeneralization.\n", "date": "2023-07-03", "categories": []}
{"id": "2306.17367", "title": "Spatially Varying Exposure with 2-by-2 Multiplexing: Optimality and\n  Universality", "abstract": "  The advancement of new digital image sensors has enabled the design of\nexposure multiplexing schemes where a single image capture can have multiple\nexposures and conversion gains in an interlaced format, similar to that of a\nBayer color filter array. In this paper, we ask the question of how to design\nsuch multiplexing schemes for adaptive high-dynamic range (HDR) imaging where\nthe multiplexing scheme can be updated according to the scenes. We present two\nnew findings.\n  (i) We address the problem of design optimality. We show that given a\nmultiplex pattern, the conventional optimality criteria based on the\ninput/output-referred signal-to-noise ratio (SNR) of the independently measured\npixels can lead to flawed decisions because it cannot encapsulate the location\nof the saturated pixels. We overcome the issue by proposing a new concept known\nas the spatially varying exposure risk (SVE-Risk) which is a pseudo-idealistic\nquantification of the amount of recoverable pixels. We present an efficient\nenumeration algorithm to select the optimal multiplex patterns.\n  (ii) We report a design universality observation that the design of the\nmultiplex pattern can be decoupled from the image reconstruction algorithm.\nThis is a significant departure from the recent literature that the multiplex\npattern should be jointly optimized with the reconstruction algorithm. Our\nfinding suggests that in the context of exposure multiplexing, an end-to-end\ntraining may not be necessary.\n", "date": "2023-06-30", "categories": []}
{"id": "2306.17618", "title": "Polarimetric iToF: Measuring High-Fidelity Depth through Scattering\n  Media", "abstract": "  Indirect time-of-flight (iToF) imaging allows us to capture dense depth\ninformation at a low cost. However, iToF imaging often suffers from multipath\ninterference (MPI) artifacts in the presence of scattering media, resulting in\nsevere depth-accuracy degradation. For instance, iToF cameras cannot measure\ndepth accurately through fog because ToF active illumination scatters back to\nthe sensor before reaching the farther target surface. In this work, we propose\na polarimetric iToF imaging method that can capture depth information robustly\nthrough scattering media. Our observations on the principle of indirect ToF\nimaging and polarization of light allow us to formulate a novel computational\nmodel of scattering-aware polarimetric phase measurements that enables us to\ncorrect MPI errors. We first devise a scattering-aware polarimetric iToF model\nthat can estimate the phase of unpolarized backscattered light. We then combine\nthe optical filtering of polarization and our computational modeling of\nunpolarized backscattered light via scattering analysis of phase and amplitude.\nThis allows us to tackle the MPI problem by estimating the scattering energy\nthrough the participating media. We validate our method on an experimental\nsetup using a customized off-the-shelf iToF camera. Our method outperforms\nbaseline methods by a significant margin by means of our scattering model and\npolarimetric phase measurements.\n", "date": "2023-06-30", "categories": []}
{"id": "2306.15953", "title": "Angle Sensitive Pixels for Lensless Imaging on Spherical Sensors", "abstract": "  We propose OrbCam, a lensless architecture for imaging with spherical\nsensors. Prior work in lensless imager techniques have focused largely on using\nplanar sensors; for such designs, it is important to use a modulation element,\ne.g. amplitude or phase masks, to construct a invertible imaging system. In\ncontrast, we show that the diversity of pixel orientations on a curved surface\nis sufficient to improve the conditioning of the mapping between the scene and\nthe sensor. Hence, when imaging on a spherical sensor, all pixels can have the\nsame angular response function such that the lensless imager is comprised of\npixels that are identical to each other and differ only in their orientations.\nWe provide the computational tools for the design of the angular response of\nthe pixels in a spherical sensor that leads to well-conditioned and\nnoise-robust measurements. We validate our design in both simulation and a lab\nprototype. The implications of our design is that the lensless imaging can be\nenabled easily for curved and flexible surfaces thereby opening up a new set of\napplication domains.\n", "date": "2023-06-28", "categories": []}
{"id": "2211.16881", "title": "Generalized Deep Learning-based Proximal Gradient Descent for MR\n  Reconstruction", "abstract": "  The data consistency for the physical forward model is crucial in inverse\nproblems, especially in MR imaging reconstruction. The standard way is to\nunroll an iterative algorithm into a neural network with a forward model\nembedded. The forward model always changes in clinical practice, so the\nlearning component's entanglement with the forward model makes the\nreconstruction hard to generalize. The deep learning-based proximal gradient\ndescent was proposed and use a network as regularization term that is\nindependent of the forward model, which makes it more generalizable for\ndifferent MR acquisition settings. This one-time pre-trained regularization is\napplied to different MR acquisition settings and was compared to conventional\nL1 regularization showing ~3 dB improvement in the peak signal-to-noise ratio.\nWe also demonstrated the flexibility of the proposed method in choosing\ndifferent undersampling patterns.\n", "date": "2022-11-30", "categories": []}
{"id": "2301.12686", "title": "GibbsDDRM: A Partially Collapsed Gibbs Sampler for Solving Blind Inverse\n  Problems with Denoising Diffusion Restoration", "abstract": "  Pre-trained diffusion models have been successfully used as priors in a\nvariety of linear inverse problems, where the goal is to reconstruct a signal\nfrom noisy linear measurements. However, existing approaches require knowledge\nof the linear operator. In this paper, we propose GibbsDDRM, an extension of\nDenoising Diffusion Restoration Models (DDRM) to a blind setting in which the\nlinear measurement operator is unknown. GibbsDDRM constructs a joint\ndistribution of the data, measurements, and linear operator by using a\npre-trained diffusion model for the data prior, and it solves the problem by\nposterior sampling with an efficient variant of a Gibbs sampler. The proposed\nmethod is problem-agnostic, meaning that a pre-trained diffusion model can be\napplied to various inverse problems without fine-tuning. In experiments, it\nachieved high performance on both blind image deblurring and vocal\ndereverberation tasks, despite the use of simple generic priors for the\nunderlying linear operators.\n", "date": "2023-01-30", "categories": []}
{"id": "2305.02074", "title": "A Vision Transformer Approach for Efficient Near-Field Irregular SAR\n  Super-Resolution", "abstract": "  In this paper, we develop a novel super-resolution algorithm for near-field\nsynthetic-aperture radar (SAR) under irregular scanning geometries. As\nfifth-generation (5G) millimeter-wave (mmWave) devices are becoming\nincreasingly affordable and available, high-resolution SAR imaging is feasible\nfor end-user applications and non-laboratory environments. Emerging\napplications such freehand imaging, wherein a handheld radar is scanned\nthroughout space by a user, unmanned aerial vehicle (UAV) imaging, and\nautomotive SAR face several unique challenges for high-resolution imaging.\nFirst, recovering a SAR image requires knowledge of the array positions\nthroughout the scan. While recent work has introduced camera-based positioning\nsystems capable of adequately estimating the position, recovering the algorithm\nefficiently is a requirement to enable edge and Internet of Things (IoT)\ntechnologies. Efficient algorithms for non-cooperative near-field SAR sampling\nhave been explored in recent work, but suffer image defocusing under position\nestimation error and can only produce medium-fidelity images. In this paper, we\nintroduce a mobile-friend vision transformer (ViT) architecture to address\nposition estimation error and perform SAR image super-resolution (SR) under\nirregular sampling geometries. The proposed algorithm, Mobile-SRViT, is the\nfirst to employ a ViT approach for SAR image enhancement and is validated in\nsimulation and via empirical studies.\n", "date": "2023-05-03", "categories": []}
{"id": "2306.13361", "title": "Neural 360$^\\circ$ Structured Light with Learned Metasurfaces", "abstract": "  Structured light has proven instrumental in 3D imaging, LiDAR, and\nholographic light projection. Metasurfaces, comprised of sub-wavelength-sized\nnanostructures, facilitate 180$^\\circ$ field-of-view (FoV) structured light,\ncircumventing the restricted FoV inherent in traditional optics like\ndiffractive optical elements. However, extant metasurface-facilitated\nstructured light exhibits sub-optimal performance in downstream tasks, due to\nheuristic pattern designs such as periodic dots that do not consider the\nobjectives of the end application. In this paper, we present neural 360$^\\circ$\nstructured light, driven by learned metasurfaces. We propose a differentiable\nframework, that encompasses a computationally-efficient 180$^\\circ$ wave\npropagation model and a task-specific reconstructor, and exploits both\ntransmission and reflection channels of the metasurface. Leveraging a\nfirst-order optimizer within our differentiable framework, we optimize the\nmetasurface design, thereby realizing neural 360$^\\circ$ structured light. We\nhave utilized neural 360$^\\circ$ structured light for holographic light\nprojection and 3D imaging. Specifically, we demonstrate the first 360$^\\circ$\nlight projection of complex patterns, enabled by our propagation model that can\nbe computationally evaluated 50,000$\\times$ faster than the Rayleigh-Sommerfeld\npropagation. For 3D imaging, we improve depth-estimation accuracy by\n5.09$\\times$ in RMSE compared to the heuristically-designed structured light.\nNeural 360$^\\circ$ structured light promises robust 360$^\\circ$ imaging and\ndisplay for robotics, extended-reality systems, and human-computer\ninteractions.\n", "date": "2023-06-23", "categories": []}
{"id": "2306.13843", "title": "Score-based Generative Models for Photoacoustic Image Reconstruction\n  with Rotation Consistency Constraints", "abstract": "  Photoacoustic tomography (PAT) is a newly emerged imaging modality which\nenables both high optical contrast and acoustic depth of penetration.\nReconstructing images of photoacoustic tomography from limited amount of senser\ndata is among one of the major challenges in photoacoustic imaging. Previous\nworks based on deep learning were trained in supervised fashion, which directly\nmap the input partially known sensor data to the ground truth reconstructed\nfrom full field of view. Recently, score-based generative models played an\nincreasingly significant role in generative modeling. Leveraging this\nprobabilistic model, we proposed Rotation Consistency Constrained Score-based\nGenerative Model (RCC-SGM), which recovers the PAT images by iterative sampling\nbetween Langevin dynamics and a constraint term utilizing the rotation\nconsistency between the images and the measurements. Our proposed method can\ngeneralize to different measurement processes (32.29 PSNR with 16 measurements\nunder random sampling, whereas 28.50 for supervised counterpart), while\nsupervised methods need to train on specific inverse mappings.\n", "date": "2023-06-24", "categories": []}
{"id": "2306.13881", "title": "Current density impedance imaging with PINNs", "abstract": "  In this paper, we introduce CDII-PINNs, a computationally efficient method\nfor solving CDII using PINNs in the framework of Tikhonov regularization. This\nmethod constructs a physics-informed loss function by merging the regularized\nleast-squares output functional with an underlying differential equation, which\ndescribes the relationship between the conductivity and voltage. A pair of\nneural networks representing the conductivity and voltage, respectively, are\ncoupled by this loss function. Then, minimizing the loss function provides a\nreconstruction. A rigorous theoretical guarantee is provided. We give an error\nanalysis for CDII-PINNs and establish a convergence rate, based on prior\nselected neural network parameters in terms of the number of samples. The\nnumerical simulations demonstrate that CDII-PINNs are efficient, accurate and\nrobust to noise levels ranging from $1\\%$ to $20\\%$.\n", "date": "2023-06-24", "categories": []}
{"id": "2306.11750", "title": "Optical Coherence Tomography Image Enhancement via Block Hankelization\n  and Low Rank Tensor Network Approximation", "abstract": "  In this paper, the problem of image super-resolution for Optical Coherence\nTomography (OCT) has been addressed. Due to the motion artifacts, OCT imaging\nis usually done with a low sampling rate and the resulting images are often\nnoisy and have low resolution. Therefore, reconstruction of high resolution OCT\nimages from the low resolution versions is an essential step for better OCT\nbased diagnosis. In this paper, we propose a novel OCT super-resolution\ntechnique using Tensor Ring decomposition in the embedded space. A new\ntensorization method based on a block Hankelization approach with overlapped\npatches, called overlapped patch Hankelization, has been proposed which allows\nus to employ Tensor Ring decomposition. The Hankelization method enables us to\nbetter exploit the inter connection of pixels and consequently achieve better\nsuper-resolution of images. The low resolution image was first patch Hankelized\nand then its Tensor Ring decomposition with rank incremental has been computed.\nSimulation results confirm that the proposed approach is effective in OCT\nsuper-resolution.\n", "date": "2023-06-19", "categories": []}
{"id": "2206.04514", "title": "SAR Despeckling using a Denoising Diffusion Probabilistic Model", "abstract": "  Speckle is a multiplicative noise which affects all coherent imaging\nmodalities including Synthetic Aperture Radar (SAR) images. The presence of\nspeckle degrades the image quality and adversely affects the performance of SAR\nimage understanding applications such as automatic target recognition and\nchange detection. Thus, SAR despeckling is an important problem in remote\nsensing. In this paper, we introduce SAR-DDPM, a denoising diffusion\nprobabilistic model for SAR despeckling. The proposed method comprises of a\nMarkov chain that transforms clean images to white Gaussian noise by repeatedly\nadding random noise. The despeckled image is recovered by a reverse process\nwhich iteratively predicts the added noise using a noise predictor which is\nconditioned on the speckled image. In addition, we propose a new inference\nstrategy based on cycle spinning to improve the despeckling performance. Our\nexperiments on both synthetic and real SAR images demonstrate that the proposed\nmethod achieves significant improvements in both quantitative and qualitative\nresults over the state-of-the-art despeckling methods.\n", "date": "2022-06-09", "categories": []}
{"id": "2212.08479", "title": "Neural Implicit k-Space for Binning-free Non-Cartesian Cardiac MR\n  Imaging", "abstract": "  In this work, we propose a novel image reconstruction framework that directly\nlearns a neural implicit representation in k-space for ECG-triggered\nnon-Cartesian Cardiac Magnetic Resonance Imaging (CMR). While existing methods\nbin acquired data from neighboring time points to reconstruct one phase of the\ncardiac motion, our framework allows for a continuous, binning-free, and\nsubject-specific k-space representation.We assign a unique coordinate that\nconsists of time, coil index, and frequency domain location to each sampled\nk-space point. We then learn the subject-specific mapping from these unique\ncoordinates to k-space intensities using a multi-layer perceptron with\nfrequency domain regularization. During inference, we obtain a complete k-space\nfor Cartesian coordinates and an arbitrary temporal resolution. A simple\ninverse Fourier transform recovers the image, eliminating the need for density\ncompensation and costly non-uniform Fourier transforms for non-Cartesian data.\nThis novel imaging framework was tested on 42 radially sampled datasets from 6\nsubjects. The proposed method outperforms other techniques qualitatively and\nquantitatively using data from four and one heartbeat(s) and 30 cardiac phases.\nOur results for one heartbeat reconstruction of 50 cardiac phases show improved\nartifact removal and spatio-temporal resolution, leveraging the potential for\nreal-time CMR.\n", "date": "2022-12-16", "categories": []}
{"id": "2305.15750", "title": "Towards Large-scale Single-shot Millimeter-wave Imaging for Low-cost\n  Security Inspection", "abstract": "  Millimeter-wave (MMW) imaging is emerging as a promising technique for safe\nsecurity inspection. It achieves a delicate balance between imaging resolution,\npenetrability and human safety, resulting in higher resolution compared to\nlow-frequency microwave, stronger penetrability compared to visible light, and\nstronger safety compared to X ray. Despite of recent advance in the last\ndecades, the high cost of requisite large-scale antenna array hinders\nwidespread adoption of MMW imaging in practice. To tackle this challenge, we\nreport a large-scale single-shot MMW imaging framework using sparse antenna\narray, achieving low-cost but high-fidelity security inspection under an\ninterpretable learning scheme. We first collected extensive full-sampled MMW\nechoes to study the statistical ranking of each element in the large-scale\narray. These elements are then sampled based on the ranking, building the\nexperimentally optimal sparse sampling strategy that reduces the cost of\nantenna array by up to one order of magnitude. Additionally, we derived an\nuntrained interpretable learning scheme, which realizes robust and accurate\nimage reconstruction from sparsely sampled echoes. Last, we developed a neural\nnetwork for automatic object detection, and experimentally demonstrated\nsuccessful detection of concealed centimeter-sized targets using 10% sparse\narray, whereas all the other contemporary approaches failed at the same sample\nsampling ratio. The performance of the reported technique presents higher than\n50% superiority over the existing MMW imaging schemes on various metrics\nincluding precision, recall, and mAP50. With such strong detection ability and\norder-of-magnitude cost reduction, we anticipate that this technique provides a\npractical way for large-scale single-shot MMW imaging, and could advocate its\nfurther practical applications.\n", "date": "2023-05-25", "categories": []}
{"id": "2306.10201", "title": "Stretched sinograms for limited-angle tomographic reconstruction with\n  neural networks", "abstract": "  We present a direct method for limited angle tomographic reconstruction using\nconvolutional networks. The key to our method is to first stretch every tilt\nview in the direction perpendicular to the tilt axis by the secant of the tilt\nangle. These stretched views are then fed into a 2-D U-Net which directly\noutputs the 3-D reconstruction. We train our networks by minimizing the mean\nsquared error between the network's generated reconstruction and a ground truth\n3-D volume. To demonstrate and evaluate our method, we synthesize tilt views\nfrom a 3-D image of fly brain tissue acquired with Focused Ion Beam Scanning\nElectron Microscopy. We compare our method to using a U-Net to directly\nreconstruct the unstretched tilt views and show that this simple stretching\nprocedure leads to significantly better reconstructions. We also compare to\nusing a network to clean up reconstructions generated by backprojection and\nfiltered backprojection, and find that this simple stretching procedure also\ngives lower mean squared error on previously unseen images.\n", "date": "2023-06-16", "categories": []}
{"id": "2306.10520", "title": "RetinexFlow for CT metal artifact reduction", "abstract": "  Metal artifacts is a major challenge in computed tomography (CT) imaging,\nsignificantly degrading image quality and making accurate diagnosis difficult.\nHowever, previous methods either require prior knowledge of the location of\nmetal implants, or have modeling deviations with the mechanism of artifact\nformation, which limits the ability to obtain high-quality CT images. In this\nwork, we formulate metal artifacts reduction problem as a combination of\ndecomposition and completion tasks. And we propose RetinexFlow, which is a\nnovel end-to-end image domain model based on Retinex theory and conditional\nnormalizing flow, to solve it. Specifically, we first design a feature\ndecomposition encoder for decomposing the metal implant component and inherent\ncomponent, and extracting the inherent feature. Then, it uses a\nfeature-to-image flow module to complete the metal artifact-free CT image step\nby step through a series of invertible transformations. These designs are\nincorporated in our model with a coarse-to-fine strategy, enabling it to\nachieve superior performance. The experimental results on on simulation and\nclinical datasets show our method achieves better quantitative and qualitative\nresults, exhibiting better visual performance in artifact removal and image\nfidelity\n", "date": "2023-06-18", "categories": []}
{"id": "2306.11316", "title": "Unfolding Framework with Prior of Convolution-Transformer Mixture and\n  Uncertainty Estimation for Video Snapshot Compressive Imaging", "abstract": "  We consider the problem of video snapshot compressive imaging (SCI), where\nsequential high-speed frames are modulated by different masks and captured by a\nsingle measurement. The underlying principle of reconstructing multi-frame\nimages from only one single measurement is to solve an ill-posed problem. By\ncombining optimization algorithms and neural networks, deep unfolding networks\n(DUNs) score tremendous achievements in solving inverse problems. In this\npaper, our proposed model is under the DUN framework and we propose a 3D\nConvolution-Transformer Mixture (CTM) module with a 3D efficient and scalable\nattention model plugged in, which helps fully learn the correlation between\ntemporal and spatial dimensions by virtue of Transformer. To our best\nknowledge, this is the first time that Transformer is employed to video SCI\nreconstruction. Besides, to further investigate the high-frequency information\nduring the reconstruction process which are neglected in previous studies, we\nintroduce variance estimation characterizing the uncertainty on a\npixel-by-pixel basis. Extensive experimental results demonstrate that our\nproposed method achieves state-of-the-art (SOTA) (with a 1.2dB gain in PSNR\nover previous SOTA algorithm) results. We will release the code.\n", "date": "2023-06-20", "categories": []}
{"id": "2306.09379", "title": "1st Solution Places for CVPR 2023 UG$^2$+ Challenge Track 2.2-Coded\n  Target Restoration through Atmospheric Turbulence", "abstract": "  In this technical report, we briefly introduce the solution of our team\nVIELab-HUST for coded target restoration through atmospheric turbulence in CVPR\n2023 UG$^2$+ Track 2.2. In this task, we propose an efficient multi-stage\nframework to restore a high quality image from distorted frames. Specifically,\neach distorted frame is initially aligned using image registration to suppress\ngeometric distortion. We subsequently select the sharpest set of registered\nframes by employing a frame selection approach based on image sharpness, and\naverage them to produce an image that is largely free of geometric distortion,\nalbeit with blurriness. A learning-based deblurring method is then applied to\nremove the residual blur in the averaged image. Finally, post-processing\ntechniques are utilized to further enhance the quality of the output image. Our\nframework is capable of handling different kinds of coded target dataset\nprovided in the final testing phase, and ranked 1st on the final leaderboard.\nOur code will be available at https://github.com/xsqhust/Turbulence_Removal.\n", "date": "2023-06-15", "categories": []}
{"id": "2306.09909", "title": "Neural Volumetric Reconstruction for Coherent Synthetic Aperture Sonar", "abstract": "  Synthetic aperture sonar (SAS) measures a scene from multiple views in order\nto increase the resolution of reconstructed imagery. Image reconstruction\nmethods for SAS coherently combine measurements to focus acoustic energy onto\nthe scene. However, image formation is typically under-constrained due to a\nlimited number of measurements and bandlimited hardware, which limits the\ncapabilities of existing reconstruction methods. To help meet these challenges,\nwe design an analysis-by-synthesis optimization that leverages recent advances\nin neural rendering to perform coherent SAS imaging. Our optimization enables\nus to incorporate physics-based constraints and scene priors into the image\nformation process. We validate our method on simulation and experimental\nresults captured in both air and water. We demonstrate both quantitatively and\nqualitatively that our method typically produces superior reconstructions than\nexisting approaches. We share code and data for reproducibility.\n", "date": "2023-06-16", "categories": []}
{"id": "2306.06408", "title": "Fast light-field 3D microscopy with out-of-distribution detection and\n  adaptation through Conditional Normalizing Flows", "abstract": "  Real-time 3D fluorescence microscopy is crucial for the spatiotemporal\nanalysis of live organisms, such as neural activity monitoring. The eXtended\nfield-of-view light field microscope (XLFM), also known as Fourier light field\nmicroscope, is a straightforward, single snapshot solution to achieve this. The\nXLFM acquires spatial-angular information in a single camera exposure. In a\nsubsequent step, a 3D volume can be algorithmically reconstructed, making it\nexceptionally well-suited for real-time 3D acquisition and potential analysis.\nUnfortunately, traditional reconstruction methods (like deconvolution) require\nlengthy processing times (0.0220 Hz), hampering the speed advantages of the\nXLFM. Neural network architectures can overcome the speed constraints at the\nexpense of lacking certainty metrics, which renders them untrustworthy for the\nbiomedical realm. This work proposes a novel architecture to perform fast 3D\nreconstructions of live immobilized zebrafish neural activity based on a\nconditional normalizing flow. It reconstructs volumes at 8 Hz spanning\n512x512x96 voxels, and it can be trained in under two hours due to the small\ndataset requirements (10 image-volume pairs). Furthermore, normalizing flows\nallow for exact Likelihood computation, enabling distribution monitoring,\nfollowed by out-of-distribution detection and retraining of the system when a\nnovel sample is detected. We evaluate the proposed method on a cross-validation\napproach involving multiple in-distribution samples (genetically identical\nzebrafish) and various out-of-distribution ones.\n", "date": "2023-06-10", "categories": []}
{"id": "2306.08344", "title": "UIERL: Internal-External Representation Learning Network for Underwater\n  Image Enhancement", "abstract": "  Underwater image enhancement (UIE) is a meaningful but challenging task, and\nmany learning-based UIE methods have been proposed in recent years. Although\nmuch progress has been made, these methods still exist two issues: (1) There\nexists a significant region-wise quality difference in a single underwater\nimage due to the underwater imaging process, especially in regions with\ndifferent scene depths. However, existing methods neglect this internal\ncharacteristic of underwater images, resulting in inferior performance; (2) Due\nto the uniqueness of the acquisition approach, underwater image acquisition\ntools usually capture multiple images in the same or similar scenes. Thus, the\nunderwater images to be enhanced in practical usage are highly correlated.\nHowever, when processing a single image, existing methods do not consider the\nrich external information provided by the related images. There is still room\nfor improvement in their performance. Motivated by these two aspects, we\npropose a novel internal-external representation learning (UIERL) network to\nbetter perform UIE tasks with internal and external information,\nsimultaneously. In the internal representation learning stage, a new\ndepth-based region feature guidance network is designed, including a region\nsegmentation based on scene depth to sense regions with different quality\nlevels, followed by a region-wise space encoder module. With performing\nregion-wise feature learning for regions with different quality separately, the\nnetwork provides an effective guidance for global features and thus guides\nintra-image differentiated enhancement. In the external representation learning\nstage, we first propose an external information extraction network to mine the\nrich external information in the related images. Then, internal and external\nfeatures interact with each other via the proposed external-assist-internal\nmodule and internal-assist-e\n", "date": "2023-06-14", "categories": []}
{"id": "2305.09996", "title": "Restoring Images Captured in Arbitrary Hybrid Adverse Weather Conditions\n  in One Go", "abstract": "  Adverse conditions typically suffer from stochastic hybrid weather\ndegradations (e.g., rainy and hazy night), while existing image restoration\nalgorithms envisage that weather degradations occur independently, thus may\nfail to handle real-world complicated scenarios. Besides, supervised training\nis not feasible due to the lack of a comprehensive paired dataset to\ncharacterize hybrid conditions. To this end, we have advanced the\naforementioned limitations with two tactics: framework and data. First, we\npresent a novel unified framework, dubbed RAHC, to Restore Arbitrary Hybrid\nadverse weather Conditions in one go. Specifically, our RAHC leverages a\nmulti-head aggregation architecture to learn multiple degradation\nrepresentation subspaces and then constrains the network to flexibly handle\nmultiple hybrid adverse weather in a unified paradigm through a discrimination\nmechanism in the output space. Furthermore, we devise a reconstruction vectors\naided scheme to provide auxiliary visual content cues for reconstruction, thus\ncan comfortably cope with hybrid scenarios with insufficient remaining image\nconstituents. Second, we construct a new dataset, termed HAC, for learning and\nbenchmarking arbitrary Hybrid Adverse Conditions restoration. HAC contains 31\nscenarios composed of an arbitrary combination of five common weather, with a\ntotal of ~316K adverse-weather/clean pairs. Extensive experiments yield\nsuperior results and establish new state-of-the-art results on both HAC and\nconventional datasets.\n", "date": "2023-05-17", "categories": []}
{"id": "2306.07233", "title": "Generative Plug and Play: Posterior Sampling for Inverse Problems", "abstract": "  Over the past decade, Plug-and-Play (PnP) has become a popular method for\nreconstructing images using a modular framework consisting of a forward and\nprior model. The great strength of PnP is that an image denoiser can be used as\na prior model while the forward model can be implemented using more traditional\nphysics-based approaches. However, a limitation of PnP is that it reconstructs\nonly a single deterministic image.\n  In this paper, we introduce Generative Plug-and-Play (GPnP), a generalization\nof PnP to sample from the posterior distribution. As with PnP, GPnP has a\nmodular framework using a physics-based forward model and an image denoising\nprior model. However, in GPnP these models are extended to become proximal\ngenerators, which sample from associated distributions. GPnP applies these\nproximal generators in alternation to produce samples from the posterior. We\npresent experimental simulations using the well-known BM3D denoiser. Our\nresults demonstrate that the GPnP method is robust, easy to implement, and\nproduces intuitively reasonable samples from the posterior for sparse\ninterpolation and tomographic reconstruction. Code to accompany this paper is\navailable at https://github.com/gbuzzard/generative-pnp-allerton .\n", "date": "2023-06-12", "categories": []}
{"id": "2207.02985", "title": "Orthogonal Matrix Retrieval with Spatial Consensus for 3D Unknown-View\n  Tomography", "abstract": "  Unknown-view tomography (UVT) reconstructs a 3D density map from its 2D\nprojections at unknown, random orientations. A line of work starting with Kam\n(1980) employs the method of moments (MoM) with rotation-invariant Fourier\nfeatures to solve UVT in the frequency domain, assuming that the orientations\nare uniformly distributed. This line of work includes the recent orthogonal\nmatrix retrieval (OMR) approaches based on matrix factorization, which, while\nelegant, either require side information about the density that is not\navailable, or fail to be sufficiently robust. For OMR to break free from those\nrestrictions, we propose to jointly recover the density map and the orthogonal\nmatrices by requiring that they be mutually consistent. We regularize the\nresulting non-convex optimization problem by a denoised reference projection\nand a nonnegativity constraint. This is enabled by the new closed-form\nexpressions for spatial autocorrelation features. Further, we design an\neasy-to-compute initial density map which effectively mitigates the\nnon-convexity of the reconstruction problem. Experimental results show that the\nproposed OMR with spatial consensus is more robust and performs significantly\nbetter than the previous state-of-the-art OMR approach in the typical low-SNR\nscenario of 3D UVT.\n", "date": "2022-07-06", "categories": []}
{"id": "2306.06748", "title": "Moving beyond simulation: data-driven quantitative photoacoustic imaging\n  using tissue-mimicking phantoms", "abstract": "  Accurate measurement of optical absorption coefficients from photoacoustic\nimaging (PAI) data would enable direct mapping of molecular concentrations,\nproviding vital clinical insight. The ill-posed nature of the problem of\nabsorption coefficient recovery has prohibited PAI from achieving this goal in\nliving systems due to the domain gap between simulation and experiment. To\nbridge this gap, we introduce a collection of experimentally well-characterised\nimaging phantoms and their digital twins. This first-of-a-kind phantom data set\nenables supervised training of a U-Net on experimental data for pixel-wise\nestimation of absorption coefficients. We show that training on simulated data\nresults in artefacts and biases in the estimates, reinforcing the existence of\na domain gap between simulation and experiment. Training on experimentally\nacquired data, however, yielded more accurate and robust estimates of optical\nabsorption coefficients. We compare the results to fluence correction with a\nMonte Carlo model from reference optical properties of the materials, which\nyields a quantification error of approximately 20%. Application of the trained\nU-Nets to a blood flow phantom demonstrated spectral biases when training on\nsimulated data, while application to a mouse model highlighted the ability of\nboth learning-based approaches to recover the depth-dependent loss of signal\nintensity. We demonstrate that training on experimental phantoms can restore\nthe correlation of signal amplitudes measured in depth. While the absolute\nquantification error remains high and further improvements are needed, our\nresults highlight the promise of deep learning to advance quantitative PAI.\n", "date": "2023-06-11", "categories": []}
{"id": "2306.05675", "title": "Illumination Controllable Dehazing Network based on Unsupervised Retinex\n  Embedding", "abstract": "  On the one hand, the dehazing task is an illposedness problem, which means\nthat no unique solution exists. On the other hand, the dehazing task should\ntake into account the subjective factor, which is to give the user selectable\ndehazed images rather than a single result. Therefore, this paper proposes a\nmulti-output dehazing network by introducing illumination controllable ability,\ncalled IC-Dehazing. The proposed IC-Dehazing can change the illumination\nintensity by adjusting the factor of the illumination controllable module,\nwhich is realized based on the interpretable Retinex theory. Moreover, the\nbackbone dehazing network of IC-Dehazing consists of a Transformer with double\ndecoders for high-quality image restoration. Further, the prior-based loss\nfunction and unsupervised training strategy enable IC-Dehazing to complete the\nparameter learning process without the need for paired data. To demonstrate the\neffectiveness of the proposed IC-Dehazing, quantitative and qualitative\nexperiments are conducted on image dehazing, semantic segmentation, and object\ndetection tasks. Code is available at\nhttps://github.com/Xiaofeng-life/ICDehazing.\n", "date": "2023-06-09", "categories": []}
{"id": "2306.06089", "title": "Computational Flash Photography through Intrinsics", "abstract": "  Flash is an essential tool as it often serves as the sole controllable light\nsource in everyday photography. However, the use of flash is a binary decision\nat the time a photograph is captured with limited control over its\ncharacteristics such as strength or color. In this work, we study the\ncomputational control of the flash light in photographs taken with or without\nflash. We present a physically motivated intrinsic formulation for flash\nphotograph formation and develop flash decomposition and generation methods for\nflash and no-flash photographs, respectively. We demonstrate that our intrinsic\nformulation outperforms alternatives in the literature and allows us to\ncomputationally control flash in in-the-wild images.\n", "date": "2023-06-09", "categories": []}
{"id": "2306.05390", "title": "HQ-50K: A Large-scale, High-quality Dataset for Image Restoration", "abstract": "  This paper introduces a new large-scale image restoration dataset, called\nHQ-50K, which contains 50,000 high-quality images with rich texture details and\nsemantic diversity. We analyze existing image restoration datasets from five\ndifferent perspectives, including data scale, resolution, compression rates,\ntexture details, and semantic coverage. However, we find that all of these\ndatasets are deficient in some aspects. In contrast, HQ-50K considers all of\nthese five aspects during the data curation process and meets all requirements.\nWe also present a new Degradation-Aware Mixture of Expert (DAMoE) model, which\nenables a single model to handle multiple corruption types and unknown levels.\nOur extensive experiments demonstrate that HQ-50K consistently improves the\nperformance on various image restoration tasks, such as super-resolution,\ndenoising, dejpeg, and deraining. Furthermore, our proposed DAMoE, trained on\nour \\dataset, outperforms existing state-of-the-art unified models designed for\nmultiple restoration tasks and levels. The dataset and code are available at\n\\url{https://github.com/littleYaang/HQ-50K}.\n", "date": "2023-06-08", "categories": []}
{"id": "2306.02644", "title": "Learned Alternating Minimization Algorithm for Dual-domain Sparse-View\n  CT Reconstruction", "abstract": "  We propose a novel Learned Alternating Minimization Algorithm (LAMA) for\ndual-domain sparse-view CT image reconstruction. LAMA is naturally induced by a\nvariational model for CT reconstruction with learnable nonsmooth nonconvex\nregularizers, which are parameterized as composite functions of deep networks\nin both image and sinogram domains. To minimize the objective of the model, we\nincorporate the smoothing technique and residual learning architecture into the\ndesign of LAMA. We show that LAMA substantially reduces network complexity,\nimproves memory efficiency and reconstruction accuracy, and is provably\nconvergent for reliable reconstructions. Extensive numerical experiments\ndemonstrate that LAMA outperforms existing methods by a wide margin on multiple\nbenchmark CT datasets.\n", "date": "2023-06-05", "categories": []}
{"id": "2306.03466", "title": "Convergent Bregman Plug-and-Play Image Restoration for Poisson Inverse\n  Problems", "abstract": "  Plug-and-Play (PnP) methods are efficient iterative algorithms for solving\nill-posed image inverse problems. PnP methods are obtained by using deep\nGaussian denoisers instead of the proximal operator or the gradient-descent\nstep within proximal algorithms. Current PnP schemes rely on data-fidelity\nterms that have either Lipschitz gradients or closed-form proximal operators,\nwhich is not applicable to Poisson inverse problems. Based on the observation\nthat the Gaussian noise is not the adequate noise model in this setting, we\npropose to generalize PnP using theBregman Proximal Gradient (BPG) method. BPG\nreplaces the Euclidean distance with a Bregman divergence that can better\ncapture the smoothness properties of the problem. We introduce the Bregman\nScore Denoiser specifically parametrized and trained for the new Bregman\ngeometry and prove that it corresponds to the proximal operator of a nonconvex\npotential. We propose two PnP algorithms based on the Bregman Score Denoiser\nfor solving Poisson inverse problems. Extending the convergence results of BPG\nin the nonconvex settings, we show that the proposed methods converge,\ntargeting stationary points of an explicit global functional. Experimental\nevaluations conducted on various Poisson inverse problems validate the\nconvergence results and showcase effective restoration performance.\n", "date": "2023-06-06", "categories": []}
{"id": "2302.10279", "title": "Image Reconstruction via Deep Image Prior Subspaces", "abstract": "  Deep learning has been widely used for solving image reconstruction tasks but\nits deployability has been held back due to the shortage of high-quality\ntraining data. Unsupervised learning methods, such as the deep image prior\n(DIP), naturally fill this gap, but bring a host of new issues: the\nsusceptibility to overfitting due to a lack of robust early stopping strategies\nand unstable convergence. We present a novel approach to tackle these issues by\nrestricting DIP optimisation to a sparse linear subspace of its parameters,\nemploying a synergy of dimensionality reduction techniques and second order\noptimisation methods. The low-dimensionality of the subspace reduces DIP's\ntendency to fit noise and allows the use of stable second order optimisation\nmethods, e.g., natural gradient descent or L-BFGS. Experiments across both\nimage restoration and tomographic tasks of different geometry and ill-posedness\nshow that second order optimisation within a low-dimensional subspace is\nfavourable in terms of optimisation stability to reconstruction fidelity\ntrade-off.\n", "date": "2023-02-20", "categories": []}
{"id": "2306.02634", "title": "Computational 3D topographic microscopy from terabytes of data per\n  sample", "abstract": "  We present a large-scale computational 3D topographic microscope that enables\n6-gigapixel profilometric 3D imaging at micron-scale resolution across $>$110\ncm$^2$ areas over multi-millimeter axial ranges. Our computational microscope,\ntermed STARCAM (Scanning Topographic All-in-focus Reconstruction with a\nComputational Array Microscope), features a parallelized, 54-camera\narchitecture with 3-axis translation to capture, for each sample of interest, a\nmulti-dimensional, 2.1-terabyte (TB) dataset, consisting of a total of 224,640\n9.4-megapixel images. We developed a self-supervised neural network-based\nalgorithm for 3D reconstruction and stitching that jointly estimates an\nall-in-focus photometric composite and 3D height map across the entire field of\nview, using multi-view stereo information and image sharpness as a focal\nmetric. The memory-efficient, compressed differentiable representation offered\nby the neural network effectively enables joint participation of the entire\nmulti-TB dataset during the reconstruction process. To demonstrate the broad\nutility of our new computational microscope, we applied STARCAM to a variety of\ndecimeter-scale objects, with applications ranging from cultural heritage to\nindustrial inspection.\n", "date": "2023-06-05", "categories": []}
{"id": "2306.02673", "title": "Cross-Modal Vertical Federated Learning for MRI Reconstruction", "abstract": "  Federated learning enables multiple hospitals to cooperatively learn a shared\nmodel without privacy disclosure. Existing methods often take a common\nassumption that the data from different hospitals have the same modalities.\nHowever, such a setting is difficult to fully satisfy in practical\napplications, since the imaging guidelines may be different between hospitals,\nwhich makes the number of individuals with the same set of modalities limited.\nTo this end, we formulate this practical-yet-challenging cross-modal vertical\nfederated learning task, in which shape data from multiple hospitals have\ndifferent modalities with a small amount of multi-modality data collected from\nthe same individuals. To tackle such a situation, we develop a novel framework,\nnamely Federated Consistent Regularization constrained Feature Disentanglement\n(Fed-CRFD), for boosting MRI reconstruction by effectively exploring the\noverlapping samples (individuals with multi-modalities) and solving the domain\nshift problem caused by different modalities. Particularly, our Fed-CRFD\ninvolves an intra-client feature disentangle scheme to decouple data into\nmodality-invariant and modality-specific features, where the modality-invariant\nfeatures are leveraged to mitigate the domain shift problem. In addition, a\ncross-client latent representation consistency constraint is proposed\nspecifically for the overlapping samples to further align the\nmodality-invariant features extracted from different modalities. Hence, our\nmethod can fully exploit the multi-source data from hospitals while alleviating\nthe domain shift problem. Extensive experiments on two typical MRI datasets\ndemonstrate that our network clearly outperforms state-of-the-art MRI\nreconstruction methods. The source code will be publicly released upon the\npublication of this work.\n", "date": "2023-06-05", "categories": []}
{"id": "2209.13300", "title": "Passive Non-line-of-sight Imaging for Moving Targets with an Event\n  Camera", "abstract": "  Non-line-of-sight (NLOS) imaging is an emerging technique for detecting\nobjects behind obstacles or around corners. Recent studies on passive NLOS\nmainly focus on steady-state measurement and reconstruction methods, which show\nlimitations in recognition of moving targets. To the best of our knowledge, we\npropose a novel event-based passive NLOS imaging method. We acquire\nasynchronous event-based data which contains detailed dynamic information of\nthe NLOS target, and efficiently ease the degradation of speckle caused by\nmovement. Besides, we create the first event-based NLOS imaging dataset,\nNLOS-ES, and the event-based feature is extracted by time-surface\nrepresentation. We compare the reconstructions through event-based data with\nframe-based data. The event-based method performs well on PSNR and LPIPS, which\nis 20% and 10% better than frame-based method, while the data volume takes only\n2% of traditional method.\n", "date": "2022-09-27", "categories": []}
{"id": "2306.01630", "title": "A Conditional Normalizing Flow for Accelerated Multi-Coil MR Imaging", "abstract": "  Accelerated magnetic resonance (MR) imaging attempts to reduce acquisition\ntime by collecting data below the Nyquist rate. As an ill-posed inverse\nproblem, many plausible solutions exist, yet the majority of deep learning\napproaches generate only a single solution. We instead focus on sampling from\nthe posterior distribution, which provides more comprehensive information for\ndownstream inference tasks. To do this, we design a novel conditional\nnormalizing flow (CNF) that infers the signal component in the measurement\noperator's nullspace, which is later combined with measured data to form\ncomplete images. Using fastMRI brain and knee data, we demonstrate fast\ninference and accuracy that surpasses recent posterior sampling techniques for\nMRI. Code is available at https://github.com/jwen307/mri_cnf/\n", "date": "2023-06-02", "categories": []}
{"id": "2303.14218", "title": "Curricular Contrastive Regularization for Physics-aware Single Image\n  Dehazing", "abstract": "  Considering the ill-posed nature, contrastive regularization has been\ndeveloped for single image dehazing, introducing the information from negative\nimages as a lower bound. However, the contrastive samples are nonconsensual, as\nthe negatives are usually represented distantly from the clear (i.e., positive)\nimage, leaving the solution space still under-constricted. Moreover, the\ninterpretability of deep dehazing models is underexplored towards the physics\nof the hazing process. In this paper, we propose a novel curricular contrastive\nregularization targeted at a consensual contrastive space as opposed to a\nnon-consensual one. Our negatives, which provide better lower-bound\nconstraints, can be assembled from 1) the hazy image, and 2) corresponding\nrestorations by other existing methods. Further, due to the different\nsimilarities between the embeddings of the clear image and negatives, the\nlearning difficulty of the multiple components is intrinsically imbalanced. To\ntackle this issue, we customize a curriculum learning strategy to reweight the\nimportance of different negatives. In addition, to improve the interpretability\nin the feature space, we build a physics-aware dual-branch unit according to\nthe atmospheric scattering model. With the unit, as well as curricular\ncontrastive regularization, we establish our dehazing network, named C2PNet.\nExtensive experiments demonstrate that our C2PNet significantly outperforms\nstate-of-the-art methods, with extreme PSNR boosts of 3.94dB and 1.50dB,\nrespectively, on SOTS-indoor and SOTS-outdoor datasets.\n", "date": "2023-03-24", "categories": []}
{"id": "2304.03895", "title": "MCDIP-ADMM: Overcoming Overfitting in DIP-based CT reconstruction", "abstract": "  This paper investigates the application of unsupervised learning methods for\ncomputed tomography (CT) reconstruction. To motivate our work, we review\nseveral existing priors, namely the truncated Gaussian prior, the $l_1$ prior,\nthe total variation prior, and the deep image prior (DIP). We find that DIP\noutperforms the other three priors in terms of representational capability and\nvisual performance. However, the performance of DIP deteriorates when the\nnumber of iterations exceeds a certain threshold due to overfitting. To address\nthis issue, we propose a novel method (MCDIP-ADMM) based on Multi-Code Deep\nImage Prior and plug-and-play Alternative Direction Method of Multipliers.\nSpecifically, MCDIP utilizes multiple latent codes to generate a series of\nfeature maps at an intermediate layer within a generator model. These maps are\nthen composed with trainable weights, representing the complete image prior.\nExperimental results demonstrate the superior performance of the proposed\nMCDIP-ADMM compared to three existing competitors. In the case of parallel beam\nprojection with Gaussian noise, MCDIP-ADMM achieves an average improvement of\n4.3 dB over DIP, 1.7 dB over ADMM DIP-WTV, and 1.2 dB over PnP-DIP in terms of\nPSNR. Similarly, for fan-beam projection with Poisson noise, MCDIP-ADMM\nachieves an average improvement of 3.09 dB over DIP, 1.86 dB over ADMM DIP-WTV,\nand 0.84 dB over PnP-DIP in terms of PSNR.\n", "date": "2023-04-08", "categories": []}
{"id": "2306.00386", "title": "Symmetric Uncertainty-Aware Feature Transmission for Depth\n  Super-Resolution", "abstract": "  Color-guided depth super-resolution (DSR) is an encouraging paradigm that\nenhances a low-resolution (LR) depth map guided by an extra high-resolution\n(HR) RGB image from the same scene. Existing methods usually use interpolation\nto upscale the depth maps before feeding them into the network and transfer the\nhigh-frequency information extracted from HR RGB images to guide the\nreconstruction of depth maps. However, the extracted high-frequency information\nusually contains textures that are not present in depth maps in the existence\nof the cross-modality gap, and the noises would be further aggravated by\ninterpolation due to the resolution gap between the RGB and depth images. To\ntackle these challenges, we propose a novel Symmetric Uncertainty-aware Feature\nTransmission (SUFT) for color-guided DSR. (1) For the resolution gap, SUFT\nbuilds an iterative up-and-down sampling pipeline, which makes depth features\nand RGB features spatially consistent while suppressing noise amplification and\nblurring by replacing common interpolated pre-upsampling. (2) For the\ncross-modality gap, we propose a novel Symmetric Uncertainty scheme to remove\nparts of RGB information harmful to the recovery of HR depth maps. Extensive\nexperiments on benchmark datasets and challenging real-world settings suggest\nthat our method achieves superior performance compared to state-of-the-art\nmethods. Our code and models are available at\nhttps://github.com/ShiWuxuan/SUFT.\n", "date": "2023-06-01", "categories": []}
{"id": "2306.00676", "title": "Hyperspectral Target Detection Based on Low-Rank Background Subspace\n  Learning and Graph Laplacian Regularization", "abstract": "  Hyperspectral target detection is good at finding dim and small objects based\non spectral characteristics. However, existing representation-based methods are\nhindered by the problem of the unknown background dictionary and insufficient\nutilization of spatial information. To address these issues, this paper\nproposes an efficient optimizing approach based on low-rank representation\n(LRR) and graph Laplacian regularization (GLR). Firstly, to obtain a complete\nand pure background dictionary, we propose a LRR-based background subspace\nlearning method by jointly mining the low-dimensional structure of all pixels.\nSecondly, to fully exploit local spatial relationships and capture the\nunderlying geometric structure, a local region-based GLR is employed to\nestimate the coefficients. Finally, the desired detection map is generated by\ncomputing the ratio of representation errors from binary hypothesis testing.\nThe experiments conducted on two benchmark datasets validate the effectiveness\nand superiority of the approach. For reproduction, the accompanying code is\navailable at https://github.com/shendb2022/LRBSL-GLR.\n", "date": "2023-06-01", "categories": []}
{"id": "2306.00834", "title": "Deformable Convolutions and LSTM-based Flexible Event Frame Fusion\n  Network for Motion Deblurring", "abstract": "  Event cameras differ from conventional RGB cameras in that they produce\nasynchronous data sequences. While RGB cameras capture every frame at a fixed\nrate, event cameras only capture changes in the scene, resulting in sparse and\nasynchronous data output. Despite the fact that event data carries useful\ninformation that can be utilized in motion deblurring of RGB cameras,\nintegrating event and image information remains a challenge. Recent\nstate-of-the-art CNN-based deblurring solutions produce multiple 2-D event\nframes based on the accumulation of event data over a time period. In most of\nthese techniques, however, the number of event frames is fixed and predefined,\nwhich reduces temporal resolution drastically, particularly for scenarios when\nfast-moving objects are present or when longer exposure times are required. It\nis also important to note that recent modern cameras (e.g., cameras in mobile\nphones) dynamically set the exposure time of the image, which presents an\nadditional problem for networks developed for a fixed number of event frames. A\nLong Short-Term Memory (LSTM)-based event feature extraction module has been\ndeveloped for addressing these challenges, which enables us to use a\ndynamically varying number of event frames. Using these modules, we constructed\na state-of-the-art deblurring network, Deformable Convolutions and LSTM-based\nFlexible Event Frame Fusion Network (DLEFNet). It is particularly useful for\nscenarios in which exposure times vary depending on factors such as lighting\nconditions or the presence of fast-moving objects in the scene. It has been\ndemonstrated through evaluation results that the proposed method can outperform\nthe existing state-of-the-art networks for deblurring task in synthetic and\nreal-world data sets.\n", "date": "2023-06-01", "categories": []}
{"id": "2207.14709", "title": "Robust Quantitative Susceptibility Mapping via Approximate Message\n  Passing with Parameter Estimation", "abstract": "  Purpose: For quantitative susceptibility mapping (QSM), the lack of\nground-truth in clinical settings makes it challenging to determine suitable\nparameters for the dipole inversion. We propose a probabilistic Bayesian\napproach for QSM with built-in parameter estimation, and incorporate the\nnonlinear formulation of the dipole inversion to achieve a robust recovery of\nthe susceptibility maps.\n  Theory: From a Bayesian perspective, the image wavelet coefficients are\napproximately sparse and modelled by the Laplace distribution. The measurement\nnoise is modelled by a Gaussian-mixture distribution with two components, where\nthe second component is used to model the noise outliers. Through probabilistic\ninference, the susceptibility map and distribution parameters can be jointly\nrecovered using approximate message passing (AMP).\n  Methods: We compare our proposed AMP with built-in parameter estimation\n(AMP-PE) to the state-of-the-art L1-QSM, FANSI and MEDI approaches on the\nsimulated and in vivo datasets, and perform experiments to explore the optimal\nsettings of AMP-PE. Reproducible code is available at\nhttps://github.com/EmoryCN2L/QSM_AMP_PE\n  Results: On the simulated Sim2Snr1 dataset, AMP-PE achieved the lowest NRMSE,\nDFCM and the highest SSIM, while MEDI achieved the lowest HFEN. On the in vivo\ndatasets, AMP-PE is robust and successfully recovers the susceptibility maps\nusing the estimated parameters, whereas L1-QSM, FANSI and MEDI typically\nrequire additional visual fine-tuning to select or double-check working\nparameters.\n  Conclusion: AMP-PE provides automatic and adaptive parameter estimation for\nQSM and avoids the subjectivity from the visual fine-tuning step, making it an\nexcellent choice for the clinical setting.\n", "date": "2022-07-29", "categories": []}
{"id": "2210.05553", "title": "Evaluating Unsupervised Denoising Requires Unsupervised Metrics", "abstract": "  Unsupervised denoising is a crucial challenge in real-world imaging\napplications. Unsupervised deep-learning methods have demonstrated impressive\nperformance on benchmarks based on synthetic noise. However, no metrics are\navailable to evaluate these methods in an unsupervised fashion. This is highly\nproblematic for the many practical applications where ground-truth clean images\nare not available. In this work, we propose two novel metrics: the unsupervised\nmean squared error (MSE) and the unsupervised peak signal-to-noise ratio\n(PSNR), which are computed using only noisy data. We provide a theoretical\nanalysis of these metrics, showing that they are asymptotically consistent\nestimators of the supervised MSE and PSNR. Controlled numerical experiments\nwith synthetic noise confirm that they provide accurate approximations in\npractice. We validate our approach on real-world data from two imaging\nmodalities: videos in raw format and transmission electron microscopy. Our\nresults demonstrate that the proposed metrics enable unsupervised evaluation of\ndenoising methods based exclusively on noisy data.\n", "date": "2022-10-11", "categories": []}
{"id": "2301.11699", "title": "Image Restoration with Mean-Reverting Stochastic Differential Equations", "abstract": "  This paper presents a stochastic differential equation (SDE) approach for\ngeneral-purpose image restoration. The key construction consists in a\nmean-reverting SDE that transforms a high-quality image into a degraded\ncounterpart as a mean state with fixed Gaussian noise. Then, by simulating the\ncorresponding reverse-time SDE, we are able to restore the origin of the\nlow-quality image without relying on any task-specific prior knowledge.\nCrucially, the proposed mean-reverting SDE has a closed-form solution, allowing\nus to compute the ground truth time-dependent score and learn it with a neural\nnetwork. Moreover, we propose a maximum likelihood objective to learn an\noptimal reverse trajectory that stabilizes the training and improves the\nrestoration results. The experiments show that our proposed method achieves\nhighly competitive performance in quantitative comparisons on image deraining,\ndeblurring, and denoising, setting a new state-of-the-art on two deraining\ndatasets. Finally, the general applicability of our approach is further\ndemonstrated via qualitative results on image super-resolution, inpainting, and\ndehazing. Code is available at\nhttps://github.com/Algolzw/image-restoration-sde.\n", "date": "2023-01-27", "categories": []}
{"id": "2305.17916", "title": "Volume Feature Rendering for Fast Neural Radiance Field Reconstruction", "abstract": "  Neural radiance fields (NeRFs) are able to synthesize realistic novel views\nfrom multi-view images captured from distinct positions and perspectives. In\nNeRF's rendering pipeline, neural networks are used to represent a scene\nindependently or transform queried learnable feature vector of a point to the\nexpected color or density. With the aid of geometry guides either in occupancy\ngrids or proposal networks, the number of neural network evaluations can be\nreduced from hundreds to dozens in the standard volume rendering framework.\nInstead of rendering yielded color after neural network evaluation, we propose\nto render the queried feature vectors of a ray first and then transform the\nrendered feature vector to the final pixel color by a neural network. This\nfundamental change to the standard volume rendering framework requires only one\nsingle neural network evaluation to render a pixel, which substantially lowers\nthe high computational complexity of the rendering framework attributed to a\nlarge number of neural network evaluations. Consequently, we can use a\ncomparably larger neural network to achieve a better rendering quality while\nmaintaining the same training and rendering time costs. Our model achieves the\nstate-of-the-art rendering quality on both synthetic and real-world datasets\nwhile requiring a training time of several minutes.\n", "date": "2023-05-29", "categories": []}
{"id": "2305.19538", "title": "Automatic Illumination Spectrum Recovery", "abstract": "  We develop a deep learning network to estimate the illumination spectrum of\nhyperspectral images under various lighting conditions. To this end, a dataset,\nIllumNet, was created. Images were captured using a Specim IQ camera under\nvarious illumination conditions, both indoor and outdoor. Outdoor images were\ncaptured in sunny, overcast, and shady conditions and at different times of the\nday. For indoor images, halogen and LED light sources were used, as well as\nmixed light sources, mainly halogen or LED and fluorescent. The ResNet18\nnetwork was employed in this study, but with the 2D kernel changed to a 3D\nkernel to suit the spectral nature of the data. As well as fitting the actual\nillumination spectrum well, the predicted illumination spectrum should also be\nsmooth, and this is achieved by the cubic smoothing spline error cost function.\nExperimental results indicate that the trained model can infer an accurate\nestimate of the illumination spectrum.\n", "date": "2023-05-31", "categories": []}
{"id": "2305.19743", "title": "Towards Monocular Shape from Refraction", "abstract": "  Refraction is a common physical phenomenon and has long been researched in\ncomputer vision. Objects imaged through a refractive object appear distorted in\nthe image as a function of the shape of the interface between the media. This\nhinders many computer vision applications, but can be utilized for obtaining\nthe geometry of the refractive interface. Previous approaches for refractive\nsurface recovery largely relied on various priors or additional information\nlike multiple images of the analyzed surface. In contrast, we claim that a\nsimple energy function based on Snell's law enables the reconstruction of an\narbitrary refractive surface geometry using just a single image and known\nbackground texture and geometry. In the case of a single point, Snell's law has\ntwo degrees of freedom, therefore to estimate a surface depth, we need\nadditional information. We show that solving for an entire surface at once\nintroduces implicit parameter-free spatial regularization and yields convincing\nresults when an intelligent initial guess is provided. We demonstrate our\napproach through simulations and real-world experiments, where the\nreconstruction shows encouraging results in the single-frame monocular setting.\n", "date": "2023-05-31", "categories": []}
{"id": "2305.19774", "title": "Ambiguity in solving imaging inverse problems with deep learning based\n  operators", "abstract": "  In recent years, large convolutional neural networks have been widely used as\ntools for image deblurring, because of their ability in restoring images very\nprecisely. It is well known that image deblurring is mathematically modeled as\nan ill-posed inverse problem and its solution is difficult to approximate when\nnoise affects the data. Really, one limitation of neural networks for\ndeblurring is their sensitivity to noise and other perturbations, which can\nlead to instability and produce poor reconstructions. In addition, networks do\nnot necessarily take into account the numerical formulation of the underlying\nimaging problem, when trained end-to-end. In this paper, we propose some\nstrategies to improve stability without losing to much accuracy to deblur\nimages with deep-learning based methods. First, we suggest a very small neural\narchitecture, which reduces the execution time for training, satisfying a green\nAI need, and does not extremely amplify noise in the computed image. Second, we\nintroduce a unified framework where a pre-processing step balances the lack of\nstability of the following, neural network-based, step. Two different\npre-processors are presented: the former implements a strong parameter-free\ndenoiser, and the latter is a variational model-based regularized formulation\nof the latent imaging problem. This framework is also formally characterized by\nmathematical analysis. Numerical experiments are performed to verify the\naccuracy and stability of the proposed approaches for image deblurring when\nunknown or not-quantified noise is present; the results confirm that they\nimprove the network stability with respect to noise. In particular, the\nmodel-based framework represents the most reliable trade-off between visual\nprecision and robustness.\n", "date": "2023-05-31", "categories": []}
{"id": "2305.20006", "title": "Physics-Informed Ensemble Representation for Light-Field Image\n  Super-Resolution", "abstract": "  Recent learning-based approaches have achieved significant progress in light\nfield (LF) image super-resolution (SR) by exploring convolution-based or\ntransformer-based network structures. However, LF imaging has many intrinsic\nphysical priors that have not been fully exploited. In this paper, we analyze\nthe coordinate transformation of the LF imaging process to reveal the geometric\nrelationship in the LF images. Based on such geometric priors, we introduce a\nnew LF subspace of virtual-slit images (VSI) that provide sub-pixel information\ncomplementary to sub-aperture images. To leverage the abundant correlation\nacross the four-dimensional data with manageable complexity, we propose\nlearning ensemble representation of all $C_4^2$ LF subspaces for more effective\nfeature extraction. To super-resolve image structures from undersampled LF\ndata, we propose a geometry-aware decoder, named EPIXformer, which constrains\nthe transformer's operational searching regions with a LF physical prior.\nExperimental results on both spatial and angular SR tasks demonstrate that the\nproposed method outperforms other state-of-the-art schemes, especially in\nhandling various disparities.\n", "date": "2023-05-31", "categories": []}
{"id": "2302.06381", "title": "Self-supervised phase unwrapping in fringe projection profilometry", "abstract": "  Fast-speed and high-accuracy three-dimensional (3D) shape measurement has\nbeen the goal all along in fringe projection profilometry (FPP). The\ndual-frequency temporal phase unwrapping method (DF-TPU) is one of the\nprominent technologies to achieve this goal. However, the period number of the\nhigh-frequency pattern of existing DF-TPU approaches is usually limited by the\ninevitable phase errors, setting a limit to measurement accuracy.\nDeep-learning-based phase unwrapping methods for single-camera FPP usually\nrequire labeled data for training. In this letter, a novel self-supervised\nphase unwrapping method for single-camera FPP systems is proposed. The trained\nnetwork can retrieve the absolute fringe order from one phase map of 64-period\nand overperform DF-TPU approaches in terms of depth accuracy. Experimental\nresults demonstrate the validation of the proposed method on real scenes of\nmotion blur, isolated objects, low reflectivity, and phase discontinuity.\n", "date": "2023-02-13", "categories": []}
{"id": "2305.18994", "title": "Toward Real-World Light Field Super-Resolution", "abstract": "  Deep learning has opened up new possibilities for light field\nsuper-resolution (SR), but existing methods trained on synthetic datasets with\nsimple degradations (e.g., bicubic downsampling) suffer from poor performance\nwhen applied to complex real-world scenarios. To address this problem, we\nintroduce LytroZoom, the first real-world light field SR dataset capturing\npaired low- and high-resolution light fields of diverse indoor and outdoor\nscenes using a Lytro ILLUM camera. Additionally, we propose the Omni-Frequency\nProjection Network (OFPNet), which decomposes the omni-frequency components and\niteratively enhances them through frequency projection operations to address\nspatially variant degradation processes present in all frequency components.\nExperiments demonstrate that models trained on LytroZoom outperform those\ntrained on synthetic datasets and are generalizable to diverse content and\ndevices. Quantitative and qualitative evaluations verify the superiority of\nOFPNet. We believe this work will inspire future research in real-world light\nfield SR.\n", "date": "2023-05-30", "categories": []}
{"id": "2208.02448", "title": "Deep Progressive Feature Aggregation Network for High Dynamic Range\n  Imaging", "abstract": "  High dynamic range (HDR) imaging is an important task in image processing\nthat aims to generate well-exposed images in scenes with varying illumination.\nAlthough existing multi-exposure fusion methods have achieved impressive\nresults, generating high-quality HDR images in dynamic scenes is still\ndifficult. The primary challenges are ghosting artifacts caused by object\nmotion between low dynamic range images and distorted content in under and\noverexposed regions. In this paper, we propose a deep progressive feature\naggregation network for improving HDR imaging quality in dynamic scenes. To\naddress the issues of object motion, our method implicitly samples\nhigh-correspondence features and aggregates them in a coarse-to-fine manner for\nalignment. In addition, our method adopts a densely connected network structure\nbased on the discrete wavelet transform, which aims to decompose the input\nfeatures into multiple frequency subbands and adaptively restore corrupted\ncontents. Experiments show that our proposed method can achieve\nstate-of-the-art performance under different scenes, compared to other\npromising HDR imaging methods. Specifically, the HDR images generated by our\nmethod contain cleaner and more detailed content, with fewer distortions,\nleading to better visual quality.\n", "date": "2022-08-04", "categories": []}
{"id": "2304.03297", "title": "Neural Operator Learning for Ultrasound Tomography Inversion", "abstract": "  Neural operator learning as a means of mapping between complex function\nspaces has garnered significant attention in the field of computational science\nand engineering (CS&E). In this paper, we apply Neural operator learning to the\ntime-of-flight ultrasound computed tomography (USCT) problem. We learn the\nmapping between time-of-flight (TOF) data and the heterogeneous sound speed\nfield using a full-wave solver to generate the training data. This novel\napplication of operator learning circumnavigates the need to solve the\ncomputationally intensive iterative inverse problem. The operator learns the\nnon-linear mapping offline and predicts the heterogeneous sound field with a\nsingle forward pass through the model. This is the first time operator learning\nhas been used for ultrasound tomography and is the first step in potential\nreal-time predictions of soft tissue distribution for tumor identification in\nbeast imaging.\n", "date": "2023-04-06", "categories": []}
{"id": "2305.17398", "title": "NeRO: Neural Geometry and BRDF Reconstruction of Reflective Objects from\n  Multiview Images", "abstract": "  We present a neural rendering-based method called NeRO for reconstructing the\ngeometry and the BRDF of reflective objects from multiview images captured in\nan unknown environment. Multiview reconstruction of reflective objects is\nextremely challenging because specular reflections are view-dependent and thus\nviolate the multiview consistency, which is the cornerstone for most multiview\nreconstruction methods. Recent neural rendering techniques can model the\ninteraction between environment lights and the object surfaces to fit the\nview-dependent reflections, thus making it possible to reconstruct reflective\nobjects from multiview images. However, accurately modeling environment lights\nin the neural rendering is intractable, especially when the geometry is\nunknown. Most existing neural rendering methods, which can model environment\nlights, only consider direct lights and rely on object masks to reconstruct\nobjects with weak specular reflections. Therefore, these methods fail to\nreconstruct reflective objects, especially when the object mask is not\navailable and the object is illuminated by indirect lights. We propose a\ntwo-step approach to tackle this problem. First, by applying the split-sum\napproximation and the integrated directional encoding to approximate the\nshading effects of both direct and indirect lights, we are able to accurately\nreconstruct the geometry of reflective objects without any object masks. Then,\nwith the object geometry fixed, we use more accurate sampling to recover the\nenvironment lights and the BRDF of the object. Extensive experiments\ndemonstrate that our method is capable of accurately reconstructing the\ngeometry and the BRDF of reflective objects from only posed RGB images without\nknowing the environment lights and the object masks. Codes and datasets are\navailable at https://github.com/liuyuan-pal/NeRO.\n", "date": "2023-05-27", "categories": []}
{"id": "2302.05872", "title": "I$^2$SB: Image-to-Image Schr\\\"odinger Bridge", "abstract": "  We propose Image-to-Image Schr\\\"odinger Bridge (I$^2$SB), a new class of\nconditional diffusion models that directly learn the nonlinear diffusion\nprocesses between two given distributions. These diffusion bridges are\nparticularly useful for image restoration, as the degraded images are\nstructurally informative priors for reconstructing the clean images. I$^2$SB\nbelongs to a tractable class of Schr\\\"odinger bridge, the nonlinear extension\nto score-based models, whose marginal distributions can be computed\nanalytically given boundary pairs. This results in a simulation-free framework\nfor nonlinear diffusions, where the I$^2$SB training becomes scalable by\nadopting practical techniques used in standard diffusion models. We validate\nI$^2$SB in solving various image restoration tasks, including inpainting,\nsuper-resolution, deblurring, and JPEG restoration on ImageNet 256x256 and show\nthat I$^2$SB surpasses standard conditional diffusion models with more\ninterpretable generative processes. Moreover, I$^2$SB matches the performance\nof inverse methods that additionally require the knowledge of the corruption\noperators. Our work opens up new algorithmic opportunities for developing\nefficient nonlinear diffusion models on a large scale. scale. Project page and\ncodes: https://i2sb.github.io/\n", "date": "2023-02-12", "categories": []}
{"id": "2305.16481", "title": "SimHaze: game engine simulated data for real-world dehazing", "abstract": "  Deep models have demonstrated recent success in single-image dehazing. Most\nprior methods consider fully supervised training and learn from paired clean\nand hazy images, where a hazy image is synthesized based on a clean image and\nits estimated depth map. This paradigm, however, can produce low-quality hazy\nimages due to inaccurate depth estimation, resulting in poor generalization of\nthe trained models. In this paper, we explore an alternative approach for\ngenerating paired clean-hazy images by leveraging computer graphics. Using a\nmodern game engine, our approach renders crisp clean images and their precise\ndepth maps, based on which high-quality hazy images can be synthesized for\ntraining dehazing models. To this end, we present SimHaze: a new synthetic haze\ndataset. More importantly, we show that training with SimHaze alone allows the\nlatest dehazing models to achieve significantly better performance in\ncomparison to previous dehazing datasets. Our dataset and code will be made\npublicly available.\n", "date": "2023-05-25", "categories": []}
{"id": "2303.00748", "title": "Efficient and Explicit Modelling of Image Hierarchies for Image\n  Restoration", "abstract": "  The aim of this paper is to propose a mechanism to efficiently and explicitly\nmodel image hierarchies in the global, regional, and local range for image\nrestoration. To achieve that, we start by analyzing two important properties of\nnatural images including cross-scale similarity and anisotropic image features.\nInspired by that, we propose the anchored stripe self-attention which achieves\na good balance between the space and time complexity of self-attention and the\nmodelling capacity beyond the regional range. Then we propose a new network\narchitecture dubbed GRL to explicitly model image hierarchies in the Global,\nRegional, and Local range via anchored stripe self-attention, window\nself-attention, and channel attention enhanced convolution. Finally, the\nproposed network is applied to 7 image restoration types, covering both real\nand synthetic settings. The proposed method sets the new state-of-the-art for\nseveral of those. Code will be available at\nhttps://github.com/ofsoundof/GRL-Image-Restoration.git.\n", "date": "2023-03-01", "categories": []}
{"id": "2305.13593", "title": "Neural Image Re-Exposure", "abstract": "  The shutter strategy applied to the photo-shooting process has a significant\ninfluence on the quality of the captured photograph. An improper shutter may\nlead to a blurry image, video discontinuity, or rolling shutter artifact.\nExisting works try to provide an independent solution for each issue. In this\nwork, we aim to re-expose the captured photo in post-processing to provide a\nmore flexible way of addressing those issues within a unified framework.\nSpecifically, we propose a neural network-based image re-exposure framework. It\nconsists of an encoder for visual latent space construction, a re-exposure\nmodule for aggregating information to neural film with a desired shutter\nstrategy, and a decoder for 'developing' neural film into a desired image. To\ncompensate for information confusion and missing frames, event streams, which\ncan capture almost continuous brightness changes, are leveraged in computing\nvisual latent content. Both self-attention layers and cross-attention layers\nare employed in the re-exposure module to promote interaction between neural\nfilm and visual latent content and information aggregation to neural film. The\nproposed unified image re-exposure framework is evaluated on several\nshutter-related image recovery tasks and performs favorably against independent\nstate-of-the-art methods.\n", "date": "2023-05-23", "categories": []}
{"id": "2305.13770", "title": "MIPI 2023 Challenge on Nighttime Flare Removal: Methods and Results", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for in-depth exchange of views from\nindustry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). With the success of the 1st MIPI Workshop@ECCV\n2022, we introduce the second MIPI challenge including four tracks focusing on\nnovel image sensors and imaging algorithms. In this paper, we summarize and\nreview the Nighttime Flare Removal track on MIPI 2023. In total, 120\nparticipants were successfully registered, and 11 teams submitted results in\nthe final testing phase. The developed solutions in this challenge achieved\nstate-of-the-art performance on Nighttime Flare Removal. A detailed description\nof all models developed in this challenge is provided in this paper. More\ndetails of this challenge and the link to the dataset can be found at\nhttps://mipi-challenge.org/MIPI2023/ .\n", "date": "2023-05-23", "categories": []}
{"id": "2305.13880", "title": "Generalized Expectation Maximization Framework for Blind Image Super\n  Resolution", "abstract": "  Learning-based methods for blind single image super resolution (SISR) conduct\nthe restoration by a learned mapping between high-resolution (HR) images and\ntheir low-resolution (LR) counterparts degraded with arbitrary blur kernels.\nHowever, these methods mostly require an independent step to estimate the blur\nkernel, leading to error accumulation between steps. We propose an end-to-end\nlearning framework for the blind SISR problem, which enables image restoration\nwithin a unified Bayesian framework with either full- or semi-supervision. The\nproposed method, namely SREMN, integrates learning techniques into the\ngeneralized expectation-maximization (GEM) algorithm and infers HR images from\nthe maximum likelihood estimation (MLE). Extensive experiments show the\nsuperiority of the proposed method with comparison to existing work and novelty\nin semi-supervised learning.\n", "date": "2023-05-23", "categories": []}
{"id": "2207.07604", "title": "Image and Texture Independent Deep Learning Noise Estimation using\n  Multiple Frames", "abstract": "  In this study, a novel multiple-frame based image and texture independent\nconvolutional Neural Network (CNN) noise estimator is introduced. The estimator\nworks.\n", "date": "2022-07-15", "categories": []}
{"id": "2305.12734", "title": "EMEF: Ensemble Multi-Exposure Image Fusion", "abstract": "  Although remarkable progress has been made in recent years, current\nmulti-exposure image fusion (MEF) research is still bounded by the lack of real\nground truth, objective evaluation function, and robust fusion strategy. In\nthis paper, we study the MEF problem from a new perspective. We don't utilize\nany synthesized ground truth, design any loss function, or develop any fusion\nstrategy. Our proposed method EMEF takes advantage of the wisdom of multiple\nimperfect MEF contributors including both conventional and deep learning-based\nmethods. Specifically, EMEF consists of two main stages: pre-train an imitator\nnetwork and tune the imitator in the runtime. In the first stage, we make a\nunified network imitate different MEF targets in a style modulation way. In the\nsecond stage, we tune the imitator network by optimizing the style code, in\norder to find an optimal fusion result for each input pair. In the experiment,\nwe construct EMEF from four state-of-the-art MEF methods and then make\ncomparisons with the individuals and several other competitive methods on the\nlatest released MEF benchmark dataset. The promising experimental results\ndemonstrate that our ensemble framework can \"get the best of all worlds\". The\ncode is available at https://github.com/medalwill/EMEF.\n", "date": "2023-05-22", "categories": []}
{"id": "2206.12745", "title": "Sequential image recovery using joint hierarchical Bayesian learning", "abstract": "  Recovering temporal image sequences (videos) based on indirect, noisy, or\nincomplete data is an essential yet challenging task. We specifically consider\nthe case where each data set is missing vital information, which prevents the\naccurate recovery of the individual images. Although some recent (variational)\nmethods have demonstrated high-resolution image recovery based on jointly\nrecovering sequential images, there remain robustness issues due to parameter\ntuning and restrictions on the type of the sequential images. Here, we present\na method based on hierarchical Bayesian learning for the joint recovery of\nsequential images that incorporates prior intra- and inter-image information.\nOur method restores the missing information in each image by \"borrowing\" it\nfrom the other images. As a result, \\emph{all} of the individual\nreconstructions yield improved accuracy. Our method can be used for various\ndata acquisitions and allows for uncertainty quantification. Some preliminary\nresults indicate its potential use for sequential deblurring and magnetic\nresonance imaging.\n", "date": "2022-06-25", "categories": []}
{"id": "2305.10006", "title": "EfficientSCI: Densely Connected Network with Space-time Factorization\n  for Large-scale Video Snapshot Compressive Imaging", "abstract": "  Video snapshot compressive imaging (SCI) uses a two-dimensional detector to\ncapture consecutive video frames during a single exposure time. Following this,\nan efficient reconstruction algorithm needs to be designed to reconstruct the\ndesired video frames. Although recent deep learning-based state-of-the-art\n(SOTA) reconstruction algorithms have achieved good results in most tasks, they\nstill face the following challenges due to excessive model complexity and GPU\nmemory limitations: 1) these models need high computational cost, and 2) they\nare usually unable to reconstruct large-scale video frames at high compression\nratios. To address these issues, we develop an efficient network for video SCI\nby using dense connections and space-time factorization mechanism within a\nsingle residual block, dubbed EfficientSCI. The EfficientSCI network can well\nestablish spatial-temporal correlation by using convolution in the spatial\ndomain and Transformer in the temporal domain, respectively. We are the first\ntime to show that an UHD color video with high compression ratio can be\nreconstructed from a snapshot 2D measurement using a single end-to-end deep\nlearning model with PSNR above 32 dB. Extensive results on both simulation and\nreal data show that our method significantly outperforms all previous SOTA\nalgorithms with better real-time performance. The code is at\nhttps://github.com/ucaswangls/EfficientSCI.git.\n", "date": "2023-05-17", "categories": []}
{"id": "2305.10889", "title": "FLIGHT Mode On: A Feather-Light Network for Low-Light Image Enhancement", "abstract": "  Low-light image enhancement (LLIE) is an ill-posed inverse problem due to the\nlack of knowledge of the desired image which is obtained under ideal\nillumination conditions. Low-light conditions give rise to two main issues: a\nsuppressed image histogram and inconsistent relative color distributions with\nlow signal-to-noise ratio. In order to address these problems, we propose a\nnovel approach named FLIGHT-Net using a sequence of neural architecture blocks.\nThe first block regulates illumination conditions through pixel-wise scene\ndependent illumination adjustment. The output image is produced in the output\nof the second block, which includes channel attention and denoising sub-blocks.\nOur highly efficient neural network architecture delivers state-of-the-art\nperformance with only 25K parameters. The method's code, pretrained models and\nresulting images will be publicly available.\n", "date": "2023-05-18", "categories": []}
{"id": "2305.09746", "title": "A Range-Null Space Decomposition Approach for Fast and Flexible Spectral\n  Compressive Imaging", "abstract": "  We present RND-SCI, a novel framework for compressive hyperspectral image\n(HSI) reconstruction. Our framework decomposes the reconstructed object into\nrange-space and null-space components, where the range-space part ensures the\nsolution conforms to the compression process, and the null-space term\nintroduces a deep HSI prior to constraining the output to have satisfactory\nproperties. RND-SCI is not only simple in design with strong interpretability\nbut also can be easily adapted to various HSI reconstruction networks,\nimproving the quality of HSIs with minimal computational overhead. RND-SCI\nsignificantly boosts the performance of HSI reconstruction networks in\nretraining, fine-tuning or plugging into a pre-trained off-the-shelf model.\nBased on the framework and SAUNet, we design an extremely fast HSI\nreconstruction network, RND-SAUNet, which achieves an astounding 91 frames per\nsecond while maintaining superior reconstruction accuracy compared to other\nless time-consuming methods. Code and models are available at\nhttps://github.com/hustvl/RND-SCI.\n", "date": "2023-05-16", "categories": []}
{"id": "2305.09236", "title": "One-shot neural band selection for spectral recovery", "abstract": "  Band selection has a great impact on the spectral recovery quality. To solve\nthis ill-posed inverse problem, most band selection methods adopt hand-crafted\npriors or exploit clustering or sparse regularization constraints to find most\nprominent bands. These methods are either very slow due to the computational\ncost of repeatedly training with respect to different selection frequencies or\ndifferent band combinations. Many traditional methods rely on the scene prior\nand thus are not applicable to other scenarios. In this paper, we present a\nnovel one-shot Neural Band Selection (NBS) framework for spectral recovery.\nUnlike conventional searching approaches with a discrete search space and a\nnon-differentiable search strategy, our NBS is based on the continuous\nrelaxation of the band selection process, thus allowing efficient band search\nusing gradient descent. To enable the compatibility for se- lecting any number\nof bands in one-shot, we further exploit the band-wise correlation matrices to\nprogressively suppress similar adjacent bands. Extensive evaluations on the\nNTIRE 2022 Spectral Reconstruction Challenge demonstrate that our NBS achieves\nconsistent performance gains over competitive baselines when examined with four\ndifferent spectral recov- ery methods. Our code will be publicly available.\n", "date": "2023-05-16", "categories": []}
{"id": "2303.15748", "title": "SVD-DIP: Overcoming the Overfitting Problem in DIP-based CT\n  Reconstruction", "abstract": "  The deep image prior (DIP) is a well-established unsupervised deep learning\nmethod for image reconstruction; yet it is far from being flawless. The DIP\noverfits to noise if not early stopped, or optimized via a regularized\nobjective. We build on the regularized fine-tuning of a pretrained DIP, by\nadopting a novel strategy that restricts the learning to the adaptation of\nsingular values. The proposed SVD-DIP uses ad hoc convolutional layers whose\npretrained parameters are decomposed via the singular value decomposition.\nOptimizing the DIP then solely consists in the fine-tuning of the singular\nvalues, while keeping the left and right singular vectors fixed. We thoroughly\nvalidate the proposed method on real-measured $\\mu$CT data of a lotus root as\nwell as two medical datasets (LoDoPaB and Mayo). We report significantly\nimproved stability of the DIP optimization, by overcoming the overfitting to\nnoise.\n", "date": "2023-03-28", "categories": []}
{"id": "2305.08585", "title": "Toward Moir\\'e-Free and Detail-Preserving Demosaicking", "abstract": "  3D convolutions are commonly employed by demosaicking neural models, in the\nsame way as solving other image restoration problems. Counter-intuitively, we\nshow that 3D convolutions implicitly impede the RGB color spectra from\nexchanging complementary information, resulting in spectral-inconsistent\ninference of the local spatial high frequency components. As a consequence,\nshallow 3D convolution networks suffer the Moir\\'e artifacts, but deep 3D\nconvolutions cause over-smoothness. We analyze the fundamental difference\nbetween demosaicking and other problems that predict lost pixels between\navailable ones (e.g., super-resolution reconstruction), and present the\nunderlying reasons for the confliction between Moir\\'e-free and\ndetail-preserving. From the new perspective, our work decouples the common\nstandard convolution procedure to spectral and spatial feature aggregations,\nwhich allow strengthening global communication in the spectral dimension while\nrespecting local contrast in the spatial dimension. We apply our demosaicking\nmodel to two tasks: Joint Demosaicking-Denoising and Independently\nDemosaicking. In both applications, our model substantially alleviates\nartifacts such as Moir\\'e and over-smoothness at similar or lower computational\ncost to currently top-performing models, as validated by diverse evaluations.\nSource code will be released along with paper publication.\n", "date": "2023-05-15", "categories": []}
{"id": "2305.07524", "title": "Joint MR sequence optimization beats pure neural network approaches for\n  spin-echo MRI super-resolution", "abstract": "  Current MRI super-resolution (SR) methods only use existing contrasts\nacquired from typical clinical sequences as input for the neural network (NN).\nIn turbo spin echo sequences (TSE) the sequence parameters can have a strong\ninfluence on the actual resolution of the acquired image and have consequently\na considera-ble impact on the performance of the NN. We propose a\nknown-operator learning approach to perform an end-to-end optimization of MR\nsequence and neural net-work parameters for SR-TSE. This MR-physics-informed\ntraining procedure jointly optimizes the radiofrequency pulse train of a proton\ndensity- (PD-) and T2-weighted TSE and a subsequently applied convolutional\nneural network to predict the corresponding PDw and T2w super-resolution TSE\nimages. The found radiofrequency pulse train designs generate an optimal signal\nfor the NN to perform the SR task. Our method generalizes from the\nsimulation-based optimi-zation to in vivo measurements and the acquired\nphysics-informed SR images show higher correlation with a time-consuming\nsegmented high-resolution TSE sequence compared to a pure network training\napproach.\n", "date": "2023-05-12", "categories": []}
{"id": "2303.11253", "title": "Zero-Shot Noise2Noise: Efficient Image Denoising without any Data", "abstract": "  Recently, self-supervised neural networks have shown excellent image\ndenoising performance. However, current dataset free methods are either\ncomputationally expensive, require a noise model, or have inadequate image\nquality. In this work we show that a simple 2-layer network, without any\ntraining data or knowledge of the noise distribution, can enable high-quality\nimage denoising at low computational cost. Our approach is motivated by\nNoise2Noise and Neighbor2Neighbor and works well for denoising pixel-wise\nindependent noise. Our experiments on artificial, real-world camera, and\nmicroscope noise show that our method termed ZS-N2N (Zero Shot Noise2Noise)\noften outperforms existing dataset-free methods at a reduced cost, making it\nsuitable for use cases with scarce data availability and limited computational\nresources. A demo of our implementation including our code and hyperparameters\ncan be found in the following colab notebook:\nhttps://colab.research.google.com/drive/1i82nyizTdszyHkaHBuKPbWnTzao8HF9b\n", "date": "2023-03-20", "categories": []}
{"id": "2305.05867", "title": "Optical Aberration Correction in Postprocessing using Imaging Simulation", "abstract": "  As the popularity of mobile photography continues to grow, considerable\neffort is being invested in the reconstruction of degraded images. Due to the\nspatial variation in optical aberrations, which cannot be avoided during the\nlens design process, recent commercial cameras have shifted some of these\ncorrection tasks from optical design to postprocessing systems. However,\nwithout engaging with the optical parameters, these systems only achieve\nlimited correction for aberrations.In this work, we propose a practical method\nfor recovering the degradation caused by optical aberrations. Specifically, we\nestablish an imaging simulation system based on our proposed optical point\nspread function model. Given the optical parameters of the camera, it generates\nthe imaging results of these specific devices. To perform the restoration, we\ndesign a spatial-adaptive network model on synthetic data pairs generated by\nthe imaging simulation system, eliminating the overhead of capturing training\ndata by a large amount of shooting and registration. Moreover, we\ncomprehensively evaluate the proposed method in simulations and experimentally\nwith a customized digital-single-lens-reflex (DSLR) camera lens and HUAWEI\nHONOR 20, respectively. The experiments demonstrate that our solution\nsuccessfully removes spatially variant blur and color dispersion. When compared\nwith the state-of-the-art deblur methods, the proposed approach achieves better\nresults with a lower computational overhead. Moreover, the reconstruction\ntechnique does not introduce artificial texture and is convenient to transfer\nto current commercial cameras. Project Page:\n\\url{https://github.com/TanGeeGo/ImagingSimulation}.\n", "date": "2023-05-10", "categories": []}
{"id": "2305.05886", "title": "Computational Optics for Mobile Terminals in Mass Production", "abstract": "  Correcting the optical aberrations and the manufacturing deviations of\ncameras is a challenging task. Due to the limitation on volume and the demand\nfor mass production, existing mobile terminals cannot rectify optical\ndegradation. In this work, we systematically construct the perturbed lens\nsystem model to illustrate the relationship between the deviated system\nparameters and the spatial frequency response measured from photographs. To\nfurther address this issue, an optimization framework is proposed based on this\nmodel to build proxy cameras from the machining samples' SFRs. Engaging with\nthe proxy cameras, we synthetic data pairs, which encode the optical\naberrations and the random manufacturing biases, for training the\nlearning-based algorithms. In correcting aberration, although promising results\nhave been shown recently with convolutional neural networks, they are hard to\ngeneralize to stochastic machining biases. Therefore, we propose a dilated\nOmni-dimensional dynamic convolution and implement it in post-processing to\naccount for the manufacturing degradation. Extensive experiments which evaluate\nmultiple samples of two representative devices demonstrate that the proposed\noptimization framework accurately constructs the proxy camera. And the dynamic\nprocessing model is well-adapted to manufacturing deviations of different\ncameras, realizing perfect computational photography. The evaluation shows that\nthe proposed method bridges the gap between optical design, system machining,\nand post-processing pipeline, shedding light on the joint of image signal\nreception (lens and sensor) and image signal processing.\n", "date": "2023-05-10", "categories": []}
{"id": "2305.05020", "title": "Domain independent post-processing with graph U-nets: Applications to\n  Electrical Impedance Tomographic Imaging", "abstract": "  Reconstruction of tomographic images from boundary measurements requires\nflexibility with respect to target domains. For instance, when the system\nequations are modeled by partial differential equations the reconstruction is\nusually done on finite element (FE) meshes, allowing for flexible geometries.\nThus, any processing of the obtained reconstructions should be ideally done on\nthe FE mesh as well. For this purpose, we extend the hugely successful U-Net\narchitecture that is limited to rectangular pixel or voxel domains to an\nequivalent that works flexibly on FE meshes. To achieve this, the FE mesh is\nconverted into a graph and we formulate a graph U-Net with a new cluster\npooling and unpooling on the graph that mimics the classic neighborhood based\nmax-pooling. We demonstrate effectiveness and flexibility of the graph U-Net\nfor improving reconstructions from electrical impedance tomographic (EIT)\nmeasurements, a nonlinear and highly ill-posed inverse problem. The performance\nis evaluated for simulated data and from three measurement devices with\ndifferent measurement geometries and instrumentations. We successfully show\nthat such networks can be trained with a simple two-dimensional simulated\ntraining set and generalize to very different domains, including measurements\nfrom a three-dimensional device and subsequent 3D reconstructions.\n", "date": "2023-05-08", "categories": []}
{"id": "2305.03899", "title": "NL-CS Net: Deep Learning with Non-Local Prior for Image Compressive\n  Sensing", "abstract": "  Deep learning has been applied to compressive sensing (CS) of images\nsuccessfully in recent years. However, existing network-based methods are often\ntrained as the black box, in which the lack of prior knowledge is often the\nbottleneck for further performance improvement. To overcome this drawback, this\npaper proposes a novel CS method using non-local prior which combines the\ninterpretability of the traditional optimization methods with the speed of\nnetwork-based methods, called NL-CS Net. We unroll each phase from iteration of\nthe augmented Lagrangian method solving non-local and sparse regularized\noptimization problem by a network. NL-CS Net is composed of the up-sampling\nmodule and the recovery module. In the up-sampling module, we use learnable\nup-sampling matrix instead of a predefined one. In the recovery module,\npatch-wise non-local network is employed to capture long-range feature\ncorrespondences. Important parameters involved (e.g. sampling matrix, nonlinear\ntransforms, shrinkage thresholds, step size, $etc.$) are learned end-to-end,\nrather than hand-crafted. Furthermore, to facilitate practical implementation,\northogonal and binary constraints on the sampling matrix are simultaneously\nadopted. Extensive experiments on natural images and magnetic resonance imaging\n(MRI) demonstrate that the proposed method outperforms the state-of-the-art\nmethods while maintaining great interpretability and speed.\n", "date": "2023-05-06", "categories": []}
{"id": "2305.04054", "title": "SST-ReversibleNet: Reversible-prior-based Spectral-Spatial Transformer\n  for Efficient Hyperspectral Image Reconstruction", "abstract": "  Spectral image reconstruction is an important task in snapshot compressed\nimaging. This paper aims to propose a new end-to-end framework with iterative\ncapabilities similar to a deep unfolding network to improve reconstruction\naccuracy, independent of optimization conditions, and to reduce the number of\nparameters. A novel framework called the reversible-prior-based method is\nproposed. Inspired by the reversibility of the optical path, the\nreversible-prior-based framework projects the reconstructions back into the\nmeasurement space, and then the residuals between the projected data and the\nreal measurements are fed into the network for iteration. The reconstruction\nsubnet in the network then learns the mapping of the residuals to the true\nvalues to improve reconstruction accuracy. In addition, a novel\nspectral-spatial transformer is proposed to account for the global correlation\nof spectral data in both spatial and spectral dimensions while balancing\nnetwork depth and computational complexity, in response to the shortcomings of\nexisting transformer-based denoising modules that ignore spatial texture\nfeatures or learn local spatial features at the expense of global spatial\nfeatures. Extensive experiments show that our SST-ReversibleNet significantly\noutperforms state-of-the-art methods on simulated and real HSI datasets, while\nrequiring lower computational and storage costs.\nhttps://github.com/caizeyu1992/SST\n", "date": "2023-05-06", "categories": []}
{"id": "2305.03226", "title": "Sign-Coded Exposure Sensing for Noise-Robust High-Speed Imaging", "abstract": "  We present a novel Fourier camera, an in-hardware optical compression of\nhigh-speed frames employing pixel-level sign-coded exposure where pixel\nintensities temporally modulated as positive and negative exposure are combined\nto yield Hadamard coefficients. The orthogonality of Walsh functions ensures\nthat the noise is not amplified during high-speed frame reconstruction, making\nit a much more attractive option for coded exposure systems aimed at very high\nframe rate operation. Frame reconstruction is carried out by a single-pass\ndemosaicking of the spatially multiplexed Walsh functions in a lattice\narrangement, significantly reducing the computational complexity. The\nsimulation prototype confirms the improved robustness to noise compared to the\nbinary-coded exposure patterns, such as one-hot encoding and pseudo-random\nencoding. Our hardware prototype demonstrated the reconstruction of 4kHz frames\nof a moving scene lit by ambient light only.\n", "date": "2023-05-05", "categories": []}
{"id": "2211.08658", "title": "Consistent Direct Time-of-Flight Video Depth Super-Resolution", "abstract": "  Direct time-of-flight (dToF) sensors are promising for next-generation\non-device 3D sensing. However, limited by manufacturing capabilities in a\ncompact module, the dToF data has a low spatial resolution (e.g., $\\sim\n20\\times30$ for iPhone dToF), and it requires a super-resolution step before\nbeing passed to downstream tasks. In this paper, we solve this super-resolution\nproblem by fusing the low-resolution dToF data with the corresponding\nhigh-resolution RGB guidance. Unlike the conventional RGB-guided depth\nenhancement approaches, which perform the fusion in a per-frame manner, we\npropose the first multi-frame fusion scheme to mitigate the spatial ambiguity\nresulting from the low-resolution dToF imaging. In addition, dToF sensors\nprovide unique depth histogram information for each local patch, and we\nincorporate this dToF-specific feature in our network design to further\nalleviate spatial ambiguity. To evaluate our models on complex dynamic indoor\nenvironments and to provide a large-scale dToF sensor dataset, we introduce\nDyDToF, the first synthetic RGB-dToF video dataset that features dynamic\nobjects and a realistic dToF simulator following the physical imaging process.\nWe believe the methods and dataset are beneficial to a broad community as dToF\ndepth sensing is becoming mainstream on mobile devices. Our code and data are\npublicly available: https://github.com/facebookresearch/DVSR/\n", "date": "2022-11-16", "categories": []}
{"id": "2206.08612", "title": "OADAT: Experimental and Synthetic Clinical Optoacoustic Data for\n  Standardized Image Processing", "abstract": "  Optoacoustic (OA) imaging is based on excitation of biological tissues with\nnanosecond-duration laser pulses followed by subsequent detection of ultrasound\nwaves generated via light-absorption-mediated thermoelastic expansion. OA\nimaging features a powerful combination between rich optical contrast and high\nresolution in deep tissues. This enabled the exploration of a number of\nattractive new applications both in clinical and laboratory settings. However,\nno standardized datasets generated with different types of experimental set-up\nand associated processing methods are available to facilitate advances in\nbroader applications of OA in clinical settings. This complicates an objective\ncomparison between new and established data processing methods, often leading\nto qualitative results and arbitrary interpretations of the data. In this\npaper, we provide both experimental and synthetic OA raw signals and\nreconstructed image domain datasets rendered with different experimental\nparameters and tomographic acquisition geometries. We further provide trained\nneural networks to tackle three important challenges related to OA image\nprocessing, namely accurate reconstruction under limited view tomographic\nconditions, removal of spatial undersampling artifacts and anatomical\nsegmentation for improved image reconstruction. Specifically, we define 44\nexperiments corresponding to the aforementioned challenges as benchmarks to be\nused as a reference for the development of more advanced processing methods.\n", "date": "2022-06-17", "categories": []}
{"id": "2305.02030", "title": "Near-Field MIMO-ISAR Millimeter-Wave Imaging", "abstract": "  Multiple-input-multiple-output (MIMO) millimeter-wave (mmWave) sensors for\nsynthetic aperture radar (SAR) and inverse SAR (ISAR) address the fundamental\nchallenges of cost-effectiveness and scalability inherent to near-field\nimaging. In this paper, near-field MIMO-ISAR mmWave imaging systems are\ndiscussed and developed. The rotational ISAR (R-ISAR) regime investigated in\nthis paper requires rotating the target at a constant radial distance from the\ntransceiver and scanning the transceiver along a vertical track. Using a 77GHz\nmmWave radar, a high resolution three-dimensional (3-D) image can be\nreconstructed from this two-dimensional scanning taking into account the\nspherical near-field wavefront. While prior work in literature consists of\nsingle-input-single-output circular synthetic aperture radar (SISO-CSAR)\nalgorithms or computationally sluggish MIMO-CSAR image reconstruction\nalgorithms, this paper proposes a novel algorithm for efficient MIMO 3-D\nholographic imaging and details the design of a MIMO R-ISAR imaging system. The\nproposed algorithm applies a multistatic-to-monostatic phase compensation to\nthe R-ISAR regime allowing for use of highly efficient monostatic algorithms.\nWe demonstrate the algorithm's performance in real-world imaging scenarios on a\nprototyped MIMO R-ISAR platform. Our fully integrated system, consisting of a\nmechanical scanner and efficient imaging algorithm, is capable of pairing the\nscanning efficiency of the MIMO regime with the computational efficiency of\nsingle pixel image reconstruction algorithms.\n", "date": "2023-05-03", "categories": []}
{"id": "2305.02092", "title": "Efficient CNN-based Super Resolution Algorithms for mmWave Mobile Radar\n  Imaging", "abstract": "  In this paper, we introduce an innovative super resolution approach to\nemerging modes of near-field synthetic aperture radar (SAR) imaging. Recent\nresearch extends convolutional neural network (CNN) architectures from the\noptical to the electromagnetic domain to achieve super resolution on images\ngenerated from radar signaling. Specifically, near-field synthetic aperture\nradar (SAR) imaging, a method for generating high-resolution images by scanning\na radar across space to create a synthetic aperture, is of interest due to its\nhigh-fidelity spatial sensing capability, low cost devices, and large\napplication space. Since SAR imaging requires large aperture sizes to achieve\nhigh resolution, super-resolution algorithms are valuable for many\napplications. Freehand smartphone SAR, an emerging sensing modality, requires\nirregular SAR apertures in the near-field and computation on mobile devices.\nAchieving efficient high-resolution SAR images from irregularly sampled data\ncollected by freehand motion of a smartphone is a challenging task. In this\npaper, we propose a novel CNN architecture to achieve SAR image\nsuper-resolution for mobile applications by employing state-of-the-art SAR\nprocessing and deep learning techniques. The proposed algorithm is verified via\nsimulation and an empirical study. Our algorithm demonstrates high-efficiency\nand high-resolution radar imaging for near-field scenarios with irregular\nscanning geometries.\n", "date": "2023-05-03", "categories": []}
{"id": "2303.10566", "title": "Dynamical Hyperspectral Unmixing with Variational Recurrent Neural\n  Networks", "abstract": "  Multitemporal hyperspectral unmixing (MTHU) is a fundamental tool in the\nanalysis of hyperspectral image sequences. It reveals the dynamical evolution\nof the materials (endmembers) and of their proportions (abundances) in a given\nscene. However, adequately accounting for the spatial and temporal variability\nof the endmembers in MTHU is challenging, and has not been fully addressed so\nfar in unsupervised frameworks. In this work, we propose an unsupervised MTHU\nalgorithm based on variational recurrent neural networks. First, a stochastic\nmodel is proposed to represent both the dynamical evolution of the endmembers\nand their abundances, as well as the mixing process. Moreover, a new model\nbased on a low-dimensional parametrization is used to represent spatial and\ntemporal endmember variability, significantly reducing the amount of variables\nto be estimated. We propose to formulate MTHU as a Bayesian inference problem.\nHowever, the solution to this problem does not have an analytical solution due\nto the nonlinearity and non-Gaussianity of the model. Thus, we propose a\nsolution based on deep variational inference, in which the posterior\ndistribution of the estimated abundances and endmembers is represented by using\na combination of recurrent neural networks and a physically motivated model.\nThe parameters of the model are learned using stochastic backpropagation.\nExperimental results show that the proposed method outperforms state of the art\nMTHU algorithms.\n", "date": "2023-03-19", "categories": []}
{"id": "2304.14610", "title": "ALL-E: Aesthetics-guided Low-light Image Enhancement", "abstract": "  Evaluating the performance of low-light image enhancement (LLE) is highly\nsubjective, thus making integrating human preferences into image enhancement a\nnecessity. Existing methods fail to consider this and present a series of\npotentially valid heuristic criteria for training enhancement models. In this\npaper, we propose a new paradigm, i.e., aesthetics-guided low-light image\nenhancement (ALL-E), which introduces aesthetic preferences to LLE and\nmotivates training in a reinforcement learning framework with an aesthetic\nreward. Each pixel, functioning as an agent, refines itself by recursive\nactions, i.e., its corresponding adjustment curve is estimated sequentially.\nExtensive experiments show that integrating aesthetic assessment improves both\nsubjective experience and objective evaluation. Our results on various\nbenchmarks demonstrate the superiority of ALL-E over state-of-the-art methods.\n", "date": "2023-04-28", "categories": []}
{"id": "2212.02025", "title": "Unidirectional Imaging using Deep Learning-Designed Materials", "abstract": "  A unidirectional imager would only permit image formation along one\ndirection, from an input field-of-view (FOV) A to an output FOV B, and in the\nreverse path, the image formation would be blocked. Here, we report the first\ndemonstration of unidirectional imagers, presenting polarization-insensitive\nand broadband unidirectional imaging based on successive diffractive layers\nthat are linear and isotropic. These diffractive layers are optimized using\ndeep learning and consist of hundreds of thousands of diffractive phase\nfeatures, which collectively modulate the incoming fields and project an\nintensity image of the input onto an output FOV, while blocking the image\nformation in the reverse direction. After their deep learning-based training,\nthe resulting diffractive layers are fabricated to form a unidirectional\nimager. As a reciprocal device, the diffractive unidirectional imager has\nasymmetric mode processing capabilities in the forward and backward directions,\nwhere the optical modes from B to A are selectively guided/scattered to miss\nthe output FOV, whereas for the forward direction such modal losses are\nminimized, yielding an ideal imaging system between the input and output FOVs.\nAlthough trained using monochromatic illumination, the diffractive\nunidirectional imager maintains its functionality over a large spectral band\nand works under broadband illumination. We experimentally validated this\nunidirectional imager using terahertz radiation, very well matching our\nnumerical results. Using the same deep learning-based design strategy, we also\ncreated a wavelength-selective unidirectional imager, where two unidirectional\nimaging operations, in reverse directions, are multiplexed through different\nillumination wavelengths. Diffractive unidirectional imaging using structured\nmaterials will have numerous applications in e.g., security, defense,\ntelecommunications and privacy protection.\n", "date": "2022-12-05", "categories": []}
{"id": "2305.00088", "title": "DD-CISENet: Dual-Domain Cross-Iteration Squeeze and Excitation Network\n  for Accelerated MRI Reconstruction", "abstract": "  Magnetic resonance imaging (MRI) is widely employed for diagnostic tests in\nneurology. However, the utility of MRI is largely limited by its long\nacquisition time. Acquiring fewer k-space data in a sparse manner is a\npotential solution to reducing the acquisition time, but it can lead to severe\naliasing reconstruction artifacts. In this paper, we present a novel\nDual-Domain Cross-Iteration Squeeze and Excitation Network (DD-CISENet) for\naccelerated sparse MRI reconstruction. The information of k-spaces and MRI\nimages can be iteratively fused and maintained using the Cross-Iteration\nResidual connection (CIR) structures. This study included 720 multi-coil brain\nMRI cases adopted from the open-source fastMRI Dataset. Results showed that the\naverage reconstruction error by DD-CISENet was 2.28 $\\pm$ 0.57%, which\noutperformed existing deep learning methods including image-domain prediction\n(6.03 $\\pm$ 1.31, p < 0.001), k-space synthesis (6.12 $\\pm$ 1.66, p < 0.001),\nand dual-domain feature fusion approaches (4.05 $\\pm$ 0.88, p < 0.001).\n", "date": "2023-04-28", "categories": []}
{"id": "2304.14736", "title": "Differentiable Sensor Layouts for End-to-End Learning of Task-Specific\n  Camera Parameters", "abstract": "  The success of deep learning is frequently described as the ability to train\nall parameters of a network on a specific application in an end-to-end fashion.\nYet, several design choices on the camera level, including the pixel layout of\nthe sensor, are considered as pre-defined and fixed, and high resolution,\nregular pixel layouts are considered to be the most generic ones in computer\nvision and graphics, treating all regions of an image as equally important.\nWhile several works have considered non-uniform, \\eg, hexagonal or foveated,\npixel layouts in hardware and image processing, the layout has not been\nintegrated into the end-to-end learning paradigm so far. In this work, we\npresent the first truly end-to-end trained imaging pipeline that optimizes the\nsize and distribution of pixels on the imaging sensor jointly with the\nparameters of a given neural network on a specific task. We derive an analytic,\ndifferentiable approach for the sensor layout parameterization that allows for\ntask-specific, local varying pixel resolutions. We present two pixel layout\nparameterization functions: rectangular and curvilinear grid shapes that retain\na regular topology. We provide a drop-in module that approximates sensor\nsimulation given existing high-resolution images to directly connect our method\nwith existing deep learning models. We show that network predictions benefit\nfrom learnable pixel layouts for two different downstream tasks, classification\nand semantic segmentation.\n", "date": "2023-04-28", "categories": []}
{"id": "2304.14841", "title": "3D shape reconstruction of semi-transparent worms", "abstract": "  3D shape reconstruction typically requires identifying object features or\ntextures in multiple images of a subject. This approach is not viable when the\nsubject is semi-transparent and moving in and out of focus. Here we overcome\nthese challenges by rendering a candidate shape with adaptive blurring and\ntransparency for comparison with the images. We use the microscopic nematode\nCaenorhabditis elegans as a case study as it freely explores a 3D complex fluid\nwith constantly changing optical properties. We model the slender worm as a 3D\ncurve using an intrinsic parametrisation that naturally admits\nbiologically-informed constraints and regularisation. To account for the\nchanging optics we develop a novel differentiable renderer to construct images\nfrom 2D projections and compare against raw images to generate a pixel-wise\nerror to jointly update the curve, camera and renderer parameters using\ngradient descent. The method is robust to interference such as bubbles and dirt\ntrapped in the fluid, stays consistent through complex sequences of postures,\nrecovers reliable estimates from blurry images and provides a significant\nimprovement on previous attempts to track C. elegans in 3D. Our results\ndemonstrate the potential of direct approaches to shape estimation in complex\nphysical environments in the absence of ground-truth data.\n", "date": "2023-04-28", "categories": []}
{"id": "2304.10089", "title": "MIPI 2023 Challenge on RGBW Fusion: Methods and Results", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for an in-depth exchange of views\nfrom industry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). With the success of the 1st MIPI Workshop@ECCV\n2022, we introduce the second MIPI challenge, including four tracks focusing on\nnovel image sensors and imaging algorithms. This paper summarizes and reviews\nthe RGBW Joint Fusion and Denoise track on MIPI 2023. In total, 69 participants\nwere successfully registered, and 4 teams submitted results in the final\ntesting phase. The final results are evaluated using objective metrics,\nincluding PSNR, SSIM, LPIPS, and KLD. A detailed description of the top three\nmodels developed in this challenge is provided in this paper. More details of\nthis challenge and the link to the dataset can be found at\nhttps://mipi-challenge.org/MIPI2023/.\n", "date": "2023-04-20", "categories": []}
{"id": "2304.12584", "title": "Learning imaging mechanism directly from optical microscopy observations", "abstract": "  Optical microscopy image plays an important role in scientific research\nthrough the direct visualization of the nanoworld, where the imaging mechanism\nis described as the convolution of the point spread function (PSF) and\nemitters. Based on a priori knowledge of the PSF or equivalent PSF, it is\npossible to achieve more precise exploration of the nanoworld. However, it is\nan outstanding challenge to directly extract the PSF from microscopy images.\nHere, with the help of self-supervised learning, we propose a physics-informed\nmasked autoencoder (PiMAE) that enables a learnable estimation of the PSF and\nemitters directly from the raw microscopy images. We demonstrate our method in\nsynthetic data and real-world experiments with significant accuracy and noise\nrobustness. PiMAE outperforms DeepSTORM and the Richardson-Lucy algorithm in\nsynthetic data tasks with an average improvement of 19.6\\% and 50.7\\% (35\ntasks), respectively, as measured by the normalized root mean square error\n(NRMSE) metric. This is achieved without prior knowledge of the PSF, in\ncontrast to the supervised approach used by DeepSTORM and the known PSF\nassumption in the Richardson-Lucy algorithm. Our method, PiMAE, provides a\nfeasible scheme for achieving the hidden imaging mechanism in optical\nmicroscopy and has the potential to learn hidden mechanisms in many more\nsystems.\n", "date": "2023-04-25", "categories": []}
{"id": "2304.12748", "title": "Inverting the Imaging Process by Learning an Implicit Camera Model", "abstract": "  Representing visual signals with implicit coordinate-based neural networks,\nas an effective replacement of the traditional discrete signal representation,\nhas gained considerable popularity in computer vision and graphics. In contrast\nto existing implicit neural representations which focus on modelling the scene\nonly, this paper proposes a novel implicit camera model which represents the\nphysical imaging process of a camera as a deep neural network. We demonstrate\nthe power of this new implicit camera model on two inverse imaging tasks: i)\ngenerating all-in-focus photos, and ii) HDR imaging. Specifically, we devise an\nimplicit blur generator and an implicit tone mapper to model the aperture and\nexposure of the camera's imaging process, respectively. Our implicit camera\nmodel is jointly learned together with implicit scene models under multi-focus\nstack and multi-exposure bracket supervision. We have demonstrated the\neffectiveness of our new model on a large number of test images and videos,\nproducing accurate and visually appealing all-in-focus and high dynamic range\nimages. In principle, our new implicit neural camera model has the potential to\nbenefit a wide array of other inverse imaging tasks.\n", "date": "2023-04-25", "categories": []}
{"id": "2304.09254", "title": "FastMRI Prostate: A Publicly Available, Biparametric MRI Dataset to\n  Advance Machine Learning for Prostate Cancer Imaging", "abstract": "  The fastMRI brain and knee dataset has enabled significant advances in\nexploring reconstruction methods for improving speed and image quality for\nMagnetic Resonance Imaging (MRI) via novel, clinically relevant reconstruction\napproaches. In this study, we describe the April 2023 expansion of the fastMRI\ndataset to include biparametric prostate MRI data acquired on a clinical\npopulation. The dataset consists of raw k-space and reconstructed images for\nT2-weighted and diffusion-weighted sequences along with slice-level labels that\nindicate the presence and grade of prostate cancer. As has been the case with\nfastMRI, increasing accessibility to raw prostate MRI data will further\nfacilitate research in MR image reconstruction and evaluation with the larger\ngoal of improving the utility of MRI for prostate cancer detection and\nevaluation. The dataset is available at https://fastmri.med.nyu.edu.\n", "date": "2023-04-18", "categories": []}
{"id": "2304.11238", "title": "Adapting model-based deep learning to multiple acquisition conditions:\n  Ada-MoDL", "abstract": "  Purpose: The aim of this work is to introduce a single model-based deep\nnetwork that can provide high-quality reconstructions from undersampled\nparallel MRI data acquired with multiple sequences, acquisition settings and\nfield strengths.\n  Methods: A single unrolled architecture, which offers good reconstructions\nfor multiple acquisition settings, is introduced. The proposed scheme adapts\nthe model to each setting by scaling the CNN features and the regularization\nparameter with appropriate weights. The scaling weights and regularization\nparameter are derived using a multi-layer perceptron model from conditional\nvectors, which represents the specific acquisition setting. The perceptron\nparameters and the CNN weights are jointly trained using data from multiple\nacquisition settings, including differences in field strengths, acceleration,\nand contrasts. The conditional network is validated using datasets acquired\nwith different acquisition settings.\n  Results: The comparison of the adaptive framework, which trains a single\nmodel using the data from all the settings, shows that it can offer\nconsistently improved performance for each acquisition condition. The\ncomparison of the proposed scheme with networks that are trained independently\nfor each acquisition setting shows that it requires less training data per\nacquisition setting to offer good performance.\n  Conclusion: The Ada-MoDL framework enables the use of a single model-based\nunrolled network for multiple acquisition settings. In addition to eliminating\nthe need to train and store multiple networks for different acquisition\nsettings, this approach reduces the training data needed for each acquisition\nsetting.\n", "date": "2023-04-21", "categories": []}
{"id": "2304.10551", "title": "MIPI 2023 Challenge on RGBW Remosaic: Methods and Results", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for an in-depth exchange of views\nfrom industry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). With the success of the 1st MIPI Workshop@ECCV\n2022, we introduce the second MIPI challenge, including four tracks focusing on\nnovel image sensors and imaging algorithms. This paper summarizes and reviews\nthe RGBW Joint Remosaic and Denoise track on MIPI 2023. In total, 81\nparticipants were successfully registered, and 4 teams submitted results in the\nfinal testing phase. The final results are evaluated using objective metrics,\nincluding PSNR, SSIM, LPIPS, and KLD. A detailed description of the top three\nmodels developed in this challenge is provided in this paper. More details of\nthis challenge and the link to the dataset can be found at\nhttps://mipi-challenge.org/MIPI2023/.\n", "date": "2023-04-20", "categories": []}
{"id": "2304.10780", "title": "Omni-Line-of-Sight Imaging for Holistic Shape Reconstruction", "abstract": "  We introduce Omni-LOS, a neural computational imaging method for conducting\nholistic shape reconstruction (HSR) of complex objects utilizing a\nSingle-Photon Avalanche Diode (SPAD)-based time-of-flight sensor. As\nillustrated in Fig. 1, our method enables new capabilities to reconstruct\nnear-$360^\\circ$ surrounding geometry of an object from a single scan spot. In\nsuch a scenario, traditional line-of-sight (LOS) imaging methods only see the\nfront part of the object and typically fail to recover the occluded back\nregions. Inspired by recent advances of non-line-of-sight (NLOS) imaging\ntechniques which have demonstrated great power to reconstruct occluded objects,\nOmni-LOS marries LOS and NLOS together, leveraging their complementary\nadvantages to jointly recover the holistic shape of the object from a single\nscan position. The core of our method is to put the object nearby diffuse walls\nand augment the LOS scan in the front view with the NLOS scans from the\nsurrounding walls, which serve as virtual ``mirrors'' to trap lights toward the\nobject. Instead of separately recovering the LOS and NLOS signals, we adopt an\nimplicit neural network to represent the object, analogous to NeRF and NeTF.\nWhile transients are measured along straight rays in LOS but over the spherical\nwavefronts in NLOS, we derive differentiable ray propagation models to\nsimultaneously model both types of transient measurements so that the NLOS\nreconstruction also takes into account the direct LOS measurements and vice\nversa. We further develop a proof-of-concept Omni-LOS hardware prototype for\nreal-world validation. Comprehensive experiments on various wall settings\ndemonstrate that Omni-LOS successfully resolves shape ambiguities caused by\nocclusions, achieves high-fidelity 3D scan quality, and manages to recover\nobjects of various scales and complexity.\n", "date": "2023-04-21", "categories": []}
{"id": "2303.06834", "title": "DarkVisionNet: Low-Light Imaging via RGB-NIR Fusion with Deep\n  Inconsistency Prior", "abstract": "  RGB-NIR fusion is a promising method for low-light imaging. However,\nhigh-intensity noise in low-light images amplifies the effect of structure\ninconsistency between RGB-NIR images, which fails existing algorithms. To\nhandle this, we propose a new RGB-NIR fusion algorithm called Dark Vision Net\n(DVN) with two technical novelties: Deep Structure and Deep Inconsistency Prior\n(DIP). The Deep Structure extracts clear structure details in deep multiscale\nfeature space rather than raw input space, which is more robust to noisy\ninputs. Based on the deep structures from both RGB and NIR domains, we\nintroduce the DIP to leverage the structure inconsistency to guide the fusion\nof RGB-NIR. Benefiting from this, the proposed DVN obtains high-quality\nlowlight images without the visual artifacts. We also propose a new dataset\ncalled Dark Vision Dataset (DVD), consisting of aligned RGB-NIR image pairs, as\nthe first public RGBNIR fusion benchmark. Quantitative and qualitative results\non the proposed benchmark show that DVN significantly outperforms other\ncomparison algorithms in PSNR and SSIM, especially in extremely low light\nconditions.\n", "date": "2023-03-13", "categories": []}
{"id": "2304.06930", "title": "Self-Supervised Scene Dynamic Recovery from Rolling Shutter Images and\n  Events", "abstract": "  Scene Dynamic Recovery (SDR) by inverting distorted Rolling Shutter (RS)\nimages to an undistorted high frame-rate Global Shutter (GS) video is a\nseverely ill-posed problem due to the missing temporal dynamic information in\nboth RS intra-frame scanlines and inter-frame exposures, particularly when\nprior knowledge about camera/object motions is unavailable. Commonly used\nartificial assumptions on scenes/motions and data-specific characteristics are\nprone to producing sub-optimal solutions in real-world scenarios. To address\nthis challenge, we propose an event-based SDR network within a self-supervised\nlearning paradigm, i.e., SelfUnroll. We leverage the extremely high temporal\nresolution of event cameras to provide accurate inter/intra-frame dynamic\ninformation. Specifically, an Event-based Inter/intra-frame Compensator (E-IC)\nis proposed to predict the per-pixel dynamic between arbitrary time intervals,\nincluding the temporal transition and spatial translation. Exploring\nconnections in terms of RS-RS, RS-GS, and GS-RS, we explicitly formulate mutual\nconstraints with the proposed E-IC, resulting in supervisions without\nground-truth GS images. Extensive evaluations over synthetic and real datasets\ndemonstrate that the proposed method achieves state-of-the-art and shows\nremarkable performance for event-based RS2GS inversion in real-world scenarios.\nThe dataset and code are available at https://w3un.github.io/selfunroll/.\n", "date": "2023-04-14", "categories": []}
{"id": "2211.01372", "title": "Investigating the robustness of a learning-based method for quantitative\n  phase retrieval from propagation-based x-ray phase contrast measurements\n  under laboratory conditions", "abstract": "  Quantitative phase retrieval (QPR) in propagation-based x-ray phase contrast\nimaging of heterogeneous and structurally complicated objects is challenging\nunder laboratory conditions due to partial spatial coherence and\npolychromaticity. A learning-based method (LBM) provides a non-linear approach\nto this problem while not being constrained by restrictive assumptions about\nobject properties and beam coherence. In this work, a LBM was assessed for its\napplicability under practical scenarios by evaluating its robustness and\ngeneralizability under typical experimental variations. Towards this end, an\nend-to-end LBM was employed for QPR under laboratory conditions and its\nrobustness was investigated across various system and object conditions. The\nrobustness of the method was tested via varying propagation distances and its\ngeneralizability with respect to object structure and experimental data was\nalso tested. Although the LBM was stable under the studied variations, its\nsuccessful deployment was found to be affected by choices pertaining to data\npre-processing, network training considerations and system modeling. To our\nknowledge, we demonstrated for the first time, the potential applicability of\nan end-to-end learning-based quantitative phase retrieval method, trained on\nsimulated data, to experimental propagation-based x-ray phase contrast\nmeasurements acquired under laboratory conditions. We considered conditions of\npolychromaticity, partial spatial coherence, and high noise levels, typical to\nlaboratory conditions. This work further explored the robustness of this method\nto practical variations in propagation distances and object structure with the\ngoal of assessing its potential for experimental use. Such an exploration of\nany LBM (irrespective of its network architecture) before practical deployment\nprovides an understanding of its potential behavior under experimental\nsettings.\n", "date": "2022-11-02", "categories": []}
{"id": "2304.07874", "title": "A Data-Centric Solution to NonHomogeneous Dehazing via Vision\n  Transformer", "abstract": "  Recent years have witnessed an increased interest in image dehazing. Many\ndeep learning methods have been proposed to tackle this challenge, and have\nmade significant accomplishments dealing with homogeneous haze. However, these\nsolutions cannot maintain comparable performance when they are applied to\nimages with non-homogeneous haze, e.g., NH-HAZE23 dataset introduced by NTIRE\nchallenges. One of the reasons for such failures is that non-homogeneous haze\ndoes not obey one of the assumptions that is required for modeling homogeneous\nhaze. In addition, a large number of pairs of non-homogeneous hazy image and\nthe clean counterpart is required using traditional end-to-end training\napproaches, while NH-HAZE23 dataset is of limited quantities. Although it is\npossible to augment the NH-HAZE23 dataset by leveraging other non-homogeneous\ndehazing datasets, we observe that it is necessary to design a proper\ndata-preprocessing approach that reduces the distribution gaps between the\ntarget dataset and the augmented one. This finding indeed aligns with the\nessence of data-centric AI. With a novel network architecture and a principled\ndata-preprocessing approach that systematically enhances data quality, we\npresent an innovative dehazing method. Specifically, we apply RGB-channel-wise\ntransformations on the augmented datasets, and incorporate the state-of-the-art\ntransformers as the backbone in the two-branch framework. We conduct extensive\nexperiments and ablation study to demonstrate the effectiveness of our proposed\nmethod.\n", "date": "2023-04-16", "categories": []}
{"id": "2304.06943", "title": "A Unified HDR Imaging Method with Pixel and Patch Level", "abstract": "  Mapping Low Dynamic Range (LDR) images with different exposures to High\nDynamic Range (HDR) remains nontrivial and challenging on dynamic scenes due to\nghosting caused by object motion or camera jitting. With the success of Deep\nNeural Networks (DNNs), several DNNs-based methods have been proposed to\nalleviate ghosting, they cannot generate approving results when motion and\nsaturation occur. To generate visually pleasing HDR images in various cases, we\npropose a hybrid HDR deghosting network, called HyHDRNet, to learn the\ncomplicated relationship between reference and non-reference images. The\nproposed HyHDRNet consists of a content alignment subnetwork and a\nTransformer-based fusion subnetwork. Specifically, to effectively avoid\nghosting from the source, the content alignment subnetwork uses patch\naggregation and ghost attention to integrate similar content from other\nnon-reference images with patch level and suppress undesired components with\npixel level. To achieve mutual guidance between patch-level and pixel-level, we\nleverage a gating module to sufficiently swap useful information both in\nghosted and saturated regions. Furthermore, to obtain a high-quality HDR image,\nthe Transformer-based fusion subnetwork uses a Residual Deformable Transformer\nBlock (RDTB) to adaptively merge information for different exposed regions. We\nexamined the proposed method on four widely used public HDR image deghosting\ndatasets. Experiments demonstrate that HyHDRNet outperforms state-of-the-art\nmethods both quantitatively and qualitatively, achieving appealing HDR\nvisualization with unified textures and colors.\n", "date": "2023-04-14", "categories": []}
{"id": "2304.07473", "title": "Hierarchical Interactive Reconstruction Network For Video Compressive\n  Sensing", "abstract": "  Deep network-based image and video Compressive Sensing(CS) has attracted\nincreasing attentions in recent years. However, in the existing deep\nnetwork-based CS methods, a simple stacked convolutional network is usually\nadopted, which not only weakens the perception of rich contextual prior\nknowledge, but also limits the exploration of the correlations between temporal\nvideo frames. In this paper, we propose a novel Hierarchical InTeractive Video\nCS Reconstruction Network(HIT-VCSNet), which can cooperatively exploit the deep\npriors in both spatial and temporal domains to improve the reconstruction\nquality. Specifically, in the spatial domain, a novel hierarchical structure is\ndesigned, which can hierarchically extract deep features from keyframes and\nnon-keyframes. In the temporal domain, a novel hierarchical interaction\nmechanism is proposed, which can cooperatively learn the correlations among\ndifferent frames in the multiscale space. Extensive experiments manifest that\nthe proposed HIT-VCSNet outperforms the existing state-of-the-art video and\nimage CS methods in a large margin.\n", "date": "2023-04-15", "categories": []}
{"id": "2304.08444", "title": "SCANet: Self-Paced Semi-Curricular Attention Network for Non-Homogeneous\n  Image Dehazing", "abstract": "  The presence of non-homogeneous haze can cause scene blurring, color\ndistortion, low contrast, and other degradations that obscure texture details.\nExisting homogeneous dehazing methods struggle to handle the non-uniform\ndistribution of haze in a robust manner. The crucial challenge of\nnon-homogeneous dehazing is to effectively extract the non-uniform distribution\nfeatures and reconstruct the details of hazy areas with high quality. In this\npaper, we propose a novel self-paced semi-curricular attention network, called\nSCANet, for non-homogeneous image dehazing that focuses on enhancing\nhaze-occluded regions. Our approach consists of an attention generator network\nand a scene reconstruction network. We use the luminance differences of images\nto restrict the attention map and introduce a self-paced semi-curricular\nlearning strategy to reduce learning ambiguity in the early stages of training.\nExtensive quantitative and qualitative experiments demonstrate that our SCANet\noutperforms many state-of-the-art methods. The code is publicly available at\nhttps://github.com/gy65896/SCANet.\n", "date": "2023-04-17", "categories": []}
{"id": "2212.07388", "title": "NoPe-NeRF: Optimising Neural Radiance Field with No Pose Prior", "abstract": "  Training a Neural Radiance Field (NeRF) without pre-computed camera poses is\nchallenging. Recent advances in this direction demonstrate the possibility of\njointly optimising a NeRF and camera poses in forward-facing scenes. However,\nthese methods still face difficulties during dramatic camera movement. We\ntackle this challenging problem by incorporating undistorted monocular depth\npriors. These priors are generated by correcting scale and shift parameters\nduring training, with which we are then able to constrain the relative poses\nbetween consecutive frames. This constraint is achieved using our proposed\nnovel loss functions. Experiments on real-world indoor and outdoor scenes show\nthat our method can handle challenging camera trajectories and outperforms\nexisting methods in terms of novel view rendering quality and pose estimation\naccuracy. Our project page is https://nope-nerf.active.vision.\n", "date": "2022-12-14", "categories": []}
{"id": "2304.06914", "title": "SMAE: Few-shot Learning for HDR Deghosting with Saturation-Aware Masked\n  Autoencoders", "abstract": "  Generating a high-quality High Dynamic Range (HDR) image from dynamic scenes\nhas recently been extensively studied by exploiting Deep Neural Networks\n(DNNs). Most DNNs-based methods require a large amount of training data with\nground truth, requiring tedious and time-consuming work. Few-shot HDR imaging\naims to generate satisfactory images with limited data. However, it is\ndifficult for modern DNNs to avoid overfitting when trained on only a few\nimages. In this work, we propose a novel semi-supervised approach to realize\nfew-shot HDR imaging via two stages of training, called SSHDR. Unlikely\nprevious methods, directly recovering content and removing ghosts\nsimultaneously, which is hard to achieve optimum, we first generate content of\nsaturated regions with a self-supervised mechanism and then address ghosts via\nan iterative semi-supervised learning framework. Concretely, considering that\nsaturated regions can be regarded as masking Low Dynamic Range (LDR) input\nregions, we design a Saturated Mask AutoEncoder (SMAE) to learn a robust\nfeature representation and reconstruct a non-saturated HDR image. We also\npropose an adaptive pseudo-label selection strategy to pick high-quality HDR\npseudo-labels in the second stage to avoid the effect of mislabeled samples.\nExperiments demonstrate that SSHDR outperforms state-of-the-art methods\nquantitatively and qualitatively within and across different datasets,\nachieving appealing HDR visualization with few labeled samples.\n", "date": "2023-04-14", "categories": []}
{"id": "2304.06955", "title": "Uncertainty-Aware Null Space Networks for Data-Consistent Image\n  Reconstruction", "abstract": "  Reconstructing an image from noisy and incomplete measurements is a central\ntask in several image processing applications. In recent years,\nstate-of-the-art reconstruction methods have been developed based on recent\nadvances in deep learning. Especially for highly underdetermined problems,\nmaintaining data consistency is a key goal. This can be achieved either by\niterative network architectures or by a subsequent projection of the network\nreconstruction. However, for such approaches to be used in safety-critical\ndomains such as medical imaging, the network reconstruction should not only\nprovide the user with a reconstructed image, but also with some level of\nconfidence in the reconstruction. In order to meet these two key requirements,\nthis paper combines deep null-space networks with uncertainty quantification.\nEvaluation of the proposed method includes image reconstruction from\nundersampled Radon measurements on a toy CT dataset and accelerated MRI\nreconstruction on the fastMRI dataset. This work is the first approach to\nsolving inverse problems that additionally models data-dependent uncertainty by\nestimating an input-dependent scale map, providing a robust assessment of\nreconstruction quality.\n", "date": "2023-04-14", "categories": []}
{"id": "2304.06976", "title": "Bitstream-Corrupted JPEG Images are Restorable: Two-stage Compensation\n  and Alignment Framework for Image Restoration", "abstract": "  In this paper, we study a real-world JPEG image restoration problem with bit\nerrors on the encrypted bitstream. The bit errors bring unpredictable color\ncasts and block shifts on decoded image contents, which cannot be resolved by\nexisting image restoration methods mainly relying on pre-defined degradation\nmodels in the pixel domain. To address these challenges, we propose a robust\nJPEG decoder, followed by a two-stage compensation and alignment framework to\nrestore bitstream-corrupted JPEG images. Specifically, the robust JPEG decoder\nadopts an error-resilient mechanism to decode the corrupted JPEG bitstream. The\ntwo-stage framework is composed of the self-compensation and alignment (SCA)\nstage and the guided-compensation and alignment (GCA) stage. The SCA adaptively\nperforms block-wise image color compensation and alignment based on the\nestimated color and block offsets via image content similarity. The GCA\nleverages the extracted low-resolution thumbnail from the JPEG header to guide\nfull-resolution pixel-wise image restoration in a coarse-to-fine manner. It is\nachieved by a coarse-guided pix2pix network and a refine-guided bi-directional\nLaplacian pyramid fusion network. We conduct experiments on three benchmarks\nwith varying degrees of bit error rates. Experimental results and ablation\nstudies demonstrate the superiority of our proposed method. The code will be\nreleased at https://github.com/wenyang001/Two-ACIR.\n", "date": "2023-04-14", "categories": []}
{"id": "2304.04773", "title": "HDR Video Reconstruction with a Large Dynamic Dataset in Raw and sRGB\n  Domains", "abstract": "  High dynamic range (HDR) video reconstruction is attracting more and more\nattention due to the superior visual quality compared with those of low dynamic\nrange (LDR) videos. The availability of LDR-HDR training pairs is essential for\nthe HDR reconstruction quality. However, there are still no real LDR-HDR pairs\nfor dynamic scenes due to the difficulty in capturing LDR-HDR frames\nsimultaneously. In this work, we propose to utilize a staggered sensor to\ncapture two alternate exposure images simultaneously, which are then fused into\nan HDR frame in both raw and sRGB domains. In this way, we build a large scale\nLDR-HDR video dataset with 85 scenes and each scene contains 60 frames. Based\non this dataset, we further propose a Raw-HDRNet, which utilizes the raw LDR\nframes as inputs. We propose a pyramid flow-guided deformation convolution to\nalign neighboring frames. Experimental results demonstrate that 1) the proposed\ndataset can improve the HDR reconstruction performance on real scenes for three\nbenchmark networks; 2) Compared with sRGB inputs, utilizing raw inputs can\nfurther improve the reconstruction quality and our proposed Raw-HDRNet is a\nstrong baseline for raw HDR reconstruction. Our dataset and code will be\nreleased after the acceptance of this paper.\n", "date": "2023-04-10", "categories": []}
{"id": "2304.06566", "title": "NeRD: Neural field-based Demosaicking", "abstract": "  We introduce NeRD, a new demosaicking method for generating full-color images\nfrom Bayer patterns. Our approach leverages advancements in neural fields to\nperform demosaicking by representing an image as a coordinate-based neural\nnetwork with sine activation functions. The inputs to the network are spatial\ncoordinates and a low-resolution Bayer pattern, while the outputs are the\ncorresponding RGB values. An encoder network, which is a blend of ResNet and\nU-net, enhances the implicit neural representation of the image to improve its\nquality and ensure spatial consistency through prior learning. Our experimental\nresults demonstrate that NeRD outperforms traditional and state-of-the-art\nCNN-based methods and significantly closes the gap to transformer-based\nmethods.\n", "date": "2023-04-13", "categories": []}
{"id": "2304.05564", "title": "Neural Invertible Variable-degree Optical Aberrations Correction", "abstract": "  Optical aberrations of optical systems cause significant degradation of\nimaging quality. Aberration correction by sophisticated lens designs and\nspecial glass materials generally incurs high cost of manufacturing and the\nincrease in the weight of optical systems, thus recent work has shifted to\naberration correction with deep learning-based post-processing. Though\nreal-world optical aberrations vary in degree, existing methods cannot\neliminate variable-degree aberrations well, especially for the severe degrees\nof degradation. Also, previous methods use a single feed-forward neural network\nand suffer from information loss in the output. To address the issues, we\npropose a novel aberration correction method with an invertible architecture by\nleveraging its information-lossless property. Within the architecture, we\ndevelop conditional invertible blocks to allow the processing of aberrations\nwith variable degrees. Our method is evaluated on both a synthetic dataset from\nphysics-based imaging simulation and a real captured dataset. Quantitative and\nqualitative experimental results demonstrate that our method outperforms\ncompared methods in correcting variable-degree optical aberrations.\n", "date": "2023-04-12", "categories": []}
{"id": "2301.06855", "title": "Event-based Shape from Polarization", "abstract": "  State-of-the-art solutions for Shape-from-Polarization (SfP) suffer from a\nspeed-resolution tradeoff: they either sacrifice the number of polarization\nangles measured or necessitate lengthy acquisition times due to framerate\nconstraints, thus compromising either accuracy or latency. We tackle this\ntradeoff using event cameras. Event cameras operate at microseconds resolution\nwith negligible motion blur, and output a continuous stream of events that\nprecisely measures how light changes over time asynchronously. We propose a\nsetup that consists of a linear polarizer rotating at high-speeds in front of\nan event camera. Our method uses the continuous event stream caused by the\nrotation to reconstruct relative intensities at multiple polarizer angles.\nExperiments demonstrate that our method outperforms physics-based baselines\nusing frames, reducing the MAE by 25% in synthetic and real-world dataset. In\nthe real world, we observe, however, that the challenging conditions (i.e.,\nwhen few events are generated) harm the performance of physics-based solutions.\nTo overcome this, we propose a learning-based approach that learns to estimate\nsurface normals even at low event-rates, improving the physics-based approach\nby 52% on the real world dataset. The proposed system achieves an acquisition\nspeed equivalent to 50 fps (>twice the framerate of the commercial polarization\nsensor) while retaining the spatial resolution of 1MP. Our evaluation is based\non the first large-scale dataset for event-based SfP\n", "date": "2023-01-17", "categories": []}
{"id": "2304.04848", "title": "Neural Lens Modeling", "abstract": "  Recent methods for 3D reconstruction and rendering increasingly benefit from\nend-to-end optimization of the entire image formation process. However, this\napproach is currently limited: effects of the optical hardware stack and in\nparticular lenses are hard to model in a unified way. This limits the quality\nthat can be achieved for camera calibration and the fidelity of the results of\n3D reconstruction. In this paper, we propose NeuroLens, a neural lens model for\ndistortion and vignetting that can be used for point projection and ray casting\nand can be optimized through both operations. This means that it can\n(optionally) be used to perform pre-capture calibration using classical\ncalibration targets, and can later be used to perform calibration or refinement\nduring 3D reconstruction, e.g., while optimizing a radiance field. To evaluate\nthe performance of our proposed model, we create a comprehensive dataset\nassembled from the Lensfun database with a multitude of lenses. Using this and\nother real-world datasets, we show that the quality of our proposed lens model\noutperforms standard packages as well as recent approaches while being much\neasier to use and extend. The model generalizes across many lens types and is\ntrivial to integrate into existing 3D reconstruction and rendering systems.\n", "date": "2023-04-10", "categories": []}
{"id": "2304.04917", "title": "Point-and-Shoot All-in-Focus Photo Synthesis from Smartphone Camera Pair", "abstract": "  All-in-Focus (AIF) photography is expected to be a commercial selling point\nfor modern smartphones. Standard AIF synthesis requires manual, time-consuming\noperations such as focal stack compositing, which is unfriendly to ordinary\npeople. To achieve point-and-shoot AIF photography with a smartphone, we expect\nthat an AIF photo can be generated from one shot of the scene, instead of from\nmultiple photos captured by the same camera. Benefiting from the multi-camera\nmodule in modern smartphones, we introduce a new task of AIF synthesis from\nmain (wide) and ultra-wide cameras. The goal is to recover sharp details from\ndefocused regions in the main-camera photo with the help of the\nultra-wide-camera one. The camera setting poses new challenges such as\nparallax-induced occlusions and inconsistent color between cameras. To overcome\nthe challenges, we introduce a predict-and-refine network to mitigate\nocclusions and propose dynamic frequency-domain alignment for color correction.\nTo enable effective training and evaluation, we also build an AIF dataset with\n2686 unique scenes. Each scene includes two photos captured by the main camera,\none photo captured by the ultrawide camera, and a synthesized AIF photo.\nResults show that our solution, termed EasyAIF, can produce high-quality AIF\nphotos and outperforms strong baselines quantitatively and qualitatively. For\nthe first time, we demonstrate point-and-shoot AIF photo synthesis successfully\nfrom main and ultra-wide cameras.\n", "date": "2023-04-11", "categories": []}
{"id": "2304.03994", "title": "RIDCP: Revitalizing Real Image Dehazing via High-Quality Codebook Priors", "abstract": "  Existing dehazing approaches struggle to process real-world hazy images owing\nto the lack of paired real data and robust priors. In this work, we present a\nnew paradigm for real image dehazing from the perspectives of synthesizing more\nrealistic hazy data and introducing more robust priors into the network.\nSpecifically, (1) instead of adopting the de facto physical scattering model,\nwe rethink the degradation of real hazy images and propose a phenomenological\npipeline considering diverse degradation types. (2) We propose a Real Image\nDehazing network via high-quality Codebook Priors (RIDCP). Firstly, a VQGAN is\npre-trained on a large-scale high-quality dataset to obtain the discrete\ncodebook, encapsulating high-quality priors (HQPs). After replacing the\nnegative effects brought by haze with HQPs, the decoder equipped with a novel\nnormalized feature alignment module can effectively utilize high-quality\nfeatures and produce clean results. However, although our degradation pipeline\ndrastically mitigates the domain gap between synthetic and real data, it is\nstill intractable to avoid it, which challenges HQPs matching in the wild.\nThus, we re-calculate the distance when matching the features to the HQPs by a\ncontrollable matching operation, which facilitates finding better counterparts.\nWe provide a recommendation to control the matching based on an explainable\nsolution. Users can also flexibly adjust the enhancement degree as per their\npreference. Extensive experiments verify the effectiveness of our data\nsynthesis pipeline and the superior performance of RIDCP in real image\ndehazing.\n", "date": "2023-04-08", "categories": []}
{"id": "2304.04297", "title": "AI-assisted Automated Workflow for Real-time X-ray Ptychography Data\n  Analysis via Federated Resources", "abstract": "  We present an end-to-end automated workflow that uses large-scale remote\ncompute resources and an embedded GPU platform at the edge to enable\nAI/ML-accelerated real-time analysis of data collected for x-ray ptychography.\nPtychography is a lensless method that is being used to image samples through a\nsimultaneous numerical inversion of a large number of diffraction patterns from\nadjacent overlapping scan positions. This acquisition method can enable\nnanoscale imaging with x-rays and electrons, but this often requires very large\nexperimental datasets and commensurately high turnaround times, which can limit\nexperimental capabilities such as real-time experimental steering and\nlow-latency monitoring. In this work, we introduce a software system that can\nautomate ptychography data analysis tasks. We accelerate the data analysis\npipeline by using a modified version of PtychoNN -- an ML-based approach to\nsolve phase retrieval problem that shows two orders of magnitude speedup\ncompared to traditional iterative methods. Further, our system coordinates and\noverlaps different data analysis tasks to minimize synchronization overhead\nbetween different stages of the workflow. We evaluate our workflow system with\nreal-world experimental workloads from the 26ID beamline at Advanced Photon\nSource and ThetaGPU cluster at Argonne Leadership Computing Resources.\n", "date": "2023-04-09", "categories": []}
{"id": "2304.04464", "title": "Exposure Fusion for Hand-held Camera Inputs with Optical Flow and\n  PatchMatch", "abstract": "  This paper proposes a hybrid synthesis method for multi-exposure image fusion\ntaken by hand-held cameras. Motions either due to the shaky camera or caused by\ndynamic scenes should be compensated before any content fusion. Any\nmisalignment can easily cause blurring/ghosting artifacts in the fused result.\nOur hybrid method can deal with such motions and maintain the exposure\ninformation of each input effectively. In particular, the proposed method first\napplies optical flow for a coarse registration, which performs well with\ncomplex non-rigid motion but produces deformations at regions with missing\ncorrespondences. The absence of correspondences is due to the occlusions of\nscene parallax or the moving contents. To correct such error registration, we\nsegment images into superpixels and identify problematic alignments based on\neach superpixel, which is further aligned by PatchMatch. The method combines\nthe efficiency of optical flow and the accuracy of PatchMatch. After PatchMatch\ncorrection, we obtain a fully aligned image stack that facilitates a\nhigh-quality fusion that is free from blurring/ghosting artifacts. We compare\nour method with existing fusion algorithms on various challenging examples,\nincluding the static/dynamic, the indoor/outdoor and the daytime/nighttime\nscenes. Experiment results demonstrate the effectiveness and robustness of our\nmethod.\n", "date": "2023-04-10", "categories": []}
{"id": "2212.05217", "title": "Snapshot Multispectral Imaging Using a Diffractive Optical Network", "abstract": "  Multispectral imaging has been used for numerous applications in e.g.,\nenvironmental monitoring, aerospace, defense, and biomedicine. Here, we present\na diffractive optical network-based multispectral imaging system trained using\ndeep learning to create a virtual spectral filter array at the output image\nfield-of-view. This diffractive multispectral imager performs\nspatially-coherent imaging over a large spectrum, and at the same time, routes\na pre-determined set of spectral channels onto an array of pixels at the output\nplane, converting a monochrome focal plane array or image sensor into a\nmultispectral imaging device without any spectral filters or image recovery\nalgorithms. Furthermore, the spectral responsivity of this diffractive\nmultispectral imager is not sensitive to input polarization states. Through\nnumerical simulations, we present different diffractive network designs that\nachieve snapshot multispectral imaging with 4, 9 and 16 unique spectral bands\nwithin the visible spectrum, based on passive spatially-structured diffractive\nsurfaces, with a compact design that axially spans ~72 times the mean\nwavelength of the spectral band of interest. Moreover, we experimentally\ndemonstrate a diffractive multispectral imager based on a 3D-printed\ndiffractive network that creates at its output image plane a\nspatially-repeating virtual spectral filter array with 2x2=4 unique bands at\nterahertz spectrum. Due to their compact form factor and computation-free,\npower-efficient and polarization-insensitive forward operation, diffractive\nmultispectral imagers can be transformative for various imaging and sensing\napplications and be used at different parts of the electromagnetic spectrum\nwhere high-density and wide-area multispectral pixel arrays are not widely\navailable.\n", "date": "2022-12-10", "categories": []}
{"id": "2304.03285", "title": "$\\text{DC}^2$: Dual-Camera Defocus Control by Learning to Refocus", "abstract": "  Smartphone cameras today are increasingly approaching the versatility and\nquality of professional cameras through a combination of hardware and software\nadvancements. However, fixed aperture remains a key limitation, preventing\nusers from controlling the depth of field (DoF) of captured images. At the same\ntime, many smartphones now have multiple cameras with different fixed apertures\n-- specifically, an ultra-wide camera with wider field of view and deeper DoF\nand a higher resolution primary camera with shallower DoF. In this work, we\npropose $\\text{DC}^2$, a system for defocus control for synthetically varying\ncamera aperture, focus distance and arbitrary defocus effects by fusing\ninformation from such a dual-camera system. Our key insight is to leverage\nreal-world smartphone camera dataset by using image refocus as a proxy task for\nlearning to control defocus. Quantitative and qualitative evaluations on\nreal-world data demonstrate our system's efficacy where we outperform\nstate-of-the-art on defocus deblurring, bokeh rendering, and image refocus.\nFinally, we demonstrate creative post-capture defocus control enabled by our\nmethod, including tilt-shift and content-based defocus effects.\n", "date": "2023-04-06", "categories": []}
{"id": "2303.16205", "title": "mHealth hyperspectral learning for instantaneous spatiospectral imaging\n  of hemodynamics", "abstract": "  Hyperspectral imaging acquires data in both the spatial and frequency domains\nto offer abundant physical or biological information. However, conventional\nhyperspectral imaging has intrinsic limitations of bulky instruments, slow data\nacquisition rate, and spatiospectral tradeoff. Here we introduce hyperspectral\nlearning for snapshot hyperspectral imaging in which sampled hyperspectral data\nin a small subarea are incorporated into a learning algorithm to recover the\nhypercube. Hyperspectral learning exploits the idea that a photograph is more\nthan merely a picture and contains detailed spectral information. A small\nsampling of hyperspectral data enables spectrally informed learning to recover\na hypercube from an RGB image. Hyperspectral learning is capable of recovering\nfull spectroscopic resolution in the hypercube, comparable to high spectral\nresolutions of scientific spectrometers. Hyperspectral learning also enables\nultrafast dynamic imaging, leveraging ultraslow video recording in an\noff-the-shelf smartphone, given that a video comprises a time series of\nmultiple RGB images. To demonstrate its versatility, an experimental model of\nvascular development is used to extract hemodynamic parameters via statistical\nand deep-learning approaches. Subsequently, the hemodynamics of peripheral\nmicrocirculation is assessed at an ultrafast temporal resolution up to a\nmillisecond, using a conventional smartphone camera. This spectrally informed\nlearning method is analogous to compressed sensing; however, it further allows\nfor reliable hypercube recovery and key feature extractions with a transparent\nlearning algorithm. This learning-powered snapshot hyperspectral imaging method\nyields high spectral and temporal resolutions and eliminates the spatiospectral\ntradeoff, offering simple hardware requirements and potential applications of\nvarious machine-learning techniques.\n", "date": "2023-03-27", "categories": []}
{"id": "2304.01686", "title": "HyperCUT: Video Sequence from a Single Blurry Image using Unsupervised\n  Ordering", "abstract": "  We consider the challenging task of training models for image-to-video\ndeblurring, which aims to recover a sequence of sharp images corresponding to a\ngiven blurry image input. A critical issue disturbing the training of an\nimage-to-video model is the ambiguity of the frame ordering since both the\nforward and backward sequences are plausible solutions. This paper proposes an\neffective self-supervised ordering scheme that allows training high-quality\nimage-to-video deblurring models. Unlike previous methods that rely on\norder-invariant losses, we assign an explicit order for each video sequence,\nthus avoiding the order-ambiguity issue. Specifically, we map each video\nsequence to a vector in a latent high-dimensional space so that there exists a\nhyperplane such that for every video sequence, the vectors extracted from it\nand its reversed sequence are on different sides of the hyperplane. The side of\nthe vectors will be used to define the order of the corresponding sequence.\nLast but not least, we propose a real-image dataset for the image-to-video\ndeblurring problem that covers a variety of popular domains, including face,\nhand, and street. Extensive experimental results confirm the effectiveness of\nour method. Code and data are available at\nhttps://github.com/VinAIResearch/HyperCUT.git\n", "date": "2023-04-04", "categories": []}
{"id": "2304.01308", "title": "Role of Transients in Two-Bounce Non-Line-of-Sight Imaging", "abstract": "  The goal of non-line-of-sight (NLOS) imaging is to image objects occluded\nfrom the camera's field of view using multiply scattered light. Recent works\nhave demonstrated the feasibility of two-bounce (2B) NLOS imaging by scanning a\nlaser and measuring cast shadows of occluded objects in scenes with two relay\nsurfaces. In this work, we study the role of time-of-flight (ToF) measurements,\n\\ie transients, in 2B-NLOS under multiplexed illumination. Specifically, we\nstudy how ToF information can reduce the number of measurements and spatial\nresolution needed for shape reconstruction. We present our findings with\nrespect to tradeoffs in (1) temporal resolution, (2) spatial resolution, and\n(3) number of image captures by studying SNR and recoverability as functions of\nsystem parameters. This leads to a formal definition of the mathematical\nconstraints for 2B lidar. We believe that our work lays an analytical\ngroundwork for design of future NLOS imaging systems, especially as ToF sensors\nbecome increasingly ubiquitous.\n", "date": "2023-04-03", "categories": []}
{"id": "2304.01351", "title": "Accelerated parallel MRI using memory efficient and robust monotone\n  operator learning (MOL)", "abstract": "  Model-based deep learning methods that combine imaging physics with learned\nregularization priors have been emerging as powerful tools for parallel MRI\nacceleration. The main focus of this paper is to determine the utility of the\nmonotone operator learning (MOL) framework in the parallel MRI setting. The MOL\nalgorithm alternates between a gradient descent step using a monotone\nconvolutional neural network (CNN) and a conjugate gradient algorithm to\nencourage data consistency. The benefits of this approach include similar\nguarantees as compressive sensing algorithms including uniqueness, convergence,\nand stability, while being significantly more memory efficient than unrolled\nmethods. We validate the proposed scheme by comparing it with different\nunrolled algorithms in the context of accelerated parallel MRI for static and\ndynamic settings.\n", "date": "2023-04-03", "categories": []}
{"id": "2304.00534", "title": "LG-BPN: Local and Global Blind-Patch Network for Self-Supervised\n  Real-World Denoising", "abstract": "  Despite the significant results on synthetic noise under simplified\nassumptions, most self-supervised denoising methods fail under real noise due\nto the strong spatial noise correlation, including the advanced self-supervised\nblind-spot networks (BSNs). For recent methods targeting real-world denoising,\nthey either suffer from ignoring this spatial correlation, or are limited by\nthe destruction of fine textures for under-considering the correlation. In this\npaper, we present a novel method called LG-BPN for self-supervised real-world\ndenoising, which takes the spatial correlation statistic into our network\ndesign for local detail restoration, and also brings the long-range\ndependencies modeling ability to previously CNN-based BSN methods. First, based\non the correlation statistic, we propose a densely-sampled patch-masked\nconvolution module. By taking more neighbor pixels with low noise correlation\ninto account, we enable a denser local receptive field, preserving more useful\ninformation for enhanced fine structure recovery. Second, we propose a dilated\nTransformer block to allow distant context exploitation in BSN. This global\nperception addresses the intrinsic deficiency of BSN, whose receptive field is\nconstrained by the blind spot requirement, which can not be fully resolved by\nthe previous CNN-based BSNs. These two designs enable LG-BPN to fully exploit\nboth the detailed structure and the global interaction in a blind manner.\nExtensive results on real-world datasets demonstrate the superior performance\nof our method. https://github.com/Wang-XIaoDingdd/LGBPN\n", "date": "2023-04-02", "categories": []}
{"id": "2304.00837", "title": "Disorder-invariant Implicit Neural Representation", "abstract": "  Implicit neural representation (INR) characterizes the attributes of a signal\nas a function of corresponding coordinates which emerges as a sharp weapon for\nsolving inverse problems. However, the expressive power of INR is limited by\nthe spectral bias in the network training. In this paper, we find that such a\nfrequency-related problem could be greatly solved by re-arranging the\ncoordinates of the input signal, for which we propose the disorder-invariant\nimplicit neural representation (DINER) by augmenting a hash-table to a\ntraditional INR backbone. Given discrete signals sharing the same histogram of\nattributes and different arrangement orders, the hash-table could project the\ncoordinates into the same distribution for which the mapped signal can be\nbetter modeled using the subsequent INR network, leading to significantly\nalleviated spectral bias. Furthermore, the expressive power of the DINER is\ndetermined by the width of the hash-table. Different width corresponds to\ndifferent geometrical elements in the attribute space, \\textit{e.g.}, 1D curve,\n2D curved-plane and 3D curved-volume when the width is set as $1$, $2$ and $3$,\nrespectively. More covered areas of the geometrical elements result in stronger\nexpressive power. Experiments not only reveal the generalization of the DINER\nfor different INR backbones (MLP vs. SIREN) and various tasks (image/video\nrepresentation, phase retrieval, refractive index recovery, and neural radiance\nfield optimization) but also show the superiority over the state-of-the-art\nalgorithms both in quality and speed. \\textit{Project page:}\n\\url{https://ezio77.github.io/DINER-website/}\n", "date": "2023-04-03", "categories": []}
{"id": "2212.05936", "title": "Encoder-Decoder Network with Guided Transmission Map: Architecture", "abstract": "  An insight into the architecture of the Encoder-Decoder Network with Guided\nTransmission Map (EDN-GTM), a novel and effective single image dehazing scheme,\nis presented in this paper. The EDN-GTM takes a conventional RGB hazy image in\nconjunction with the corresponding transmission map estimated by the dark\nchannel prior (DCP) approach as inputs of the network. The EDN-GTM adopts an\nenhanced structure of U-Net developed for dehazing tasks and the resulting\nEDN-GDM has shown state-of-the-art performances on benchmark dehazing datasets\nin terms of PSNR and SSIM metrics. In order to give an in-depth understanding\nof the well-designed architecture which largely contributes to the success of\nthe EDN-GTM, extensive experiments and analysis from selecting the core\nstructure of the scheme to investigating advanced network designs are presented\nin this paper.\n", "date": "2022-12-07", "categories": []}
{"id": "2303.16181", "title": "Learning Federated Visual Prompt in Null Space for MRI Reconstruction", "abstract": "  Federated Magnetic Resonance Imaging (MRI) reconstruction enables multiple\nhospitals to collaborate distributedly without aggregating local data, thereby\nprotecting patient privacy. However, the data heterogeneity caused by different\nMRI protocols, insufficient local training data, and limited communication\nbandwidth inevitably impair global model convergence and updating. In this\npaper, we propose a new algorithm, FedPR, to learn federated visual prompts in\nthe null space of global prompt for MRI reconstruction. FedPR is a new\nfederated paradigm that adopts a powerful pre-trained model while only learning\nand communicating the prompts with few learnable parameters, thereby\nsignificantly reducing communication costs and achieving competitive\nperformance on limited local data. Moreover, to deal with catastrophic\nforgetting caused by data heterogeneity, FedPR also updates efficient federated\nvisual prompts that project the local prompts into an approximate null space of\nthe global prompt, thereby suppressing the interference of gradients on the\nserver performance. Extensive experiments on federated MRI show that FedPR\nsignificantly outperforms state-of-the-art FL algorithms with <6% of\ncommunication costs when given the limited amount of local training data.\n", "date": "2023-03-28", "categories": []}
{"id": "2303.17583", "title": "TiDy-PSFs: Computational Imaging with Time-Averaged Dynamic\n  Point-Spread-Functions", "abstract": "  Point-spread-function (PSF) engineering is a powerful computational imaging\ntechniques wherein a custom phase mask is integrated into an optical system to\nencode additional information into captured images. Used in combination with\ndeep learning, such systems now offer state-of-the-art performance at monocular\ndepth estimation, extended depth-of-field imaging, lensless imaging, and other\ntasks. Inspired by recent advances in spatial light modulator (SLM) technology,\nthis paper answers a natural question: Can one encode additional information\nand achieve superior performance by changing a phase mask dynamically over\ntime? We first prove that the set of PSFs described by static phase masks is\nnon-convex and that, as a result, time-averaged PSFs generated by dynamic phase\nmasks are fundamentally more expressive. We then demonstrate, in simulation,\nthat time-averaged dynamic (TiDy) phase masks can offer substantially improved\nmonocular depth estimation and extended depth-of-field imaging performance.\n", "date": "2023-03-30", "categories": []}
{"id": "2211.10732", "title": "Passive Micron-scale Time-of-Flight with Sunlight Interferometry", "abstract": "  We introduce an interferometric technique for passive time-of-flight imaging\nand depth sensing at micrometer axial resolutions. Our technique uses a\nfull-field Michelson interferometer, modified to use sunlight as the only light\nsource. The large spectral bandwidth of sunlight makes it possible to acquire\nmicrometer-resolution time-resolved scene responses, through a simple axial\nscanning operation. Additionally, the angular bandwidth of sunlight makes it\npossible to capture time-of-flight measurements insensitive to indirect\nillumination effects, such as interreflections and subsurface scattering. We\nbuild an experimental prototype that we operate outdoors, under direct\nsunlight, and in adverse environment conditions such as machine vibrations and\nvehicle traffic. We use this prototype to demonstrate, for the first time,\npassive imaging capabilities such as micrometer-scale depth sensing robust to\nindirect illumination, direct-only imaging, and imaging through diffusers.\n", "date": "2022-11-19", "categories": []}
{"id": "2212.04441", "title": "The Differentiable Lens: Compound Lens Search over Glass Surfaces and\n  Materials for Object Detection", "abstract": "  Most camera lens systems are designed in isolation, separately from\ndownstream computer vision methods. Recently, joint optimization approaches\nthat design lenses alongside other components of the image acquisition and\nprocessing pipeline -- notably, downstream neural networks -- have achieved\nimproved imaging quality or better performance on vision tasks. However, these\nexisting methods optimize only a subset of lens parameters and cannot optimize\nglass materials given their categorical nature. In this work, we develop a\ndifferentiable spherical lens simulation model that accurately captures\ngeometrical aberrations. We propose an optimization strategy to address the\nchallenges of lens design -- notorious for non-convex loss function landscapes\nand many manufacturing constraints -- that are exacerbated in joint\noptimization tasks. Specifically, we introduce quantized continuous glass\nvariables to facilitate the optimization and selection of glass materials in an\nend-to-end design context, and couple this with carefully designed constraints\nto support manufacturability. In automotive object detection, we report\nimproved detection performance over existing designs even when simplifying\ndesigns to two- or three-element lenses, despite significantly degrading the\nimage quality.\n", "date": "2022-12-08", "categories": []}
{"id": "2303.15101", "title": "DANI-Net: Uncalibrated Photometric Stereo by Differentiable Shadow\n  Handling, Anisotropic Reflectance Modeling, and Neural Inverse Rendering", "abstract": "  Uncalibrated photometric stereo (UPS) is challenging due to the inherent\nambiguity brought by the unknown light. Although the ambiguity is alleviated on\nnon-Lambertian objects, the problem is still difficult to solve for more\ngeneral objects with complex shapes introducing irregular shadows and general\nmaterials with complex reflectance like anisotropic reflectance. To exploit\ncues from shadow and reflectance to solve UPS and improve performance on\ngeneral materials, we propose DANI-Net, an inverse rendering framework with\ndifferentiable shadow handling and anisotropic reflectance modeling. Unlike\nmost previous methods that use non-differentiable shadow maps and assume\nisotropic material, our network benefits from cues of shadow and anisotropic\nreflectance through two differentiable paths. Experiments on multiple\nreal-world datasets demonstrate our superior and robust performance.\n", "date": "2023-03-27", "categories": []}
{"id": "2303.15724", "title": "Scalable, Detailed and Mask-Free Universal Photometric Stereo", "abstract": "  In this paper, we introduce SDM-UniPS, a groundbreaking Scalable, Detailed,\nMask-free, and Universal Photometric Stereo network. Our approach can recover\nastonishingly intricate surface normal maps, rivaling the quality of 3D\nscanners, even when images are captured under unknown, spatially-varying\nlighting conditions in uncontrolled environments. We have extended previous\nuniversal photometric stereo networks to extract spatial-light features,\nutilizing all available information in high-resolution input images and\naccounting for non-local interactions among surface points. Moreover, we\npresent a new synthetic training dataset that encompasses a diverse range of\nshapes, materials, and illumination scenarios found in real-world scenes.\nThrough extensive evaluation, we demonstrate that our method not only surpasses\ncalibrated, lighting-specific techniques on public benchmarks, but also excels\nwith a significantly smaller number of input images even without object masks.\n", "date": "2023-03-28", "categories": []}
{"id": "2211.12285", "title": "Exact-NeRF: An Exploration of a Precise Volumetric Parameterization for\n  Neural Radiance Fields", "abstract": "  Neural Radiance Fields (NeRF) have attracted significant attention due to\ntheir ability to synthesize novel scene views with great accuracy. However,\ninherent to their underlying formulation, the sampling of points along a ray\nwith zero width may result in ambiguous representations that lead to further\nrendering artifacts such as aliasing in the final scene. To address this issue,\nthe recent variant mip-NeRF proposes an Integrated Positional Encoding (IPE)\nbased on a conical view frustum. Although this is expressed with an integral\nformulation, mip-NeRF instead approximates this integral as the expected value\nof a multivariate Gaussian distribution. This approximation is reliable for\nshort frustums but degrades with highly elongated regions, which arises when\ndealing with distant scene objects under a larger depth of field. In this\npaper, we explore the use of an exact approach for calculating the IPE by using\na pyramid-based integral formulation instead of an approximated conical-based\none. We denote this formulation as Exact-NeRF and contribute the first approach\nto offer a precise analytical solution to the IPE within the NeRF domain. Our\nexploratory work illustrates that such an exact formulation Exact-NeRF matches\nthe accuracy of mip-NeRF and furthermore provides a natural extension to more\nchallenging scenarios without further modification, such as in the case of\nunbounded scenes. Our contribution aims to both address the hitherto unexplored\nissues of frustum approximation in earlier NeRF work and additionally provide\ninsight into the potential future consideration of analytical solutions in\nfuture NeRF extensions.\n", "date": "2022-11-22", "categories": []}
{"id": "2301.10241", "title": "K-Planes: Explicit Radiance Fields in Space, Time, and Appearance", "abstract": "  We introduce k-planes, a white-box model for radiance fields in arbitrary\ndimensions. Our model uses d choose 2 planes to represent a d-dimensional\nscene, providing a seamless way to go from static (d=3) to dynamic (d=4)\nscenes. This planar factorization makes adding dimension-specific priors easy,\ne.g. temporal smoothness and multi-resolution spatial structure, and induces a\nnatural decomposition of static and dynamic components of a scene. We use a\nlinear feature decoder with a learned color basis that yields similar\nperformance as a nonlinear black-box MLP decoder. Across a range of synthetic\nand real, static and dynamic, fixed and varying appearance scenes, k-planes\nyields competitive and often state-of-the-art reconstruction fidelity with low\nmemory usage, achieving 1000x compression over a full 4D grid, and fast\noptimization with a pure PyTorch implementation. For video results and code,\nplease see https://sarafridov.github.io/K-Planes.\n", "date": "2023-01-24", "categories": []}
{"id": "2303.10752", "title": "Fully Self-Supervised Depth Estimation from Defocus Clue", "abstract": "  Depth-from-defocus (DFD), modeling the relationship between depth and defocus\npattern in images, has demonstrated promising performance in depth estimation.\nRecently, several self-supervised works try to overcome the difficulties in\nacquiring accurate depth ground-truth. However, they depend on the all-in-focus\n(AIF) images, which cannot be captured in real-world scenarios. Such limitation\ndiscourages the applications of DFD methods. To tackle this issue, we propose a\ncompletely self-supervised framework that estimates depth purely from a sparse\nfocal stack. We show that our framework circumvents the needs for the depth and\nAIF image ground-truth, and receives superior predictions, thus closing the gap\nbetween the theoretical success of DFD works and their applications in the real\nworld. In particular, we propose (i) a more realistic setting for DFD tasks,\nwhere no depth or AIF image ground-truth is available; (ii) a novel\nself-supervision framework that provides reliable predictions of depth and AIF\nimage under the challenging setting. The proposed framework uses a neural model\nto predict the depth and AIF image, and utilizes an optical model to validate\nand refine the prediction. We verify our framework on three benchmark datasets\nwith rendered focal stacks and real focal stacks. Qualitative and quantitative\nevaluations show that our method provides a strong baseline for self-supervised\nDFD tasks.\n", "date": "2023-03-19", "categories": []}
{"id": "2303.14478", "title": "DBARF: Deep Bundle-Adjusting Generalizable Neural Radiance Fields", "abstract": "  Recent works such as BARF and GARF can bundle adjust camera poses with neural\nradiance fields (NeRF) which is based on coordinate-MLPs. Despite the\nimpressive results, these methods cannot be applied to Generalizable NeRFs\n(GeNeRFs) which require image feature extractions that are often based on more\ncomplicated 3D CNN or transformer architectures. In this work, we first analyze\nthe difficulties of jointly optimizing camera poses with GeNeRFs, and then\nfurther propose our DBARF to tackle these issues. Our DBARF which bundle\nadjusts camera poses by taking a cost feature map as an implicit cost function\ncan be jointly trained with GeNeRFs in a self-supervised manner. Unlike BARF\nand its follow-up works, which can only be applied to per-scene optimized NeRFs\nand need accurate initial camera poses with the exception of forward-facing\nscenes, our method can generalize across scenes and does not require any good\ninitialization. Experiments show the effectiveness and generalization ability\nof our DBARF when evaluated on real-world datasets. Our code is available at\n\\url{https://aibluefisher.github.io/dbarf}.\n", "date": "2023-03-25", "categories": []}
{"id": "2303.15046", "title": "Nighttime Smartphone Reflective Flare Removal Using Optical Center\n  Symmetry Prior", "abstract": "  Reflective flare is a phenomenon that occurs when light reflects inside\nlenses, causing bright spots or a \"ghosting effect\" in photos, which can impact\ntheir quality. Eliminating reflective flare is highly desirable but\nchallenging. Many existing methods rely on manually designed features to detect\nthese bright spots, but they often fail to identify reflective flares created\nby various types of light and may even mistakenly remove the light sources in\nscenarios with multiple light sources. To address these challenges, we propose\nan optical center symmetry prior, which suggests that the reflective flare and\nlight source are always symmetrical around the lens's optical center. This\nprior helps to locate the reflective flare's proposal region more accurately\nand can be applied to most smartphone cameras. Building on this prior, we\ncreate the first reflective flare removal dataset called BracketFlare, which\ncontains diverse and realistic reflective flare patterns. We use continuous\nbracketing to capture the reflective flare pattern in the underexposed image\nand combine it with a normally exposed image to synthesize a pair of\nflare-corrupted and flare-free images. With the dataset, neural networks can be\ntrained to remove the reflective flares effectively. Extensive experiments\ndemonstrate the effectiveness of our method on both synthetic and real-world\ndatasets.\n", "date": "2023-03-27", "categories": []}
{"id": "2302.10927", "title": "Spatial gradient consistency for unsupervised learning of hyperspectral\n  demosaicking: Application to surgical imaging", "abstract": "  Hyperspectral imaging has the potential to improve intraoperative decision\nmaking if tissue characterisation is performed in real-time and with\nhigh-resolution. Hyperspectral snapshot mosaic sensors offer a promising\napproach due to their fast acquisition speed and compact size. However, a\ndemosaicking algorithm is required to fully recover the spatial and spectral\ninformation of the snapshot images. Most state-of-the-art demosaicking\nalgorithms require ground-truth training data with paired snapshot and\nhigh-resolution hyperspectral images, but such imagery pairs with the exact\nsame scene are physically impossible to acquire in intraoperative settings. In\nthis work, we present a fully unsupervised hyperspectral image demosaicking\nalgorithm which only requires exemplar snapshot images for training purposes.\nWe regard hyperspectral demosaicking as an ill-posed linear inverse problem\nwhich we solve using a deep neural network. We take advantage of the spectral\ncorrelation occurring in natural scenes to design a novel inter spectral band\nregularisation term based on spatial gradient consistency. By combining our\nproposed term with standard regularisation techniques and exploiting a standard\ndata fidelity term, we obtain an unsupervised loss function for training deep\nneural networks, which allows us to achieve real-time hyperspectral image\ndemosaicking. Quantitative results on hyperspetral image datasets show that our\nunsupervised demosaicking approach can achieve similar performance to its\nsupervised counter-part, and significantly outperform linear demosaicking. A\nqualitative user study on real snapshot hyperspectral surgical images confirms\nthe results from the quantitative analysis. Our results suggest that the\nproposed unsupervised algorithm can achieve promising hyperspectral\ndemosaicking in real-time thus advancing the suitability of the modality for\nintraoperative use.\n", "date": "2023-02-21", "categories": []}
{"id": "2303.09071", "title": "Joint Multi-Scale Tone Mapping and Denoising for HDR Image Enhancement", "abstract": "  An image processing unit (IPU), or image signal processor (ISP) for high\ndynamic range (HDR) imaging usually consists of demosaicing, white balancing,\nlens shading correction, color correction, denoising, and tone-mapping. Besides\nnoise from the imaging sensors, almost every step in the ISP introduces or\namplifies noise in different ways, and denoising operators are designed to\nreduce the noise from these sources. Designed for dynamic range compressing,\ntone-mapping operators in an ISP can significantly amplify the noise level,\nespecially for images captured in low-light conditions, making denoising very\ndifficult. Therefore, we propose a joint multi-scale denoising and tone-mapping\nframework that is designed with both operations in mind for HDR images. Our\njoint network is trained in an end-to-end format that optimizes both operators\ntogether, to prevent the tone-mapping operator from overwhelming the denoising\noperator. Our model outperforms existing HDR denoising and tone-mapping\noperators both quantitatively and qualitatively on most of our benchmarking\ndatasets.\n", "date": "2023-03-16", "categories": []}
{"id": "2303.13571", "title": "Inheriting Bayer's Legacy-Joint Remosaicing and Denoising for Quad Bayer\n  Image Sensor", "abstract": "  Pixel binning based Quad sensors have emerged as a promising solution to\novercome the hardware limitations of compact cameras in low-light imaging.\nHowever, binning results in lower spatial resolution and non-Bayer CFA\nartifacts. To address these challenges, we propose a dual-head joint\nremosaicing and denoising network (DJRD), which enables the conversion of noisy\nQuad Bayer and standard noise-free Bayer pattern without any resolution loss.\nDJRD includes a newly designed Quad Bayer remosaicing (QB-Re) block, integrated\ndenoising modules based on Swin-transformer and multi-scale wavelet transform.\nThe QB-Re block constructs the convolution kernel based on the CFA pattern to\nachieve a periodic color distribution in the perceptual field, which is used to\nextract exact spectral information and reduce color misalignment. The\nintegrated Swin-Transformer and multi-scale wavelet transform capture non-local\ndependencies, frequency and location information to effectively reduce\npractical noise. By identifying challenging patches utilizing Moire and zipper\ndetection metrics, we enable our model to concentrate on difficult patches\nduring the post-training phase, which enhances the model's performance in hard\ncases. Our proposed model outperforms competing models by approximately 3dB,\nwithout additional complexity in hardware or software.\n", "date": "2023-03-23", "categories": []}
{"id": "2303.13916", "title": "Self-Supervised Reversed Image Signal Processing via Reference-Guided\n  Dynamic Parameter Selection", "abstract": "  Unprocessed sensor outputs (RAW images) potentially improve both low-level\nand high-level computer vision algorithms, but the lack of large-scale RAW\nimage datasets is a barrier to research. Thus, reversed Image Signal Processing\n(ISP) which converts existing RGB images into RAW images has been studied.\nHowever, most existing methods require camera-specific metadata or paired RGB\nand RAW images to model the conversion, and they are not always available. In\naddition, there are issues in handling diverse ISPs and recovering global\nillumination. To tackle these limitations, we propose a self-supervised\nreversed ISP method that does not require metadata and paired images. The\nproposed method converts a RGB image into a RAW-like image taken in the same\nenvironment with the same sensor as a reference RAW image by dynamically\nselecting parameters of the reversed ISP pipeline based on the reference RAW\nimage. The parameter selection is trained via pseudo paired data created from\nunpaired RGB and RAW images. We show that the proposed method is able to learn\nvarious reversed ISPs with comparable accuracy to other state-of-the-art\nsupervised methods and convert unknown RGB images from COCO and Flickr1M to\ntarget RAW-like images more accurately in terms of pixel distribution. We also\ndemonstrate that our generated RAW images improve performance on real RAW image\nobject detection task.\n", "date": "2023-03-24", "categories": []}
{"id": "2303.07150", "title": "Multi PILOT: Learned Feasible Multiple Acquisition Trajectories for\n  Dynamic MRI", "abstract": "  Dynamic Magnetic Resonance Imaging (MRI) is known to be a powerful and\nreliable technique for the dynamic imaging of internal organs and tissues,\nmaking it a leading diagnostic tool. A major difficulty in using MRI in this\nsetting is the relatively long acquisition time (and, hence, increased cost)\nrequired for imaging in high spatio-temporal resolution, leading to the\nappearance of related motion artifacts and decrease in resolution. Compressed\nSensing (CS) techniques have become a common tool to reduce MRI acquisition\ntime by subsampling images in the k-space according to some acquisition\ntrajectory. Several studies have particularly focused on applying deep learning\ntechniques to learn these acquisition trajectories in order to attain better\nimage reconstruction, rather than using some predefined set of trajectories. To\nthe best of our knowledge, learning acquisition trajectories has been only\nexplored in the context of static MRI. In this study, we consider acquisition\ntrajectory learning in the dynamic imaging setting. We design an end-to-end\npipeline for the joint optimization of multiple per-frame acquisition\ntrajectories along with a reconstruction neural network, and demonstrate\nimproved image reconstruction quality in shorter acquisition times. The code\nfor reproducing all experiments is accessible at\nhttps://github.com/tamirshor7/MultiPILOT.\n", "date": "2023-03-13", "categories": []}
{"id": "2303.13132", "title": "Masked Image Training for Generalizable Deep Image Denoising", "abstract": "  When capturing and storing images, devices inevitably introduce noise.\nReducing this noise is a critical task called image denoising. Deep learning\nhas become the de facto method for image denoising, especially with the\nemergence of Transformer-based models that have achieved notable\nstate-of-the-art results on various image tasks. However, deep learning-based\nmethods often suffer from a lack of generalization ability. For example, deep\nmodels trained on Gaussian noise may perform poorly when tested on other noise\ndistributions. To address this issue, we present a novel approach to enhance\nthe generalization performance of denoising networks, known as masked training.\nOur method involves masking random pixels of the input image and reconstructing\nthe missing information during training. We also mask out the features in the\nself-attention layers to avoid the impact of training-testing inconsistency.\nOur approach exhibits better generalization ability than other deep learning\nmodels and is directly applicable to real-world scenarios. Additionally, our\ninterpretability analysis demonstrates the superiority of our method.\n", "date": "2023-03-23", "categories": []}
{"id": "2303.13404", "title": "MSFA-Frequency-Aware Transformer for Hyperspectral Images Demosaicing", "abstract": "  Hyperspectral imaging systems that use multispectral filter arrays (MSFA)\ncapture only one spectral component in each pixel. Hyperspectral demosaicing is\nused to recover the non-measured components. While deep learning methods have\nshown promise in this area, they still suffer from several challenges,\nincluding limited modeling of non-local dependencies, lack of consideration of\nthe periodic MSFA pattern that could be linked to periodic artifacts, and\ndifficulty in recovering high-frequency details. To address these challenges,\nthis paper proposes a novel de-mosaicing framework, the MSFA-frequency-aware\nTransformer network (FDM-Net). FDM-Net integrates a novel MSFA-frequency-aware\nmulti-head self-attention mechanism (MaFormer) and a filter-based Fourier\nzero-padding method to reconstruct high pass components with greater difficulty\nand low pass components with relative ease, separately. The advantage of\nMaformer is that it can leverage the MSFA information and non-local\ndependencies present in the data. Additionally, we introduce a joint spatial\nand frequency loss to transfer MSFA information and enhance training on\nfrequency components that are hard to recover. Our experimental results\ndemonstrate that FDM-Net outperforms state-of-the-art methods with 6dB PSNR,\nand reconstructs high-fidelity details successfully.\n", "date": "2023-03-23", "categories": []}
{"id": "2303.12217", "title": "Image Reconstruction without Explicit Priors", "abstract": "  We consider solving ill-posed imaging inverse problems without access to an\nexplicit image prior or ground-truth examples. An overarching challenge in\ninverse problems is that there are many undesired images that fit to the\nobserved measurements, thus requiring image priors to constrain the space of\npossible solutions to more plausible reconstructions. However, in many\napplications it is difficult or potentially impossible to obtain ground-truth\nimages to learn an image prior. Thus, inaccurate priors are often used, which\ninevitably result in biased solutions. Rather than solving an inverse problem\nusing priors that encode the explicit structure of any one image, we propose to\nsolve a set of inverse problems jointly by incorporating prior constraints on\nthe collective structure of the underlying images.The key assumption of our\nwork is that the ground-truth images we aim to reconstruct share common,\nlow-dimensional structure. We show that such a set of inverse problems can be\nsolved simultaneously by learning a shared image generator with a\nlow-dimensional latent space. The parameters of the generator and latent\nembedding are learned by maximizing a proxy for the Evidence Lower Bound\n(ELBO). Once learned, the generator and latent embeddings can be combined to\nprovide reconstructions for each inverse problem. The framework we propose can\nhandle general forward model corruptions, and we show that measurements derived\nfrom only a few ground-truth images (O(10)) are sufficient for image\nreconstruction without explicit priors.\n", "date": "2023-03-21", "categories": []}
{"id": "2303.12234", "title": "Pre-NeRF 360: Enriching Unbounded Appearances for Neural Radiance Fields", "abstract": "  Neural radiance fields (NeRF) appeared recently as a powerful tool to\ngenerate realistic views of objects and confined areas. Still, they face\nserious challenges with open scenes, where the camera has unrestricted movement\nand content can appear at any distance. In such scenarios, current\nNeRF-inspired models frequently yield hazy or pixelated outputs, suffer slow\ntraining times, and might display irregularities, because of the challenging\ntask of reconstructing an extensive scene from a limited number of images. We\npropose a new framework to boost the performance of NeRF-based architectures\nyielding significantly superior outcomes compared to the prior work. Our\nsolution overcomes several obstacles that plagued earlier versions of NeRF,\nincluding handling multiple video inputs, selecting keyframes, and extracting\nposes from real-world frames that are ambiguous and symmetrical. Furthermore,\nwe applied our framework, dubbed as \"Pre-NeRF 360\", to enable the use of the\nNutrition5k dataset in NeRF and introduce an updated version of this dataset,\nknown as the N5k360 dataset.\n", "date": "2023-03-21", "categories": []}
{"id": "2207.07705", "title": "Untrained, physics-informed neural networks for structured illumination\n  microscopy", "abstract": "  In recent years there has been great interest in using deep neural networks\n(DNN) for super-resolution image reconstruction including for structured\nillumination microscopy (SIM). While these methods have shown very promising\nresults, they all rely on data-driven, supervised training strategies that need\na large number of ground truth images, which is experimentally difficult to\nrealize. For SIM imaging, there exists a need for a flexible, general, and\nopen-source reconstruction method that can be readily adapted to different\nforms of structured illumination. We demonstrate that we can combine a deep\nneural network with the forward model of the structured illumination process to\nreconstruct sub-diffraction images without training data. The resulting\nphysics-informed neural network (PINN) can be optimized on a single set of\ndiffraction limited sub-images and thus doesn't require any training set. We\nshow with simulated and experimental data that this PINN can be applied to a\nwide variety of SIM methods by simply changing the known illumination patterns\nused in the loss function and can achieve resolution improvements that match\nwell with theoretical expectations.\n", "date": "2022-07-15", "categories": []}
{"id": "2211.10206", "title": "Multi-view Inverse Rendering for Large-scale Real-world Indoor Scenes", "abstract": "  We present a efficient multi-view inverse rendering method for large-scale\nreal-world indoor scenes that reconstructs global illumination and\nphysically-reasonable SVBRDFs. Unlike previous representations, where the\nglobal illumination of large scenes is simplified as multiple environment maps,\nwe propose a compact representation called Texture-based Lighting (TBL). It\nconsists of 3D mesh and HDR textures, and efficiently models direct and\ninfinite-bounce indirect lighting of the entire large scene. Based on TBL, we\nfurther propose a hybrid lighting representation with precomputed irradiance,\nwhich significantly improves the efficiency and alleviates the rendering noise\nin the material optimization. To physically disentangle the ambiguity between\nmaterials, we propose a three-stage material optimization strategy based on the\npriors of semantic segmentation and room segmentation. Extensive experiments\nshow that the proposed method outperforms the state-of-the-art quantitatively\nand qualitatively, and enables physically-reasonable mixed-reality applications\nsuch as material editing, editable novel view synthesis and relighting. The\nproject page is at https://lzleejean.github.io/TexIR.\n", "date": "2022-11-18", "categories": []}
{"id": "2301.02239", "title": "Robust Dynamic Radiance Fields", "abstract": "  Dynamic radiance field reconstruction methods aim to model the time-varying\nstructure and appearance of a dynamic scene. Existing methods, however, assume\nthat accurate camera poses can be reliably estimated by Structure from Motion\n(SfM) algorithms. These methods, thus, are unreliable as SfM algorithms often\nfail or produce erroneous poses on challenging videos with highly dynamic\nobjects, poorly textured surfaces, and rotating camera motion. We address this\nrobustness issue by jointly estimating the static and dynamic radiance fields\nalong with the camera parameters (poses and focal length). We demonstrate the\nrobustness of our approach via extensive quantitative and qualitative\nexperiments. Our results show favorable performance over the state-of-the-art\ndynamic view synthesis methods.\n", "date": "2023-01-05", "categories": []}
{"id": "2303.11724", "title": "Task-based Generation of Optimized Projection Sets using Differentiable\n  Ranking", "abstract": "  We present a method for selecting valuable projections in computed tomography\n(CT) scans to enhance image reconstruction and diagnosis. The approach\nintegrates two important factors, projection-based detectability and data\ncompleteness, into a single feed-forward neural network. The network evaluates\nthe value of projections, processes them through a differentiable ranking\nfunction and makes the final selection using a straight-through estimator. Data\ncompleteness is ensured through the label provided during training. The\napproach eliminates the need for heuristically enforcing data completeness,\nwhich may exclude valuable projections. The method is evaluated on simulated\ndata in a non-destructive testing scenario, where the aim is to maximize the\nreconstruction quality within a specified region of interest. We achieve\ncomparable results to previous methods, laying the foundation for using\nreconstruction-based loss functions to learn the selection of projections.\n", "date": "2023-03-21", "categories": []}
{"id": "2303.11212", "title": "Fluctuation-based deconvolution in fluorescence microscopy using\n  plug-and-play denoisers", "abstract": "  The spatial resolution of images of living samples obtained by fluorescence\nmicroscopes is physically limited due to the diffraction of visible light,\nwhich makes the study of entities of size less than the diffraction barrier\n(around 200 nm in the x-y plane) very challenging. To overcome this limitation,\nseveral deconvolution and super-resolution techniques have been proposed.\nWithin the framework of inverse problems, modern approaches in fluorescence\nmicroscopy reconstruct a super-resolved image from a temporal stack of frames\nby carefully designing suitable hand-crafted sparsity-promoting regularisers.\nNumerically, such approaches are solved by proximal gradient-based iterative\nschemes. Aiming at obtaining a reconstruction more adapted to sample geometries\n(e.g. thin filaments), we adopt a plug-and-play denoising approach with\nconvergence guarantees and replace the proximity operator associated with the\nexplicit image regulariser with an image denoiser (i.e. a pre-trained network)\nwhich, upon appropriate training, mimics the action of an implicit prior. To\naccount for the independence of the fluctuations between molecules, the model\nrelies on second-order statistics. The denoiser is then trained on covariance\nimages coming from data representing sequences of fluctuating fluorescent\nmolecules with filament structure. The method is evaluated on both simulated\nand real fluorescence microscopy images, showing its ability to correctly\nreconstruct filament structures with high values of peak signal-to-noise ratio\n(PSNR).\n", "date": "2023-03-20", "categories": []}
{"id": "2303.11265", "title": "Convergence Guarantees of Overparametrized Wide Deep Inverse Prior", "abstract": "  Neural networks have become a prominent approach to solve inverse problems in\nrecent years. Amongst the different existing methods, the Deep Image/Inverse\nPriors (DIPs) technique is an unsupervised approach that optimizes a highly\noverparametrized neural network to transform a random input into an object\nwhose image under the forward model matches the observation. However, the level\nof overparametrization necessary for such methods remains an open problem. In\nthis work, we aim to investigate this question for a two-layers neural network\nwith a smooth activation function. We provide overparametrization bounds under\nwhich such network trained via continuous-time gradient descent will converge\nexponentially fast with high probability which allows to derive recovery\nprediction bounds. This work is thus a first step towards a theoretical\nunderstanding of overparametrized DIP networks, and more broadly it\nparticipates to the theoretical understanding of neural networks in inverse\nproblem settings.\n", "date": "2023-03-20", "categories": []}
{"id": "2303.09757", "title": "Video Dehazing via a Multi-Range Temporal Alignment Network with\n  Physical Prior", "abstract": "  Video dehazing aims to recover haze-free frames with high visibility and\ncontrast. This paper presents a novel framework to effectively explore the\nphysical haze priors and aggregate temporal information. Specifically, we\ndesign a memory-based physical prior guidance module to encode the\nprior-related features into long-range memory. Besides, we formulate a\nmulti-range scene radiance recovery module to capture space-time dependencies\nin multiple space-time ranges, which helps to effectively aggregate temporal\ninformation from adjacent frames. Moreover, we construct the first large-scale\noutdoor video dehazing benchmark dataset, which contains videos in various\nreal-world scenarios. Experimental results on both synthetic and real\nconditions show the superiority of our proposed method.\n", "date": "2023-03-17", "categories": []}
{"id": "2303.09773", "title": "Progressive Content-aware Coded Hyperspectral Compressive Imaging", "abstract": "  Hyperspectral imaging plays a pivotal role in a wide range of applications,\nlike remote sensing, medicine, and cytology. By acquiring 3D hyperspectral\nimages (HSIs) via 2D sensors, the coded aperture snapshot spectral imaging\n(CASSI) has achieved great success due to its hardware-friendly implementation\nand fast imaging speed. However, for some less spectrally sparse scenes, single\nsnapshot and unreasonable coded aperture design tend to make HSI recovery more\nill-posed and yield poor spatial and spectral fidelity. In this paper, we\npropose a novel Progressive Content-Aware CASSI framework, dubbed PCA-CASSI,\nwhich captures HSIs with multiple optimized content-aware coded apertures and\nfuses all the snapshots for reconstruction progressively. Simultaneously, by\nmapping the Range-Null space Decomposition (RND) into a deep network with\nseveral phases, an RND-HRNet is proposed for HSI recovery. Each recovery phase\ncan fully exploit the hidden physical information in the coded apertures via\nexplicit $\\mathcal{R}$$-$$\\mathcal{N}$ decomposition and explore the\nspatial-spectral correlation by dual transformer blocks. Our method is\nvalidated to surpass other state-of-the-art methods on both multiple- and\nsingle-shot HSI imaging tasks by large margins.\n", "date": "2023-03-17", "categories": []}
{"id": "2212.04319", "title": "On the Robustness of Normalizing Flows for Inverse Problems in Imaging", "abstract": "  Conditional normalizing flows can generate diverse image samples for solving\ninverse problems. Most normalizing flows for inverse problems in imaging employ\nthe conditional affine coupling layer that can generate diverse images quickly.\nHowever, unintended severe artifacts are occasionally observed in the output of\nthem. In this work, we address this critical issue by investigating the origins\nof these artifacts and proposing the conditions to avoid them. First of all, we\nempirically and theoretically reveal that these problems are caused by\n\"exploding inverse\" in the conditional affine coupling layer for certain\nout-of-distribution (OOD) conditional inputs. Then, we further validated that\nthe probability of causing erroneous artifacts in pixels is highly correlated\nwith a Mahalanobis distance-based OOD score for inverse problems in imaging.\nLastly, based on our investigations, we propose a remark to avoid exploding\ninverse and then based on it, we suggest a simple remedy that substitutes the\naffine coupling layers with the modified rational quadratic spline coupling\nlayers in normalizing flows, to encourage the robustness of generated image\nsamples. Our experimental results demonstrated that our suggested methods\neffectively suppressed critical artifacts occurring in normalizing flows for\nsuper-resolution space generation and low-light image enhancement.\n", "date": "2022-12-08", "categories": []}
{"id": "2212.04968", "title": "SupeRVol: Super-Resolution Shape and Reflectance Estimation in Inverse\n  Volume Rendering", "abstract": "  We propose an end-to-end inverse rendering pipeline called SupeRVol that\nallows us to recover 3D shape and material parameters from a set of color\nimages in a super-resolution manner. To this end, we represent both the\nbidirectional reflectance distribution function (BRDF) and the signed distance\nfunction (SDF) by multi-layer perceptrons. In order to obtain both the surface\nshape and its reflectance properties, we revert to a differentiable volume\nrenderer with a physically based illumination model that allows us to decouple\nreflectance and lighting. This physical model takes into account the effect of\nthe camera's point spread function thereby enabling a reconstruction of shape\nand material in a super-resolution quality. Experimental validation confirms\nthat SupeRVol achieves state of the art performance in terms of inverse\nrendering quality. It generates reconstructions that are sharper than the\nindividual input images, making this method ideally suited for 3D modeling from\nlow-resolution imagery.\n", "date": "2022-12-09", "categories": []}
{"id": "2303.08156", "title": "Nonlinear Hyperspectral Unmixing based on Multilinear Mixing Model using\n  Convolutional Autoencoders", "abstract": "  Unsupervised spectral unmixing consists of representing each observed pixel\nas a combination of several pure materials called endmembers with their\ncorresponding abundance fractions. Beyond the linear assumption, various\nnonlinear unmixing models have been proposed, with the associated optimization\nproblems solved either by traditional optimization algorithms or deep learning\ntechniques. Current deep learning-based nonlinear unmixing focuses on the\nmodels in additive, bilinear-based formulations. By interpreting the reflection\nprocess using the discrete Markov chain, the multilinear mixing model (MLM)\nsuccessfully accounts for the up to infinite-order interactions between\nendmembers. However, to simulate the physics process of MLM by neural networks\nexplicitly is a challenging problem that has not been approached by far. In\nthis article, we propose a novel autoencoder-based network for unsupervised\nunmixing based on MLM. Benefitting from an elaborate network design, the\nrelationships among all the model parameters {\\em i.e.}, endmembers,\nabundances, and transition probability parameters are explicitly modeled. There\nare two modes: MLM-1DAE considers only pixel-wise spectral information, and\nMLM-3DAE exploits the spectral-spatial correlations within input patches.\nExperiments on both the synthetic and real datasets demonstrate the\neffectiveness of the proposed method as it achieves competitive performance to\nthe classic solutions of MLM.\n", "date": "2023-03-14", "categories": []}
{"id": "2208.14125", "title": "A Diffusion Model Predicts 3D Shapes from 2D Microscopy Images", "abstract": "  Diffusion models are a special type of generative model, capable of\nsynthesising new data from a learnt distribution. We introduce DISPR, a\ndiffusion-based model for solving the inverse problem of three-dimensional (3D)\ncell shape prediction from two-dimensional (2D) single cell microscopy images.\nUsing the 2D microscopy image as a prior, DISPR is conditioned to predict\nrealistic 3D shape reconstructions. To showcase the applicability of DISPR as a\ndata augmentation tool in a feature-based single cell classification task, we\nextract morphological features from the red blood cells grouped into six highly\nimbalanced classes. Adding features from the DISPR predictions to the three\nminority classes improved the macro F1 score from $F1_\\text{macro} = 55.2 \\pm\n4.6\\%$ to $F1_\\text{macro} = 72.2 \\pm 4.9\\%$. We thus demonstrate that\ndiffusion models can be successfully applied to inverse biomedical problems,\nand that they learn to reconstruct 3D shapes with realistic morphological\nfeatures from 2D microscopy images.\n", "date": "2022-08-30", "categories": []}
{"id": "2303.05183", "title": "Blind2Sound: Self-Supervised Image Denoising without Residual Noise", "abstract": "  Self-supervised blind denoising for Poisson-Gaussian noise remains a\nchallenging task. Pseudo-supervised pairs constructed from single noisy images\nre-corrupt the signal and degrade the performance. The visible blindspots solve\nthe information loss in masked inputs. However, without explicitly noise\nsensing, mean square error as an objective function cannot adjust denoising\nintensities for dynamic noise levels, leading to noticeable residual noise. In\nthis paper, we propose Blind2Sound, a simple yet effective approach to overcome\nresidual noise in denoised images. The proposed adaptive re-visible loss senses\nnoise levels and performs personalized denoising without noise residues while\nretaining the signal lossless. The theoretical analysis of intermediate medium\ngradients guarantees stable training, while the Cramer Gaussian loss acts as a\nregularization to facilitate the accurate perception of noise levels and\nimprove the performance of the denoiser. Experiments on synthetic and\nreal-world datasets show the superior performance of our method, especially for\nsingle-channel images.\n", "date": "2023-03-09", "categories": []}
{"id": "2303.07606", "title": "PSNet: a deep learning model based digital phase shifting algorithm from\n  a single fringe image", "abstract": "  As the gold standard for phase retrieval, phase-shifting algorithm (PS) has\nbeen widely used in optical interferometry, fringe projection profilometry,\netc. However, capturing multiple fringe patterns in PS limits the algorithm to\nonly a narrow range of application. To this end, a deep learning (DL) model\nbased digital PS algorithm from only a single fringe image is proposed. By\ntraining on a simulated dataset of PS fringe patterns, the learnt model,\ndenoted PSNet, can predict fringe patterns with other PS steps when given a\npattern with the first PS step. Simulation and experiment results demonstrate\nthe PSNet's promising performance on accurate prediction of digital PS\npatterns, and robustness to complex scenarios such as surfaces with varying\ncurvature and reflectance.\n", "date": "2023-03-14", "categories": []}
{"id": "2208.03524", "title": "Deep Learning-enabled Spatial Phase Unwrapping for 3D Measurement", "abstract": "  In terms of 3D imaging speed and system cost, the single-camera system\nprojecting single-frequency patterns is the ideal option among all proposed\nFringe Projection Profilometry (FPP) systems. This system necessitates a robust\nspatial phase unwrapping (SPU) algorithm. However, robust SPU remains a\nchallenge in complex scenes. Quality-guided SPU algorithms need more efficient\nways to identify the unreliable points in phase maps before unwrapping.\nEnd-to-end deep learning SPU methods face generality and interpretability\nproblems. This paper proposes a hybrid method combining deep learning and\ntraditional path-following for robust SPU in FPP. This hybrid SPU scheme\ndemonstrates better robustness than traditional quality-guided SPU methods,\nbetter interpretability than end-to-end deep learning scheme, and generality on\nunseen data. Experiments on the real dataset of multiple illumination\nconditions and multiple FPP systems differing in image resolution, the number\nof fringes, fringe direction, and optics wavelength verify the effectiveness of\nthe proposed method.\n", "date": "2022-08-06", "categories": []}
{"id": "2303.06418", "title": "Rethinking the Multi-view Stereo from the Perspective of Rendering-based\n  Augmentation", "abstract": "  GigaMVS presents several challenges to existing Multi-View Stereo (MVS)\nalgorithms for its large scale, complex occlusions, and gigapixel images. To\naddress these problems, we first apply one of the state-of-the-art\nlearning-based MVS methods, --MVSFormer, to overcome intractable scenarios such\nas textureless and reflections regions suffered by traditional PatchMatch\nmethods, but it fails in a few large scenes' reconstructions. Moreover,\ntraditional PatchMatch algorithms such as ACMMP, OpenMVS, and RealityCapture\nare leveraged to further improve the completeness in large scenes. Furthermore,\nto unify both advantages of deep learning methods and the traditional\nPatchMatch, we propose to render depth and color images to further fine-tune\nthe MVSFormer model. Notably, we find that the MVS method could produce much\nbetter predictions through rendered images due to the coincident illumination,\nwhich we believe is significant for the MVS community. Thus, MVSFormer is\ncapable of generalizing to large-scale scenes and complementarily solves the\ntextureless reconstruction problem. Finally, we have assembled all point clouds\nmentioned above \\textit{except ones from RealityCapture} and ranked Top-1 on\nthe competitive GigaReconstruction.\n", "date": "2023-03-11", "categories": []}
{"id": "2303.06575", "title": "Scale-aware Two-stage High Dynamic Range Imaging", "abstract": "  Deep high dynamic range (HDR) imaging as an image translation issue has\nachieved great performance without explicit optical flow alignment. However,\nchallenges remain over content association ambiguities especially caused by\nsaturation and large-scale movements. To address the ghosting issue and enhance\nthe details in saturated regions, we propose a scale-aware two-stage high\ndynamic range imaging framework (STHDR) to generate high-quality ghost-free HDR\nimage. The scale-aware technique and two-stage fusion strategy can\nprogressively and effectively improve the HDR composition performance.\nSpecifically, our framework consists of feature alignment and two-stage fusion.\nIn feature alignment, we propose a spatial correct module (SCM) to better\nexploit useful information among non-aligned features to avoid ghosting and\nsaturation. In the first stage of feature fusion, we obtain a preliminary\nfusion result with little ghosting. In the second stage, we conflate the\nresults of the first stage with aligned features to further reduce residual\nartifacts and thus improve the overall quality. Extensive experimental results\non the typical test dataset validate the effectiveness of the proposed STHDR in\nterms of speed and quality.\n", "date": "2023-03-12", "categories": []}
{"id": "2303.05879", "title": "Handheld Burst Super-Resolution Meets Multi-Exposure Satellite Imagery", "abstract": "  Image resolution is an important criterion for many applications based on\nsatellite imagery. In this work, we adapt a state-of-the-art kernel regression\ntechnique for smartphone camera burst super-resolution to satellites. This\ntechnique leverages the local structure of the image to optimally steer the\nfusion kernels, limiting blur in the final high-resolution prediction,\ndenoising the image, and recovering details up to a zoom factor of 2. We extend\nthis approach to the multi-exposure case to predict from a sequence of\nmulti-exposure low-resolution frames a high-resolution and noise-free one.\nExperiments on both single and multi-exposure scenarios show the merits of the\napproach. Since the fusion is learning-free, the proposed method is ensured to\nnot hallucinate details, which is crucial for many remote sensing applications.\n", "date": "2023-03-10", "categories": []}
{"id": "2210.09821", "title": "On-the-go Reflectance Transformation Imaging with Ordinary Smartphones", "abstract": "  Reflectance Transformation Imaging (RTI) is a popular technique that allows\nthe recovery of per-pixel reflectance information by capturing an object under\ndifferent light conditions. This can be later used to reveal surface details\nand interactively relight the subject. Such process, however, typically\nrequires dedicated hardware setups to recover the light direction from multiple\nlocations, making the process tedious when performed outside the lab.\n  We propose a novel RTI method that can be carried out by recording videos\nwith two ordinary smartphones. The flash led-light of one device is used to\nilluminate the subject while the other captures the reflectance. Since the led\nis mounted close to the camera lenses, we can infer the light direction for\nthousands of images by freely moving the illuminating device while observing a\nfiducial marker surrounding the subject. To deal with such amount of data, we\npropose a neural relighting model that reconstructs object appearance for\narbitrary light directions from extremely compact reflectance distribution data\ncompressed via Principal Components Analysis (PCA). Experiments shows that the\nproposed technique can be easily performed on the field with a resulting RTI\nmodel that can outperform state-of-the-art approaches involving dedicated\nhardware setups.\n", "date": "2022-10-18", "categories": []}
{"id": "2303.04667", "title": "STPDnet: Spatial-temporal convolutional primal dual network for dynamic\n  PET image reconstruction", "abstract": "  Dynamic positron emission tomography (dPET) image reconstruction is extremely\nchallenging due to the limited counts received in individual frame. In this\npaper, we propose a spatial-temporal convolutional primal dual network\n(STPDnet) for dynamic PET image reconstruction. Both spatial and temporal\ncorrelations are encoded by 3D convolution operators. The physical projection\nof PET is embedded in the iterative learning process of the network, which\nprovides the physical constraints and enhances interpretability. The\nexperiments of real rat scan data have shown that the proposed method can\nachieve substantial noise reduction in both temporal and spatial domains and\noutperform the maximum likelihood expectation maximization (MLEM),\nspatial-temporal kernel method (KEM-ST), DeepPET and Learned Primal Dual (LPD).\n", "date": "2023-03-08", "categories": []}
{"id": "2211.11423", "title": "Blur Interpolation Transformer for Real-World Motion from Blur", "abstract": "  This paper studies the challenging problem of recovering motion from blur,\nalso known as joint deblurring and interpolation or blur temporal\nsuper-resolution. The challenges are twofold: 1) the current methods still\nleave considerable room for improvement in terms of visual quality even on the\nsynthetic dataset, and 2) poor generalization to real-world data. To this end,\nwe propose a blur interpolation transformer (BiT) to effectively unravel the\nunderlying temporal correlation encoded in blur. Based on multi-scale residual\nSwin transformer blocks, we introduce dual-end temporal supervision and\ntemporally symmetric ensembling strategies to generate effective features for\ntime-varying motion rendering. In addition, we design a hybrid camera system to\ncollect the first real-world dataset of one-to-many blur-sharp video pairs.\nExperimental results show that BiT has a significant gain over the\nstate-of-the-art methods on the public dataset Adobe240. Besides, the proposed\nreal-world dataset effectively helps the model generalize well to real blurry\nscenarios. Code and data are available at https://github.com/zzh-tech/BiT.\n", "date": "2022-11-21", "categories": []}
{"id": "2303.03478", "title": "Amortized Normalizing Flows for Transcranial Ultrasound with Uncertainty\n  Quantification", "abstract": "  We present a novel approach to transcranial ultrasound computed tomography\nthat utilizes normalizing flows to improve the speed of imaging and provide\nBayesian uncertainty quantification. Our method combines physics-informed\nmethods and data-driven methods to accelerate the reconstruction of the final\nimage. We make use of a physics-informed summary statistic to incorporate the\nknown ultrasound physics with the goal of compressing large incoming\nobservations. This compression enables efficient training of the normalizing\nflow and standardizes the size of the data regardless of imaging\nconfigurations. The combinations of these methods results in fast\nuncertainty-aware image reconstruction that generalizes to a variety of\ntransducer configurations. We evaluate our approach with in silico experiments\nand demonstrate that it can significantly improve the imaging speed while\nquantifying uncertainty. We validate the quality of our image reconstructions\nby comparing against the traditional physics-only method and also verify that\nour provided uncertainty is calibrated with the error.\n", "date": "2023-03-06", "categories": []}
{"id": "2303.03555", "title": "Hyperspectral Compressive Wavefront Sensing", "abstract": "  Presented is a novel way to combine snapshot compressive imaging and lateral\nshearing interferometry in order to capture the spatio-spectral phase of an\nultrashort laser pulse in a single shot. A deep unrolling algorithm is utilised\nfor the snapshot compressive imaging reconstruction due to its parameter\nefficiency and superior speed relative to other methods, potentially allowing\nfor online reconstruction. The algorithm's regularisation term is represented\nusing neural network with 3D convolutional layers, to exploit the\nspatio-spectral correlations that exist in laser wavefronts. Compressed sensing\nis not typically applied to modulated signals, but we demonstrate its success\nhere. Furthermore, we train a neural network to predict the wavefronts from a\nlateral shearing interferogram in terms of Zernike polynomials, which again\nincreases the speed of our technique without sacrificing fidelity. This method\nis supported with simulation-based results. While applied to the example of\nlateral shearing interferometry, the methods presented here are generally\napplicable to a wide range of signals, including Shack-Hartmann-type sensors.\nThe results may be of interest beyond the context of laser wavefront\ncharacterization, including within quantitative phase imaging.\n", "date": "2023-03-06", "categories": []}
{"id": "2303.04025", "title": "DeepSeeColor: Realtime Adaptive Color Correction for Autonomous\n  Underwater Vehicles via Deep Learning Methods", "abstract": "  Successful applications of complex vision-based behaviours underwater have\nlagged behind progress in terrestrial and aerial domains. This is largely due\nto the degraded image quality resulting from the physical phenomena involved in\nunderwater image formation. Spectrally-selective light attenuation drains some\ncolors from underwater images while backscattering adds others, making it\nchallenging to perform vision-based tasks underwater. State-of-the-art methods\nfor underwater color correction optimize the parameters of image formation\nmodels to restore the full spectrum of color to underwater imagery. However,\nthese methods have high computational complexity that is unfavourable for\nrealtime use by autonomous underwater vehicles (AUVs), as a result of having\nbeen primarily designed for offline color correction. Here, we present\nDeepSeeColor, a novel algorithm that combines a state-of-the-art underwater\nimage formation model with the computational efficiency of deep learning\nframeworks. In our experiments, we show that DeepSeeColor offers comparable\nperformance to the popular \"Sea-Thru\" algorithm (Akkaynak & Treibitz, 2019)\nwhile being able to rapidly process images at up to 60Hz, thus making it\nsuitable for use onboard AUVs as a preprocessing step to enable more robust\nvision-based behaviours.\n", "date": "2023-03-07", "categories": []}
{"id": "2206.12455", "title": "Ev-NeRF: Event Based Neural Radiance Field", "abstract": "  We present Ev-NeRF, a Neural Radiance Field derived from event data. While\nevent cameras can measure subtle brightness changes in high frame rates, the\nmeasurements in low lighting or extreme motion suffer from significant domain\ndiscrepancy with complex noise. As a result, the performance of event-based\nvision tasks does not transfer to challenging environments, where the event\ncameras are expected to thrive over normal cameras. We find that the multi-view\nconsistency of NeRF provides a powerful self-supervision signal for eliminating\nthe spurious measurements and extracting the consistent underlying structure\ndespite highly noisy input. Instead of posed images of the original NeRF, the\ninput to Ev-NeRF is the event measurements accompanied by the movements of the\nsensors. Using the loss function that reflects the measurement model of the\nsensor, Ev-NeRF creates an integrated neural volume that summarizes the\nunstructured and sparse data points captured for about 2-4 seconds. The\ngenerated neural volume can also produce intensity images from novel views with\nreasonable depth estimates, which can serve as a high-quality input to various\nvision-based tasks. Our results show that Ev-NeRF achieves competitive\nperformance for intensity image reconstruction under extreme noise conditions\nand high-dynamic-range imaging.\n", "date": "2022-06-24", "categories": []}
{"id": "2303.02584", "title": "Super-Resolution Neural Operator", "abstract": "  We propose Super-resolution Neural Operator (SRNO), a deep operator learning\nframework that can resolve high-resolution (HR) images at arbitrary scales from\nthe low-resolution (LR) counterparts. Treating the LR-HR image pairs as\ncontinuous functions approximated with different grid sizes, SRNO learns the\nmapping between the corresponding function spaces. From the perspective of\napproximation theory, SRNO first embeds the LR input into a higher-dimensional\nlatent representation space, trying to capture sufficient basis functions, and\nthen iteratively approximates the implicit image function with a kernel\nintegral mechanism, followed by a final dimensionality reduction step to\ngenerate the RGB representation at the target coordinates. The key\ncharacteristics distinguishing SRNO from prior continuous SR works are: 1) the\nkernel integral in each layer is efficiently implemented via the Galerkin-type\nattention, which possesses non-local properties in the spatial domain and\ntherefore benefits the grid-free continuum; and 2) the multilayer attention\narchitecture allows for the dynamic latent basis update, which is crucial for\nSR problems to \"hallucinate\" high-frequency information from the LR image.\nExperiments show that SRNO outperforms existing continuous SR methods in terms\nof both accuracy and running time. Our code is at\nhttps://github.com/2y7c3/Super-Resolution-Neural-Operator\n", "date": "2023-03-05", "categories": []}
{"id": "2303.01515", "title": "Optimization-Based Deep learning methods for Magnetic Resonance Imaging\n  Reconstruction and Synthesis", "abstract": "  This dissertation is devoted to provide advanced nonconvex nonsmooth\nvariational models of (Magnetic Resonance Image) MRI reconstruction, efficient\nlearnable image reconstruction algorithms and parameter training algorithms\nthat improve the accuracy and robustness of the optimization-based deep\nlearning methods for compressed sensing MRI reconstruction and synthesis. The\nfirst part introduces a novel optimization based deep neural network whose\narchitecture is inspired by proximal gradient descent for solving a variational\nmodel. The second part is a substantial extension of the preliminary work in\nthe first part by solving the calibration-free fast pMRI reconstruction problem\nin a discrete-time optimal control framework. The third part aims at developing\na generalizable Magnetic Resonance Imaging (MRI) reconstruction method in the\nmeta-learning framework. The last part aims to synthesize target modality of\nMRI by using partially scanned k-space data from source modalities instead of\nfully scanned data that is used in the state-of-the-art multimodal synthesis.\n", "date": "2023-03-02", "categories": []}
{"id": "2303.01661", "title": "Longwave infrared multispectral image sensor system using\n  aluminum-germanium plasmonic filter arrays", "abstract": "  A multispectral camera records image data in various wavelengths across the\nelectromagnetic spectrum to acquire additional information that a conventional\ncamera fails to capture. With the advent of high-resolution image sensors and\ncolour filter technologies, multispectral imagers in the visible wavelengths\nhave become popular with increasing commercial viability in the last decade.\nHowever, multispectral imaging in longwave infrared (LWIR: 8 to 14 microns) is\nstill an emerging area due to the limited availability of optical materials,\nfilter technologies, and high-resolution sensors. Images from LWIR\nmultispectral cameras can capture emission spectra of objects to extract\nadditional information that a human eye fails to capture and thus have\nimportant applications in precision agriculture, forestry, medicine, and object\nidentification. In this work, we experimentally demonstrate an LWIR\nmultispectral image sensor with three wavelength bands using optical elements\nmade of an aluminum-based plasmonic filter array sandwiched in germanium. To\nrealize the multispectral sensor, the filter arrays are then integrated into a\n3D printed wheel stacked on a low-resolution monochrome thermal sensor. Our\nprototype device is calibrated using a blackbody and its thermal output has\nbeen enhanced with computer vision methods. By applying a state-of-the-art deep\nlearning method, we have also reconstructed multispectral images to a better\nspatial resolution. Scientifically, our work demonstrates a versatile spectral\nthermography technique for detecting target signatures in the LWIR range and\nother advanced spectral analyses.\n", "date": "2023-03-03", "categories": []}
{"id": "2303.01678", "title": "Nonlinear ill-posed problem in low-dose dental cone-beam computed\n  tomography", "abstract": "  This paper describes the mathematical structure of the ill-posed nonlinear\ninverse problem of low-dose dental cone-beam computed tomography (CBCT) and\nexplains the advantages of a deep learning-based approach to the reconstruction\nof computed tomography images over conventional regularization methods. This\npaper explains the underlying reasons why dental CBCT is more ill-posed than\nstandard computed tomography. Despite this severe ill-posedness, the demand for\ndental CBCT systems is rapidly growing because of their cost competitiveness\nand low radiation dose. We then describe the limitations of existing methods in\nthe accurate restoration of the morphological structures of teeth using dental\nCBCT data severely damaged by metal implants. We further discuss the usefulness\nof panoramic images generated from CBCT data for accurate tooth segmentation.\nWe also discuss the possibility of utilizing radiation-free intra-oral scan\ndata as prior information in CBCT image reconstruction to compensate for the\ndamage to data caused by metal implants.\n", "date": "2023-03-03", "categories": []}
{"id": "2303.02033", "title": "Single-photon Image Super-resolution via Self-supervised Learning", "abstract": "  Single-Photon Image Super-Resolution (SPISR) aims to recover a\nhigh-resolution volumetric photon counting cube from a noisy low-resolution one\nby computational imaging algorithms. In real-world scenarios, pairs of training\nsamples are often expensive or impossible to obtain. By extending Equivariant\nImaging (EI) to volumetric single-photon data, we propose a self-supervised\nlearning framework for the SPISR task. Particularly, using the Poisson unbiased\nKullback-Leibler risk estimator and equivariance, our method is able to learn\nfrom noisy measurements without ground truths. Comprehensive experiments on\nsimulated and real-world dataset demonstrate that the proposed method achieves\ncomparable performance with supervised learning and outperforms\ninterpolation-based methods.\n", "date": "2023-03-03", "categories": []}
{"id": "2303.00882", "title": "X-Ray2EM: Uncertainty-Aware Cross-Modality Image Reconstruction from\n  X-Ray to Electron Microscopy in Connectomics", "abstract": "  Comprehensive, synapse-resolution imaging of the brain will be crucial for\nunderstanding neuronal computations and function. In connectomics, this has\nbeen the sole purview of volume electron microscopy (EM), which entails an\nexcruciatingly difficult process because it requires cutting tissue into many\nthin, fragile slices that then need to be imaged, aligned, and reconstructed.\nUnlike EM, hard X-ray imaging is compatible with thick tissues, eliminating the\nneed for thin sectioning, and delivering fast acquisition, intrinsic alignment,\nand isotropic resolution. Unfortunately, current state-of-the-art X-ray\nmicroscopy provides much lower resolution, to the extent that segmenting\nmembranes is very challenging. We propose an uncertainty-aware 3D\nreconstruction model that translates X-ray images to EM-like images with\nenhanced membrane segmentation quality, showing its potential for developing\nsimpler, faster, and more accurate X-ray based connectomics pipelines.\n", "date": "2023-03-02", "categories": []}
{"id": "2303.00354", "title": "Unlimited-Size Diffusion Restoration", "abstract": "  Recently, using diffusion models for zero-shot image restoration (IR) has\nbecome a new hot paradigm. This type of method only needs to use the\npre-trained off-the-shelf diffusion models, without any finetuning, and can\ndirectly handle various IR tasks. The upper limit of the restoration\nperformance depends on the pre-trained diffusion models, which are in rapid\nevolution. However, current methods only discuss how to deal with fixed-size\nimages, but dealing with images of arbitrary sizes is very important for\npractical applications. This paper focuses on how to use those diffusion-based\nzero-shot IR methods to deal with any size while maintaining the excellent\ncharacteristics of zero-shot. A simple way to solve arbitrary size is to divide\nit into fixed-size patches and solve each patch independently. But this may\nyield significant artifacts since it neither considers the global semantics of\nall patches nor the local information of adjacent patches. Inspired by the\nRange-Null space Decomposition, we propose the Mask-Shift Restoration to\naddress local incoherence and propose the Hierarchical Restoration to alleviate\nout-of-domain issues. Our simple, parameter-free approaches can be used not\nonly for image restoration but also for image generation of unlimited sizes,\nwith the potential to be a general tool for diffusion models. Code:\nhttps://github.com/wyhuai/DDNM/tree/main/hq_demo\n", "date": "2023-03-01", "categories": []}
{"id": "2302.14247", "title": "Sequential edge detection using joint hierarchical Bayesian learning", "abstract": "  This paper introduces a new sparse Bayesian learning (SBL) algorithm that\njointly recovers a temporal sequence of edge maps from noisy and under-sampled\nFourier data. The new method is cast in a Bayesian framework and uses a prior\nthat simultaneously incorporates intra-image information to promote sparsity in\neach individual edge map with inter-image information to promote similarities\nin any unchanged regions. By treating both the edges as well as the similarity\nbetween adjacent images as random variables, there is no need to separately\nform regions of change. Thus we avoid both additional computational cost as\nwell as any information loss resulting from pre-processing the image. Our\nnumerical examples demonstrate that our new method compares favorably with more\nstandard SBL approaches.\n", "date": "2023-02-28", "categories": []}
{"id": "2301.03162", "title": "eFIN: Enhanced Fourier Imager Network for generalizable autofocusing and\n  pixel super-resolution in holographic imaging", "abstract": "  The application of deep learning techniques has greatly enhanced holographic\nimaging capabilities, leading to improved phase recovery and image\nreconstruction. Here, we introduce a deep neural network termed enhanced\nFourier Imager Network (eFIN) as a highly generalizable framework for hologram\nreconstruction with pixel super-resolution and image autofocusing. Through\nholographic microscopy experiments involving lung, prostate and salivary gland\ntissue sections and Papanicolau (Pap) smears, we demonstrate that eFIN has a\nsuperior image reconstruction quality and exhibits external generalization to\nnew types of samples never seen during the training phase. This network\nachieves a wide autofocusing axial range of 0.35 mm, with the capability to\naccurately predict the hologram axial distances by physics-informed learning.\neFIN enables 3x pixel super-resolution imaging and increases the\nspace-bandwidth product of the reconstructed images by 9-fold with almost no\nperformance loss, which allows for significant time savings in holographic\nimaging and data processing steps. Our results showcase the advancements of\neFIN in pushing the boundaries of holographic imaging for various applications\nin e.g., quantitative phase imaging and label-free microscopy.\n", "date": "2023-01-09", "categories": []}
{"id": "2302.13207", "title": "Stereo X-ray Tomography", "abstract": "  X-ray tomography is a powerful volumetric imaging technique, but detailed\nthree dimensional (3D) imaging requires the acquisition of a large number of\nindividual X-ray images, which is time consuming. For applications where\nspatial information needs to be collected quickly, for example, when studying\ndynamic processes, standard X-ray tomography is therefore not applicable.\nInspired by stereo vision, in this paper, we develop X-ray imaging methods that\nwork with two X-ray projection images. In this setting, without the use of\nadditional strong prior information, we no longer have enough information to\nfully recover the 3D tomographic images. However, up to a point, we are\nnevertheless able to extract spatial locations of point and line features. From\nstereo vision, it is well known that, for a known imaging geometry, once the\nsame point is identified in two images taken from different directions, then\nthe point's location in 3D space is exactly specified. The challenge is the\nmatching of points between images. As X-ray transmission images are\nfundamentally different from the surface reflection images used in standard\ncomputer vision, we here develop a different feature identification and\nmatching approach. In fact, once point like features are identified, if there\nare limited points in the image, then they can often be matched exactly. In\nfact, by utilising a third observation from an appropriate direction, matching\nbecomes unique. Once matched, point locations in 3D space are easily computed\nusing geometric considerations. Linear features, with clear end points, can be\nlocated using a similar approach.\n", "date": "2023-02-26", "categories": []}
{"id": "2302.11570", "title": "Plug-and-Play Deep Energy Model for Inverse problems", "abstract": "  We introduce a novel energy formulation for Plug- and-Play (PnP) image\nrecovery. Traditional PnP methods that use a convolutional neural network (CNN)\ndo not have an energy based formulation. The primary focus of this work is to\nintroduce an energy-based PnP formulation, which relies on a CNN that learns\nthe log of the image prior from training data. The score function is evaluated\nas the gradient of the energy model, which resembles a UNET with shared encoder\nand decoder weights. The proposed score function is thus constrained to a\nconservative vector field, which is the key difference with classical PnP\nmodels. The energy-based formulation offers algorithms with convergence\nguarantees, even when the learned score model is not a contraction. The\nrelaxation of the contraction constraint allows the proposed model to learn\nmore complex priors, thus offering improved performance over traditional PnP\nschemes. Our experiments in magnetic resonance image reconstruction\ndemonstrates the improved performance offered by the proposed energy model over\ntraditional PnP methods.\n", "date": "2023-02-15", "categories": []}
{"id": "2210.15948", "title": "Matching entropy based disparity estimation from light field", "abstract": "  A major challenge for matching-based depth estimation is to prevent\nmismatches in occlusion and smooth regions. An effective matching window\nsatisfying three characteristics: texture richness, disparity consistency and\nanti-occlusion should be able to prevent mismatches to some extent. According\nto these characteristics, we propose matching entropy in the spatial domain of\nlight field to measure the amount of correct information in a matching window,\nwhich provides the criterion for matching window selection. Based on matching\nentropy regularization, we establish an optimization model for depth estimation\nwith a matching cost fidelity term. To find the optimum, we propose a two-step\nadaptive matching algorithm. First, the region type is adaptively determined to\nidentify occluding, occluded, smooth and textured regions. Then, the matching\nentropy criterion is used to adaptively select the size and shape of matching\nwindows, as well as the visible viewpoints. The two-step process can reduce\nmismatches and redundant calculations by selecting effective matching windows.\nThe experimental results on synthetic and real data show that the proposed\nmethod can effectively improve the accuracy of depth estimation in occlusion\nand smooth regions and has strong robustness for different noise levels.\nTherefore, high-precision depth estimation from 4D light field data is\nachieved.\n", "date": "2022-10-28", "categories": []}
{"id": "2302.10309", "title": "Hierarchical Perception Adversarial Learning Framework for Compressed\n  Sensing MRI", "abstract": "  The long acquisition time has limited the accessibility of magnetic resonance\nimaging (MRI) because it leads to patient discomfort and motion artifacts.\nAlthough several MRI techniques have been proposed to reduce the acquisition\ntime, compressed sensing in magnetic resonance imaging (CS-MRI) enables fast\nacquisition without compromising SNR and resolution. However, existing CS-MRI\nmethods suffer from the challenge of aliasing artifacts. This challenge results\nin the noise-like textures and missing the fine details, thus leading to\nunsatisfactory reconstruction performance. To tackle this challenge, we propose\na hierarchical perception adversarial learning framework (HP-ALF). HP-ALF can\nperceive the image information in the hierarchical mechanism: image-level\nperception and patch-level perception. The former can reduce the visual\nperception difference in the entire image, and thus achieve aliasing artifact\nremoval. The latter can reduce this difference in the regions of the image, and\nthus recover fine details. Specifically, HP-ALF achieves the hierarchical\nmechanism by utilizing multilevel perspective discrimination. This\ndiscrimination can provide the information from two perspectives (overall and\nregional) for adversarial learning. It also utilizes a global and local\ncoherent discriminator to provide structure information to the generator during\ntraining. In addition, HP-ALF contains a context-aware learning block to\neffectively exploit the slice information between individual images for better\nreconstruction performance. The experiments validated on three datasets\ndemonstrate the effectiveness of HP-ALF and its superiority to the comparative\nmethods.\n", "date": "2023-01-27", "categories": []}
{"id": "2302.10481", "title": "LMPDNet: TOF-PET list-mode image reconstruction using model-based deep\n  learning method", "abstract": "  The integration of Time-of-Flight (TOF) information in the reconstruction\nprocess of Positron Emission Tomography (PET) yields improved image properties.\nHowever, implementing the cutting-edge model-based deep learning methods for\nTOF-PET reconstruction is challenging due to the substantial memory\nrequirements. In this study, we present a novel model-based deep learning\napproach, LMPDNet, for TOF-PET reconstruction from list-mode data. We address\nthe issue of real-time parallel computation of the projection matrix for\nlist-mode data, and propose an iterative model-based module that utilizes a\ndedicated network model for list-mode data. Our experimental results indicate\nthat the proposed LMPDNet outperforms traditional iteration-based TOF-PET\nlist-mode reconstruction algorithms. Additionally, we compare the spatial and\ntemporal consumption of list-mode data and sinogram data in model-based deep\nlearning methods, demonstrating the superiority of list-mode data in\nmodel-based TOF-PET reconstruction.\n", "date": "2023-02-21", "categories": []}
{"id": "2302.10502", "title": "Learning Gradually Non-convex Image Priors Using Score Matching", "abstract": "  In this paper, we propose a unified framework of denoising score-based models\nin the context of graduated non-convex energy minimization. We show that for\nsufficiently large noise variance, the associated negative log density -- the\nenergy -- becomes convex. Consequently, denoising score-based models\nessentially follow a graduated non-convexity heuristic. We apply this framework\nto learning generalized Fields of Experts image priors that approximate the\njoint density of noisy images and their associated variances. These priors can\nbe easily incorporated into existing optimization algorithms for solving\ninverse problems and naturally implement a fast and robust graduated\nnon-convexity mechanism.\n", "date": "2023-02-21", "categories": []}
{"id": "2206.00850", "title": "Dynamic MRI using Learned Transform-based Tensor Low-Rank Network\n  (LT$^2$LR-Net)", "abstract": "  While low-rank matrix prior has been exploited in dynamic MR image\nreconstruction and has obtained satisfying performance, tensor low-rank models\nhave recently emerged as powerful alternative representations for\nthree-dimensional dynamic MR datasets. In this paper, we introduce a novel deep\nunrolling network for dynamic MRI, namely the learned transform-based tensor\nlow-rank network (LT$^2$LR-Net). First, we generalize the tensor singular value\ndecomposition (t-SVD) into an arbitrary unitary transform-based version and\nsubsequently propose the novel transformed tensor nuclear norm (TTNN). Then, we\ndesign a novel TTNN-based iterative optimization algorithm based on the\nalternating direction method of multipliers (ADMM) to exploit the tensor\nlow-rank prior in the transformed domain. The corresponding iterative steps are\nunrolled into the proposed LT$^2$LR-Net, where the convolutional neural network\n(CNN) is incorporated to adaptively learn the transformation from the dynamic\nMR dataset for more robust and accurate tensor low-rank representations.\nExperimental results on the cardiac cine MR dataset demonstrate that the\nproposed framework can provide improved recovery results compared with the\nstate-of-the-art methods.\n", "date": "2022-06-02", "categories": []}
{"id": "2302.08765", "title": "On the Regularising Levenberg-Marquardt Method for Blinn-Phong\n  Photometric Stereo", "abstract": "  Photometric stereo refers to the process to compute the 3D shape of an object\nusing information on illumination and reflectance from several input images\nfrom the same point of view. The most often used reflectance model is the\nLambertian reflectance, however this does not include specular highlights in\ninput images. In this paper we consider the arising non-linear optimisation\nproblem when employing Blinn-Phong reflectance for modeling specular effects.\nTo this end we focus on the regularising Levenberg-Marquardt scheme. We show\nhow to derive an explicit bound that gives information on the convergence\nreliability of the method depending on given data, and we show how to gain\nexperimental evidence of numerical correctness of the iteration by making use\nof the Scherzer condition. The theoretical investigations that are at the heart\nof this paper are supplemented by some tests with real-world imagery.\n", "date": "2023-02-17", "categories": []}
{"id": "2302.07689", "title": "Event-guided Multi-patch Network with Self-supervision for Non-uniform\n  Motion Deblurring", "abstract": "  Contemporary deep learning multi-scale deblurring models suffer from many\nissues: 1) They perform poorly on non-uniformly blurred images/videos; 2)\nSimply increasing the model depth with finer-scale levels cannot improve\ndeblurring; 3) Individual RGB frames contain a limited motion information for\ndeblurring; 4) Previous models have a limited robustness to spatial\ntransformations and noise. Below, we extend the DMPHN model by several\nmechanisms to address the above issues: I) We present a novel self-supervised\nevent-guided deep hierarchical Multi-patch Network (MPN) to deal with blurry\nimages and videos via fine-to-coarse hierarchical localized representations;\nII) We propose a novel stacked pipeline, StackMPN, to improve the deblurring\nperformance under the increased network depth; III) We propose an event-guided\narchitecture to exploit motion cues contained in videos to tackle complex blur\nin videos; IV) We propose a novel self-supervised step to expose the model to\nrandom transformations (rotations, scale changes), and make it robust to\nGaussian noises. Our MPN achieves the state of the art on the GoPro and\nVideoDeblur datasets with a 40x faster runtime compared to current multi-scale\nmethods. With 30ms to process an image at 1280x720 resolution, it is the first\nreal-time deep motion deblurring model for 720p images at 30fps. For StackMPN,\nwe obtain significant improvements over 1.2dB on the GoPro dataset by\nincreasing the network depth. Utilizing the event information and\nself-supervision further boost results to 33.83dB.\n", "date": "2023-02-14", "categories": []}
{"id": "2211.16356", "title": "Real-time Blind Deblurring Based on Lightweight Deep-Wiener-Network", "abstract": "  In this paper, we address the problem of blind deblurring with high\nefficiency. We propose a set of lightweight deep-wiener-network to finish the\ntask with real-time speed. The Network contains a deep neural network for\nestimating parameters of wiener networks and a wiener network for deblurring.\nExperimental evaluations show that our approaches have an edge on State of the\nArt in terms of inference times and numbers of parameters. Two of our models\ncan reach a speed of 100 images per second, which is qualified for real-time\ndeblurring. Further research may focus on some real-world applications of\ndeblurring with our models.\n", "date": "2022-11-29", "categories": []}
{"id": "2302.06091", "title": "Boosted ab initio Cryo-EM 3D Reconstruction with ACE-EM", "abstract": "  The central problem in cryo-electron microscopy (cryo-EM) is to recover the\n3D structure from noisy 2D projection images which requires estimating the\nmissing projection angles (poses). Recent methods attempted to solve the 3D\nreconstruction problem with the autoencoder architecture, which suffers from\nthe latent vector space sampling problem and frequently produces suboptimal\npose inferences and inferior 3D reconstructions. Here we present an improved\nautoencoder architecture called ACE (Asymmetric Complementary autoEncoder),\nbased on which we designed the ACE-EM method for cryo-EM 3D reconstructions.\nCompared to previous methods, ACE-EM reached higher pose space coverage within\nthe same training time and boosted the reconstruction performance regardless of\nthe choice of decoders. With this method, the Nyquist resolution (highest\npossible resolution) was reached for 3D reconstructions of both simulated and\nexperimental cryo-EM datasets. Furthermore, ACE-EM is the only amortized\ninference method that reached the Nyquist resolution.\n", "date": "2023-02-13", "categories": []}
{"id": "2302.06895", "title": "SpeckleNN: A unified embedding for real-time speckle pattern\n  classification in X-ray single-particle imaging with limited labeled examples", "abstract": "  With X-ray free-electron lasers (XFELs), it is possible to determine the\nthree-dimensional structure of noncrystalline nanoscale particles using X-ray\nsingle-particle imaging (SPI) techniques at room temperature. Classifying SPI\nscattering patterns, or \"speckles\", to extract single hits that are needed for\nreal-time vetoing and three-dimensional reconstruction poses a challenge for\nhigh data rate facilities like European XFEL and LCLS-II-HE. Here, we introduce\nSpeckleNN, a unified embedding model for real-time speckle pattern\nclassification with limited labeled examples that can scale linearly with\ndataset size. Trained with twin neural networks, SpeckleNN maps speckle\npatterns to a unified embedding vector space, where similarity is measured by\nEuclidean distance. We highlight its few-shot classification capability on new\nnever-seen samples and its robust performance despite only tens of labels per\nclassification category even in the presence of substantial missing detector\nareas. Without the need for excessive manual labeling or even a full detector\nimage, our classification method offers a great solution for real-time\nhigh-throughput SPI experiments.\n", "date": "2023-02-14", "categories": []}
{"id": "2302.07135", "title": "Fast-MC-PET: A Novel Deep Learning-aided Motion Correction and\n  Reconstruction Framework for Accelerated PET", "abstract": "  Patient motion during PET is inevitable. Its long acquisition time not only\nincreases the motion and the associated artifacts but also the patient's\ndiscomfort, thus PET acceleration is desirable. However, accelerating PET\nacquisition will result in reconstructed images with low SNR, and the image\nquality will still be degraded by motion-induced artifacts. Most of the\nprevious PET motion correction methods are motion type specific that require\nmotion modeling, thus may fail when multiple types of motion present together.\nAlso, those methods are customized for standard long acquisition and could not\nbe directly applied to accelerated PET. To this end, modeling-free universal\nmotion correction reconstruction for accelerated PET is still highly\nunder-explored. In this work, we propose a novel deep learning-aided motion\ncorrection and reconstruction framework for accelerated PET, called\nFast-MC-PET. Our framework consists of a universal motion correction (UMC) and\na short-to-long acquisition reconstruction (SL-Reon) module. The UMC enables\nmodeling-free motion correction by estimating quasi-continuous motion from\nultra-short frame reconstructions and using this information for\nmotion-compensated reconstruction. Then, the SL-Recon converts the accelerated\nUMC image with low counts to a high-quality image with high counts for our\nfinal reconstruction output. Our experimental results on human studies show\nthat our Fast-MC-PET can enable 7-fold acceleration and use only 2 minutes\nacquisition to generate high-quality reconstruction images that\noutperform/match previous motion correction reconstruction methods using\nstandard 15 minutes long acquisition data.\n", "date": "2023-02-14", "categories": []}
{"id": "2301.08749", "title": "CSwin2SR: Circular Swin2SR for Compressed Image Super-Resolution", "abstract": "  Closed-loop negative feedback mechanism is extensively utilized in automatic\ncontrol systems and brings about extraordinary dynamic and static performance.\nIn order to further improve the reconstruction capability of current methods of\ncompressed image super-resolution, a circular Swin2SR (CSwin2SR) approach is\nproposed. The CSwin2SR contains a serial Swin2SR for initial super-resolution\nreestablishment and circular Swin2SR for enhanced super-resolution\nreestablishment. Simulated experimental results show that the proposed CSwin2SR\ndramatically outperforms the classical Swin2SR in the capacity of\nsuper-resolution recovery. On DIV2K test and valid datasets, the average\nincrement of PSNR is greater than 0.18 dB and the related average increment of\nSSIM is greater than 0.01.\n", "date": "2023-01-20", "categories": []}
{"id": "2302.05213", "title": "CEN-HDR: Computationally Efficient neural Network for real-time High\n  Dynamic Range imaging", "abstract": "  High dynamic range (HDR) imaging is still a challenging task in modern\ndigital photography. Recent research proposes solutions that provide\nhigh-quality acquisition but at the cost of a very large number of operations\nand a slow inference time that prevent the implementation of these solutions on\nlightweight real-time systems. In this paper, we propose CEN-HDR, a new\ncomputationally efficient neural network by providing a novel architecture\nbased on a light attention mechanism and sub-pixel convolution operations for\nreal-time HDR imaging. We also provide an efficient training scheme by applying\nnetwork compression using knowledge distillation. We performed extensive\nqualitative and quantitative comparisons to show that our approach produces\ncompetitive results in image quality while being faster than state-of-the-art\nsolutions, allowing it to be practically deployed under real-time constraints.\nExperimental results show our method obtains a score of 43.04 mu-PSNR on the\nKalantari2017 dataset with a framerate of 33 FPS using a Macbook M1 NPU.\n", "date": "2023-02-10", "categories": []}
{"id": "2207.04236", "title": "Sparse Ellipsometry: Portable Acquisition of Polarimetric SVBRDF and\n  Shape with Unstructured Flash Photography", "abstract": "  Ellipsometry techniques allow to measure polarization information of\nmaterials, requiring precise rotations of optical components with different\nconfigurations of lights and sensors. This results in cumbersome capture\ndevices, carefully calibrated in lab conditions, and in very long acquisition\ntimes, usually in the order of a few days per object. Recent techniques allow\nto capture polarimetric spatially-varying reflectance information, but limited\nto a single view, or to cover all view directions, but limited to spherical\nobjects made of a single homogeneous material. We present sparse ellipsometry,\na portable polarimetric acquisition method that captures both polarimetric\nSVBRDF and 3D shape simultaneously. Our handheld device consists of\noff-the-shelf, fixed optical components. Instead of days, the total acquisition\ntime varies between twenty and thirty minutes per object. We develop a complete\npolarimetric SVBRDF model that includes diffuse and specular components, as\nwell as single scattering, and devise a novel polarimetric inverse rendering\nalgorithm with data augmentation of specular reflection samples via generative\nmodeling. Our results show a strong agreement with a recent ground-truth\ndataset of captured polarimetric BRDFs of real-world objects.\n", "date": "2022-07-09", "categories": []}
{"id": "2207.03064", "title": "Shadow-Background-Noise 3D Spatial Decomposition Using Sparse Low-Rank\n  Gaussian Properties for Video-SAR Moving Target Shadow Enhancement", "abstract": "  Moving target shadows among video synthetic aperture radar (Video-SAR) images\nare always interfered by low scattering backgrounds and cluttered noises,\ncausing poor detec-tion-tracking accuracy. Thus, a shadow-background-noise 3D\nspatial decomposition (SBN-3D-SD) model is proposed to enhance shadows for\nhigher detection-tracking accuracy. It leverages the sparse property of\nshadows, the low-rank property of back-grounds, and the Gaussian property of\nnoises to perform 3D spatial three-decomposition. It separates shadows from\nback-grounds and noises by the alternating direction method of multi-pliers\n(ADMM). Results on the Sandia National Laboratories (SNL) data verify its\neffectiveness. It boosts the shadow saliency from the qualitative and\nquantitative evaluation. It boosts the shadow detection accuracy of Faster\nR-CNN, RetinaNet and YOLOv3. It also boosts the shadow tracking accuracy of\nTransTrack, FairMOT and ByteTrack.\n", "date": "2022-07-07", "categories": []}
{"id": "2211.01825", "title": "Fast Noise Removal in Hyperspectral Images via Representative\n  Coefficient Total Variation", "abstract": "  Mining structural priors in data is a widely recognized technique for\nhyperspectral image (HSI) denoising tasks, whose typical ways include\nmodel-based methods and data-based methods. The model-based methods have good\ngeneralization ability, while the runtime cannot meet the fast processing\nrequirements of the practical situations due to the large size of an HSI data $\n\\mathbf{X} \\in \\mathbb{R}^{MN\\times B}$. For the data-based methods, they\nperform very fast on new test data once they have been trained. However, their\ngeneralization ability is always insufficient. In this paper, we propose a fast\nmodel-based HSI denoising approach. Specifically, we propose a novel\nregularizer named Representative Coefficient Total Variation (RCTV) to\nsimultaneously characterize the low rank and local smooth properties. The RCTV\nregularizer is proposed based on the observation that the representative\ncoefficient matrix $\\mathbf{U}\\in\\mathbb{R}^{MN\\times R} (R\\ll B)$ obtained by\northogonally transforming the original HSI $\\mathbf{X}$ can inherit the strong\nlocal-smooth prior of $\\mathbf{X}$. Since $R/B$ is very small, the HSI\ndenoising model based on the RCTV regularizer has lower time complexity.\nAdditionally, we find that the representative coefficient matrix $\\mathbf{U}$\nis robust to noise, and thus the RCTV regularizer can somewhat promote the\nrobustness of the HSI denoising model. Extensive experiments on mixed noise\nremoval demonstrate the superiority of the proposed method both in denoising\nperformance and denoising speed compared with other state-of-the-art methods.\nRemarkably, the denoising speed of our proposed method outperforms all the\nmodel-based techniques and is comparable with the deep learning-based\napproaches.\n", "date": "2022-11-03", "categories": []}
{"id": "2301.11482", "title": "Diffusion Denoising for Low-Dose-CT Model", "abstract": "  Low-dose Computed Tomography (LDCT) reconstruction is an important task in\nmedical image analysis. Recent years have seen many deep learning based\nmethods, proved to be effective in this area. However, these methods mostly\nfollow a supervised architecture, which needs paired CT image of full dose and\nquarter dose, and the solution is highly dependent on specific measurements. In\nthis work, we introduce Denoising Diffusion LDCT Model, dubbed as DDLM,\ngenerating noise-free CT image using conditioned sampling. DDLM uses pretrained\nmodel, and need no training nor tuning process, thus our proposal is in\nunsupervised manner. Experiments on LDCT images have shown comparable\nperformance of DDLM using less inference time, surpassing other\nstate-of-the-art methods, proving both accurate and efficient. Implementation\ncode will be set to public soon.\n", "date": "2023-01-27", "categories": []}
{"id": "2302.02519", "title": "RDFNet: Regional Dynamic FISTA-Net for Spectral Snapshot Compressive\n  Imaging", "abstract": "  Deep convolutional neural networks have recently shown promising results in\ncompressive spectral reconstruction. Previous methods, however, usually adopt a\nsingle mapping function for sparse representation. Considering that different\nregions have distinct characteristics, it is desirable to apply various mapping\nfunctions to adjust different regions' transformations dynamically. With this\nin mind, we first introduce a regional dynamic way of using Fast Iterative\nShrinkage-Thresholding Algorithm (FISTA) to exploit regional characteristics\nand derive dynamic sparse representations. Then, we propose to unfold the\nprocess into a hierarchical dynamic deep network, dubbed RDFNet. The network\ncomprises multiple regional dynamic blocks and corresponding pixel-wise\nadaptive soft-thresholding modules, respectively in charge of region-based\ndynamic mapping and pixel-wise soft-thresholding selection. The regional\ndynamic block guides the network to adjust the transformation domain for\ndifferent regions. Equipped with the adaptive soft-thresholding, our proposed\nregional dynamic architecture can also learn appropriate shrinkage scale in a\npixel-wise manner.\n  Extensive experiments on both simulated and real data demonstrate that our\nmethod outperforms prior state-of-the-arts.\n", "date": "2023-02-06", "categories": []}
{"id": "2302.02849", "title": "An Unsupervised Framework for Joint MRI Super Resolution and Gibbs\n  Artifact Removal", "abstract": "  The k-space data generated from magnetic resonance imaging (MRI) is only a\nfinite sampling of underlying signals. Therefore, MRI images often suffer from\nlow spatial resolution and Gibbs ringing artifacts. Previous studies tackled\nthese two problems separately, where super resolution methods tend to enhance\nGibbs artifacts, whereas Gibbs ringing removal methods tend to blur the images.\nIt is also a challenge that high resolution ground truth is hard to obtain in\nclinical MRI. In this paper, we propose an unsupervised learning framework for\nboth MRI super resolution and Gibbs artifacts removal without using high\nresolution ground truth. Furthermore, we propose regularization methods to\nimprove the model's generalizability across out-of-distribution MRI images. We\nevaluated our proposed methods with other state-of-the-art methods on eight MRI\ndatasets with various contrasts and anatomical structures. Our method not only\nachieves the best SR performance but also significantly reduces the Gibbs\nartifacts. Our method also demonstrates good generalizability across different\ndatasets, which is beneficial to clinical applications where training data are\nusually scarce and biased.\n", "date": "2023-02-06", "categories": []}
{"id": "2302.00675", "title": "NDJIR: Neural Direct and Joint Inverse Rendering for Geometry, Lights,\n  and Materials of Real Object", "abstract": "  The goal of inverse rendering is to decompose geometry, lights, and materials\ngiven pose multi-view images. To achieve this goal, we propose neural direct\nand joint inverse rendering, NDJIR. Different from prior works which relies on\nsome approximations of the rendering equation, NDJIR directly addresses the\nintegrals in the rendering equation and jointly decomposes geometry: signed\ndistance function, lights: environment and implicit lights, materials: base\ncolor, roughness, specular reflectance using the powerful and flexible volume\nrendering framework, voxel grid feature, and Bayesian prior. Our method\ndirectly uses the physically-based rendering, so we can seamlessly export an\nextracted mesh with materials to DCC tools and show material conversion\nexamples. We perform intensive experiments to show that our proposed method can\ndecompose semantically well for real object in photogrammetric setting and what\nfactors contribute towards accurate inverse rendering.\n", "date": "2023-02-02", "categories": []}
{"id": "2302.01144", "title": "UW-CVGAN: UnderWater Image Enhancement with Capsules Vectors\n  Quantization", "abstract": "  The degradation in the underwater images is due to wavelength-dependent light\nattenuation, scattering, and to the diversity of the water types in which they\nare captured. Deep neural networks take a step in this field, providing\nautonomous models able to achieve the enhancement of underwater images. We\nintroduce Underwater Capsules Vectors GAN UWCVGAN based on the discrete\nfeatures quantization paradigm from VQGAN for this task. The proposed UWCVGAN\ncombines an encoding network, which compresses the image into its latent\nrepresentation, with a decoding network, able to reconstruct the enhancement of\nthe image from the only latent representation. In contrast with VQGAN, UWCVGAN\nachieves feature quantization by exploiting the clusterization ability of\ncapsule layer, making the model completely trainable and easier to manage. The\nmodel obtains enhanced underwater images with high quality and fine details.\nMoreover, the trained encoder is independent of the decoder giving the\npossibility to be embedded onto the collector as compressing algorithm to\nreduce the memory space required for the images, of factor $3\\times$.\n\\myUWCVGAN{ }is validated with quantitative and qualitative analysis on\nbenchmark datasets, and we present metrics results compared with the state of\nthe art.\n", "date": "2023-02-02", "categories": []}
{"id": "2301.13358", "title": "Hierarchical Disentangled Representation for Invertible Image Denoising\n  and Beyond", "abstract": "  Image denoising is a typical ill-posed problem due to complex degradation.\nLeading methods based on normalizing flows have tried to solve this problem\nwith an invertible transformation instead of a deterministic mapping. However,\nthe implicit bijective mapping is not explored well. Inspired by a latent\nobservation that noise tends to appear in the high-frequency part of the image,\nwe propose a fully invertible denoising method that injects the idea of\ndisentangled learning into a general invertible neural network to split noise\nfrom the high-frequency part. More specifically, we decompose the noisy image\ninto clean low-frequency and hybrid high-frequency parts with an invertible\ntransformation and then disentangle case-specific noise and high-frequency\ncomponents in the latent space. In this way, denoising is made tractable by\ninversely merging noiseless low and high-frequency parts. Furthermore, we\nconstruct a flexible hierarchical disentangling framework, which aims to\ndecompose most of the low-frequency image information while disentangling noise\nfrom the high-frequency part in a coarse-to-fine manner. Extensive experiments\non real image denoising, JPEG compressed artifact removal, and medical low-dose\nCT image restoration have demonstrated that the proposed method achieves\ncompeting performance on both quantitative metrics and visual quality, with\nsignificantly less computational cost.\n", "date": "2023-01-31", "categories": []}
{"id": "2301.12332", "title": "Towards Vision Transformer Unrolling Fixed-Point Algorithm: a Case Study\n  on Image Restoration", "abstract": "  The great success of Deep Neural Networks (DNNs) has inspired the algorithmic\ndevelopment of DNN-based Fixed-Point (DNN-FP) for computer vision tasks. DNN-FP\nmethods, trained by Back-Propagation Through Time or computing the inaccurate\ninversion of the Jacobian, suffer from inferior representation ability.\nMotivated by the representation power of the Transformer, we propose a\nframework to unroll the FP and approximate each unrolled process via\nTransformer blocks, called FPformer. To reduce the high consumption of memory\nand computation, we come up with FPRformer by sharing parameters between the\nsuccessive blocks. We further design a module to adapt Anderson acceleration to\nFPRformer to enlarge the unrolled iterations and improve the performance,\ncalled FPAformer. In order to fully exploit the capability of the Transformer,\nwe apply the proposed model to image restoration, using self-supervised\npre-training and supervised fine-tuning. 161 tasks from 4 categories of image\nrestoration problems are used in the pre-training phase. Hereafter, the\npre-trained FPformer, FPRformer, and FPAformer are further fine-tuned for the\ncomparison scenarios. Using self-supervised pre-training and supervised\nfine-tuning, the proposed FPformer, FPRformer, and FPAformer achieve\ncompetitive performance with state-of-the-art image restoration methods and\nbetter training efficiency. FPAformer employs only 29.82% parameters used in\nSwinIR models, and provides superior performance after fine-tuning. To train\nthese comparison models, it takes only 26.9% time used for training SwinIR\nmodels. It provides a promising way to introduce the Transformer in low-level\nvision tasks.\n", "date": "2023-01-29", "categories": []}
{"id": "2301.11525", "title": "Mixed Attention Network for Hyperspectral Image Denoising", "abstract": "  Hyperspectral image denoising is unique for the highly similar and correlated\nspectral information that should be properly considered. However, existing\nmethods show limitations in exploring the spectral correlations across\ndifferent bands and feature interactions within each band. Besides, the low-\nand high-level features usually exhibit different importance for different\nspatial-spectral regions, which is not fully explored for current algorithms as\nwell. In this paper, we present a Mixed Attention Network (MAN) that\nsimultaneously considers the inter- and intra-spectral correlations as well as\nthe interactions between low- and high-level spatial-spectral meaningful\nfeatures. Specifically, we introduce a multi-head recurrent spectral attention\nthat efficiently integrates the inter-spectral features across all the spectral\nbands. These features are further enhanced with a progressive spectral channel\nattention by exploring the intra-spectral relationships. Moreover, we propose\nan attentive skip-connection that adaptively controls the proportion of the\nlow- and high-level spatial-spectral features from the encoder and decoder to\nbetter enhance the aggregated features. Extensive experiments show that our MAN\noutperforms existing state-of-the-art methods on simulated and real noise\nsettings while maintaining a low cost of parameters and running time.\n", "date": "2023-01-27", "categories": []}
{"id": "2301.11813", "title": "Biomedical Image Reconstruction: A Survey", "abstract": "  Biomedical image reconstruction research has been developed for more than\nfive decades, giving rise to various techniques such as central and filtered\nback projection. With the rise of deep learning technology, biomedical image\nreconstruction field has undergone a massive paradigm shift from analytical and\niterative methods to deep learning methods To drive scientific discussion on\nadvanced deep learning techniques for biomedical image reconstruction, a\nworkshop focusing on deep biomedical image reconstruction, MLMIR, is introduced\nand is being held yearly since 2018. This survey paper is aimed to provide\nbasic knowledge in biomedical image reconstruction and the current research\ntrend in biomedical image reconstruction based on the publications in MLMIR.\nThis survey paper is intended for machine learning researchers to grasp a\ngeneral understanding of the biomedical image reconstruction field and the\ncurrent research trend in deep biomedical image reconstruction.\n", "date": "2023-01-04", "categories": []}
{"id": "2301.10208", "title": "A Simple Adaptive Unfolding Network for Hyperspectral Image\n  Reconstruction", "abstract": "  We present a simple, efficient, and scalable unfolding network, SAUNet, to\nsimplify the network design with an adaptive alternate optimization framework\nfor hyperspectral image (HSI) reconstruction. SAUNet customizes a Residual\nAdaptive ADMM Framework (R2ADMM) to connect each stage of the network via a\ngroup of learnable parameters to promote the usage of mask prior, which greatly\nstabilizes training and solves the accuracy degradation issue. Additionally, we\nintroduce a simple convolutional modulation block (CMB), which leads to\nefficient training, easy scale-up, and less computation. Coupling these two\ndesigns, SAUNet can be scaled to non-trivial 13 stages with continuous\nimprovement. Without bells and whistles, SAUNet improves both performance and\nspeed compared with the previous state-of-the-art counterparts, which makes it\nfeasible for practical high-resolution HSI reconstruction scenarios. We set new\nrecords on CAVE and KAIST HSI reconstruction benchmarks. Code and models are\navailable at https://github.com/hustvl/SAUNet.\n", "date": "2023-01-24", "categories": []}
{"id": "2206.07128", "title": "Stability of Image-Reconstruction Algorithms", "abstract": "  Robustness and stability of image-reconstruction algorithms have recently\ncome under scrutiny. Their importance to medical imaging cannot be overstated.\nWe review the known results for the topical variational regularization\nstrategies ($\\ell_2$ and $\\ell_1$ regularization) and present novel stability\nresults for $\\ell_p$-regularized linear inverse problems for $p\\in(1,\\infty)$.\nOur results guarantee Lipschitz continuity for small $p$ and H\\\"{o}lder\ncontinuity for larger $p$. They generalize well to the $L_p(\\Omega)$ function\nspaces.\n", "date": "2022-06-14", "categories": []}
{"id": "2301.09452", "title": "Fast and robust single particle reconstruction in 3D fluorescence\n  microscopy", "abstract": "  Single particle reconstruction has recently emerged in 3D fluorescence\nmicroscopy as a powerful technique to improve the axial resolution and the\ndegree of fluorescent labeling. It is based on the reconstruction of an average\nvolume of a biological particle from the acquisition multiple views with\nunknown poses. Current methods are limited either by template bias, restriction\nto 2D data, high computational cost or a lack of robustness to low fluorescent\nlabeling. In this work, we propose a single particle reconstruction method\ndedicated to convolutional models in 3D fluorescence microscopy that overcome\nthese issues. We address the joint reconstruction and estimation of the poses\nof the particles, which translates into a challenging non-convex optimization\nproblem. Our approach is based on a multilevel reformulation of this problem,\nand the development of efficient optimization techniques at each level. We\ndemonstrate on synthetic data that our method outperforms the standard\napproaches in terms of resolution and reconstruction error, while achieving a\nlow computational cost. We also perform successful reconstruction on real\ndatasets of centrioles to show the potential of our method in concrete\napplications.\n", "date": "2023-01-23", "categories": []}
{"id": "2301.06392", "title": "I See-Through You: A Framework for Removing Foreground Occlusion in Both\n  Sparse and Dense Light Field Images", "abstract": "  Light field (LF) camera captures rich information from a scene. Using the\ninformation, the LF de-occlusion (LF-DeOcc) task aims to reconstruct the\nocclusion-free center view image. Existing LF-DeOcc studies mainly focus on the\nsparsely sampled (sparse) LF images where most of the occluded regions are\nvisible in other views due to the large disparity. In this paper, we expand\nLF-DeOcc in more challenging datasets, densely sampled (dense) LF images, which\nare taken by a micro-lens-based portable LF camera. Due to the small disparity\nranges of dense LF images, most of the background regions are invisible in any\nview. To apply LF-DeOcc in both LF datasets, we propose a framework, ISTY,\nwhich is defined and divided into three roles: (1) extract LF features, (2)\ndefine the occlusion, and (3) inpaint occluded regions. By dividing the\nframework into three specialized components according to the roles, the\ndevelopment and analysis can be easier. Furthermore, an explainable\nintermediate representation, an occlusion mask, can be obtained in the proposed\nframework. The occlusion mask is useful for comprehensive analysis of the model\nand other applications by manipulating the mask. In experiments, qualitative\nand quantitative results show that the proposed framework outperforms\nstate-of-the-art LF-DeOcc methods in both sparse and dense LF datasets.\n", "date": "2023-01-16", "categories": []}
{"id": "2301.05434", "title": "LVRNet: Lightweight Image Restoration for Aerial Images under Low\n  Visibility", "abstract": "  Learning to recover clear images from images having a combination of\ndegrading factors is a challenging task. That being said, autonomous\nsurveillance in low visibility conditions caused by high pollution/smoke, poor\nair quality index, low light, atmospheric scattering, and haze during a\nblizzard becomes even more important to prevent accidents. It is thus crucial\nto form a solution that can result in a high-quality image and is efficient\nenough to be deployed for everyday use. However, the lack of proper datasets\navailable to tackle this task limits the performance of the previous methods\nproposed. To this end, we generate the LowVis-AFO dataset, containing 3647\npaired dark-hazy and clear images. We also introduce a lightweight deep\nlearning model called Low-Visibility Restoration Network (LVRNet). It\noutperforms previous image restoration methods with low latency, achieving a\nPSNR value of 25.744 and an SSIM of 0.905, making our approach scalable and\nready for practical use. The code and data can be found at\nhttps://github.com/Achleshwar/LVRNet.\n", "date": "2023-01-13", "categories": []}
{"id": "2301.04805", "title": "DEA-Net: Single image dehazing based on detail-enhanced convolution and\n  content-guided attention", "abstract": "  Single image dehazing is a challenging ill-posed problem which estimates\nlatent haze-free images from observed hazy images. Some existing deep learning\nbased methods are devoted to improving the model performance via increasing the\ndepth or width of convolution. The learning ability of convolutional neural\nnetwork (CNN) structure is still under-explored. In this paper, a\ndetail-enhanced attention block (DEAB) consisting of the detail-enhanced\nconvolution (DEConv) and the content-guided attention (CGA) is proposed to\nboost the feature learning for improving the dehazing performance.\nSpecifically, the DEConv integrates prior information into normal convolution\nlayer to enhance the representation and generalization capacity. Then by using\nthe re-parameterization technique, DEConv is equivalently converted into a\nvanilla convolution with NO extra parameters and computational cost. By\nassigning unique spatial importance map (SIM) to every channel, CGA can attend\nmore useful information encoded in features. In addition, a CGA-based mixup\nfusion scheme is presented to effectively fuse the features and aid the\ngradient flow. By combining above mentioned components, we propose our\ndetail-enhanced attention network (DEA-Net) for recovering high-quality\nhaze-free images. Extensive experimental results demonstrate the effectiveness\nof our DEA-Net, outperforming the state-of-the-art (SOTA) methods by boosting\nthe PSNR index over 41 dB with only 3.653 M parameters. The source code of our\nDEA-Net will be made available at https://github.com/cecret3350/DEA-Net.\n", "date": "2023-01-12", "categories": []}
{"id": "2210.15492", "title": "Reconstruction of compressed spectral imaging based on global structure\n  and spectral correlation", "abstract": "  In this paper, a convolutional sparse coding method based on global structure\ncharacteristics and spectral correlation is proposed for the reconstruction of\ncompressive spectral images. The spectral data is regarded as the convolution\nsum of the convolution kernel and the corresponding coefficients, using the\nconvolution kernel operates the global image information, preserving the\nstructure information of the spectral image in the spatial dimension. To take\nfull exploration of the constraints between spectra, the coefficients\ncorresponding to the convolution kernel are constrained by the L_(2,1)norm to\nimprove spectral accuracy. And, to solve the problem that convolutional sparse\ncoding is insensitive to low frequency, the global total-variation (TV)\nconstraint is added to estimate the low-frequency components. It not only\nensures the effective estimation of the low-frequency but also transforms the\nconvolutional sparse coding into a de-noising process, which makes the\nreconstructing process simpler. Simulations show that compared with the current\nmainstream optimization methods, the proposed method can improve the\nreconstruction quality by up to 4 dB in PSNR and 10% in SSIM, and has a great\nimprovement in the details of the reconstructed image.\n", "date": "2022-10-27", "categories": []}
{"id": "2301.03027", "title": "Annealed Score-Based Diffusion Model for MR Motion Artifact Reduction", "abstract": "  Motion artifact reduction is one of the important research topics in MR\nimaging, as the motion artifact degrades image quality and makes diagnosis\ndifficult. Recently, many deep learning approaches have been studied for motion\nartifact reduction. Unfortunately, most existing models are trained in a\nsupervised manner, requiring paired motion-corrupted and motion-free images, or\nare based on a strict motion-corruption model, which limits their use for\nreal-world situations. To address this issue, here we present an annealed\nscore-based diffusion model for MRI motion artifact reduction. Specifically, we\ntrain a score-based model using only motion-free images, and then motion\nartifacts are removed by applying forward and reverse diffusion processes\nrepeatedly to gradually impose a low-frequency data consistency. Experimental\nresults verify that the proposed method successfully reduces both simulated and\nin vivo motion artifacts, outperforming the state-of-the-art deep learning\nmethods.\n", "date": "2023-01-08", "categories": []}
{"id": "2301.03047", "title": "Large-scale Global Low-rank Optimization for Computational Compressed\n  Imaging", "abstract": "  Computational reconstruction plays a vital role in computer vision and\ncomputational photography. Most of the conventional optimization and deep\nlearning techniques explore local information for reconstruction. Recently,\nnonlocal low-rank (NLR) reconstruction has achieved remarkable success in\nimproving accuracy and generalization. However, the computational cost has\ninhibited NLR from seeking global structural similarity, which consequentially\nkeeps it trapped in the tradeoff between accuracy and efficiency and prevents\nit from high-dimensional large-scale tasks. To address this challenge, we\nreport here the global low-rank (GLR) optimization technique, realizing\nhighly-efficient large-scale reconstruction with global self-similarity.\nInspired by the self-attention mechanism in deep learning, GLR extracts\nexemplar image patches by feature detection instead of conventional uniform\nselection. This directly produces key patches using structural features to\navoid burdensome computational redundancy. Further, it performs patch matching\nacross the entire image via neural-based convolution, which produces the global\nsimilarity heat map in parallel, rather than conventional sequential block-wise\nmatching. As such, GLR improves patch grouping efficiency by more than one\norder of magnitude. We experimentally demonstrate GLR's effectiveness on\ntemporal, frequency, and spectral dimensions, including different computational\nimaging modalities of compressive temporal imaging, magnetic resonance imaging,\nand multispectral filter array demosaicing. This work presents the superiority\nof inherent fusion of deep learning strategies and iterative optimization, and\nbreaks the persistent dilemma of the tradeoff between accuracy and efficiency\nfor various large-scale reconstruction tasks.\n", "date": "2023-01-08", "categories": []}
{"id": "2301.03362", "title": "Image Denoising: The Deep Learning Revolution and Beyond -- A Survey\n  Paper --", "abstract": "  Image denoising (removal of additive white Gaussian noise from an image) is\none of the oldest and most studied problems in image processing. An extensive\nwork over several decades has led to thousands of papers on this subject, and\nto many well-performing algorithms for this task. Indeed, 10 years ago, these\nachievements have led some researchers to suspect that \"Denoising is Dead\", in\nthe sense that all that can be achieved in this domain has already been\nobtained. However, this turned out to be far from the truth, with the\npenetration of deep learning (DL) into image processing. The era of DL brought\na revolution to image denoising, both by taking the lead in today's ability for\nnoise removal in images, and by broadening the scope of denoising problems\nbeing treated. Our paper starts by describing this evolution, highlighting in\nparticular the tension and synergy that exist between classical approaches and\nmodern DL-based alternatives in design of image denoisers.\n  The recent transitions in the field of image denoising go far beyond the\nability to design better denoisers. In the 2nd part of this paper we focus on\nrecently discovered abilities and prospects of image denoisers. We expose the\npossibility of using denoisers to serve other problems, such as regularizing\ngeneral inverse problems and serving as the prime engine in diffusion-based\nimage synthesis. We also unveil the idea that denoising and other inverse\nproblems might not have a unique solution as common algorithms would have us\nbelieve. Instead, we describe constructive ways to produce randomized and\ndiverse high quality results for inverse problems, all fueled by the progress\nthat DL brought to image denoising.\n  This survey paper aims to provide a broad view of the history of image\ndenoising and closely related topics. Our aim is to give a better context to\nrecent discoveries, and to the influence of DL in our domain.\n", "date": "2023-01-09", "categories": []}
{"id": "2301.01355", "title": "Holistic Multi-Slice Framework for Dynamic Simultaneous Multi-Slice MRI\n  Reconstruction", "abstract": "  Dynamic Magnetic Resonance Imaging (dMRI) is widely used to assess various\ncardiac conditions such as cardiac motion and blood flow. To accelerate MR\nacquisition, techniques such as undersampling and Simultaneous Multi-Slice\n(SMS) are often used. Special reconstruction algorithms are needed to\nreconstruct multiple SMS image slices from the entangled information. Deep\nlearning (DL)-based methods have shown promising results for single-slice MR\nreconstruction, but the addition of SMS acceleration raises unique challenges\ndue to the composite k-space signals and the resulting images with strong\ninter-slice artifacts. Furthermore, many dMRI applications lack sufficient data\nfor training reconstruction neural networks. In this study, we propose a novel\nDL-based framework for dynamic SMS reconstruction. Our main contributions are\n1) a combination of data transformation steps and network design that\neffectively leverages the unique characteristics of undersampled dynamic SMS\ndata, and 2) an MR physics-guided transfer learning strategy that addresses the\ndata scarcity issue. Thorough comparisons with multiple baseline methods\nillustrate the strengths of our proposed methods.\n", "date": "2023-01-03", "categories": []}
{"id": "2207.01466", "title": "Physics-informed compressed sensing for PC-MRI: an inverse Navier-Stokes\n  problem", "abstract": "  We formulate a physics-informed compressed sensing (PICS) method for the\nreconstruction of velocity fields from noisy and sparse phase-contrast magnetic\nresonance signals. The method solves an inverse Navier-Stokes boundary value\nproblem, which permits us to jointly reconstruct and segment the velocity\nfield, and at the same time infer hidden quantities such as the hydrodynamic\npressure and the wall shear stress. Using a Bayesian framework, we regularize\nthe problem by introducing a priori information about the unknown parameters in\nthe form of Gaussian random fields. This prior information is updated using the\nNavier-Stokes problem, an energy-based segmentation functional, and by\nrequiring that the reconstruction is consistent with the $k$-space signals. We\ncreate an algorithm that solves this reconstruction problem, and test it for\nnoisy and sparse $k$-space signals of the flow through a converging nozzle. We\nfind that the method is capable of reconstructing and segmenting the velocity\nfields from sparsely-sampled (15% $k$-space coverage), low ($\\sim$$10$)\nsignal-to-noise ratio (SNR) signals, and that the reconstructed velocity field\ncompares well with that derived from fully-sampled (100% $k$-space coverage)\nhigh ($>40$) SNR signals of the same flow.\n", "date": "2022-07-04", "categories": []}
{"id": "2211.10563", "title": "Real-World Image Super Resolution via Unsupervised Bi-directional Cycle\n  Domain Transfer Learning based Generative Adversarial Network", "abstract": "  Deep Convolutional Neural Networks (DCNNs) have exhibited impressive\nperformance on image super-resolution tasks. However, these deep learning-based\nsuper-resolution methods perform poorly in real-world super-resolution tasks,\nwhere the paired high-resolution and low-resolution images are unavailable and\nthe low-resolution images are degraded by complicated and unknown kernels. To\nbreak these limitations, we propose the Unsupervised Bi-directional Cycle\nDomain Transfer Learning-based Generative Adversarial Network (UBCDTL-GAN),\nwhich consists of an Unsupervised Bi-directional Cycle Domain Transfer Network\n(UBCDTN) and the Semantic Encoder guided Super Resolution Network (SESRN).\nFirst, the UBCDTN is able to produce an approximated real-like LR image through\ntransferring the LR image from an artificially degraded domain to the\nreal-world LR image domain. Second, the SESRN has the ability to super-resolve\nthe approximated real-like LR image to a photo-realistic HR image. Extensive\nexperiments on unpaired real-world image benchmark datasets demonstrate that\nthe proposed method achieves superior performance compared to state-of-the-art\nmethods.\n", "date": "2022-11-19", "categories": []}
{"id": "2206.06070", "title": "Annular Computational Imaging: Capture Clear Panoramic Images through\n  Simple Lens", "abstract": "  Panoramic Annular Lens (PAL) composed of few lenses has great potential in\npanoramic surrounding sensing tasks for mobile and wearable devices because of\nits tiny size and large Field of View (FoV). However, the image quality of\ntiny-volume PAL confines to optical limit due to the lack of lenses for\naberration correction. In this paper, we propose an Annular Computational\nImaging (ACI) framework to break the optical limit of light-weight PAL design.\nTo facilitate learning-based image restoration, we introduce a wave-based\nsimulation pipeline for panoramic imaging and tackle the synthetic-to-real gap\nthrough multiple data distributions. The proposed pipeline can be easily\nadapted to any PAL with design parameters and is suitable for loose-tolerance\ndesigns. Furthermore, we design the Physics Informed Image Restoration Network\n(PI2RNet) considering the physical priors of panoramic imaging and single-pass\nphysics-informed engine. At the dataset level, we create the DIVPano dataset\nand the extensive experiments on it illustrate that our proposed network sets\nthe new state of the art in the panoramic image restoration under\nspatially-variant degradation. In addition, the evaluation of the proposed ACI\non a simple PAL with only 3 spherical lenses reveals the delicate balance\nbetween high-quality panoramic imaging and compact design. To the best of our\nknowledge, we are the first to explore Computational Imaging (CI) in PAL. Code\nand datasets are publicly available at\nhttps://github.com/zju-jiangqi/ACI-PI2RNet.\n", "date": "2022-06-13", "categories": []}
{"id": "2211.05018", "title": "The Best of Both Worlds: a Framework for Combining Degradation\n  Prediction with High Performance Super-Resolution Networks", "abstract": "  To date, the best-performing blind super-resolution (SR) techniques follow\none of two paradigms: A) generate and train a standard SR network on synthetic\nlow-resolution - high-resolution (LR - HR) pairs or B) attempt to predict the\ndegradations an LR image has suffered and use these to inform a customised SR\nnetwork. Despite significant progress, subscribers to the former miss out on\nuseful degradation information that could be used to improve the SR process. On\nthe other hand, followers of the latter rely on weaker SR networks, which are\nsignificantly outperformed by the latest architectural advancements. In this\nwork, we present a framework for combining any blind SR prediction mechanism\nwith any deep SR network, using a metadata insertion block to insert prediction\nvectors into SR network feature maps. Through comprehensive testing, we prove\nthat state-of-the-art contrastive and iterative prediction schemes can be\nsuccessfully combined with high-performance SR networks such as RCAN and HAN\nwithin our framework. We show that our hybrid models consistently achieve\nstronger SR performance than both their non-blind and blind counterparts.\nFurthermore, we demonstrate our framework's robustness by predicting\ndegradations and super-resolving images from a complex pipeline of blurring,\nnoise and compression.\n", "date": "2022-11-09", "categories": []}
{"id": "2210.04866", "title": "PoGaIN: Poisson-Gaussian Image Noise Modeling from Paired Samples", "abstract": "  Image noise can often be accurately fitted to a Poisson-Gaussian\ndistribution. However, estimating the distribution parameters from a noisy\nimage only is a challenging task. Here, we study the case when paired noisy and\nnoise-free samples are accessible. No method is currently available to exploit\nthe noise-free information, which may help to achieve more accurate\nestimations. To fill this gap, we derive a novel, cumulant-based, approach for\nPoisson-Gaussian noise modeling from paired image samples. We show its improved\nperformance over different baselines, with special emphasis on MSE, effect of\noutliers, image dependence, and bias. We additionally derive the log-likelihood\nfunction for further insights and discuss real-world applicability.\n", "date": "2022-10-10", "categories": []}
{"id": "2212.13654", "title": "Large-scale single-photon imaging", "abstract": "  Benefiting from its single-photon sensitivity, single-photon avalanche diode\n(SPAD) array has been widely applied in various fields such as fluorescence\nlifetime imaging and quantum computing. However, large-scale high-fidelity\nsingle-photon imaging remains a big challenge, due to the complex hardware\nmanufacture craft and heavy noise disturbance of SPAD arrays. In this work, we\nintroduce deep learning into SPAD, enabling super-resolution single-photon\nimaging over an order of magnitude, with significant enhancement of bit depth\nand imaging quality. We first studied the complex photon flow model of SPAD\nelectronics to accurately characterize multiple physical noise sources, and\ncollected a real SPAD image dataset (64 $\\times$ 32 pixels, 90 scenes, 10\ndifferent bit depth, 3 different illumination flux, 2790 images in total) to\ncalibrate noise model parameters. With this real-world physical noise model, we\nfor the first time synthesized a large-scale realistic single-photon image\ndataset (image pairs of 5 different resolutions with maximum megapixels, 17250\nscenes, 10 different bit depth, 3 different illumination flux, 2.6 million\nimages in total) for subsequent network training. To tackle the severe\nsuper-resolution challenge of SPAD inputs with low bit depth, low resolution,\nand heavy noise, we further built a deep transformer network with a\ncontent-adaptive self-attention mechanism and gated fusion modules, which can\ndig global contextual features to remove multi-source noise and extract\nfull-frequency details. We applied the technique on a series of experiments\nincluding macroscopic and microscopic imaging, microfluidic inspection, and\nFourier ptychography. The experiments validate the technique's state-of-the-art\nsuper-resolution SPAD imaging performance, with more than 5 dB superiority on\nPSNR compared to the existing methods.\n", "date": "2022-12-28", "categories": []}
{"id": "2210.10892", "title": "DEEP$^2$: Deep Learning Powered De-scattering with Excitation Patterning", "abstract": "  Limited throughput is a key challenge in in-vivo deep-tissue imaging using\nnonlinear optical microscopy. Point scanning multiphoton microscopy, the\ncurrent gold standard, is slow especially compared to the wide-field imaging\nmodalities used for optically cleared or thin specimens. We recently introduced\n'De-scattering with Excitation Patterning or DEEP', as a widefield alternative\nto point-scanning geometries. Using patterned multiphoton excitation, DEEP\nencodes spatial information inside tissue before scattering. However, to\nde-scatter at typical depths, hundreds of such patterned excitations are\nneeded. In this work, we present DEEP$^2$, a deep learning based model, that\ncan de-scatter images from just tens of patterned excitations instead of\nhundreds. Consequently, we improve DEEP's throughput by almost an order of\nmagnitude. We demonstrate our method in multiple numerical and physical\nexperiments including in-vivo cortical vasculature imaging up to four\nscattering lengths deep, in alive mice.\n", "date": "2022-10-19", "categories": []}
{"id": "2207.11406", "title": "PS-NeRF: Neural Inverse Rendering for Multi-view Photometric Stereo", "abstract": "  Traditional multi-view photometric stereo (MVPS) methods are often composed\nof multiple disjoint stages, resulting in noticeable accumulated errors. In\nthis paper, we present a neural inverse rendering method for MVPS based on\nimplicit representation. Given multi-view images of a non-Lambertian object\nilluminated by multiple unknown directional lights, our method jointly\nestimates the geometry, materials, and lights. Our method first employs\nmulti-light images to estimate per-view surface normal maps, which are used to\nregularize the normals derived from the neural radiance field. It then jointly\noptimizes the surface normals, spatially-varying BRDFs, and lights based on a\nshadow-aware differentiable rendering layer. After optimization, the\nreconstructed object can be used for novel-view rendering, relighting, and\nmaterial editing. Experiments on both synthetic and real datasets demonstrate\nthat our method achieves far more accurate shape reconstruction than existing\nMVPS and neural rendering methods. Our code and model can be found at\nhttps://ywq.github.io/psnerf.\n", "date": "2022-07-23", "categories": []}
{"id": "2212.11274", "title": "SPIRiT-Diffusion: SPIRiT-driven Score-Based Generative Modeling for\n  Vessel Wall imaging", "abstract": "  Diffusion model is the most advanced method in image generation and has been\nsuccessfully applied to MRI reconstruction. However, the existing methods do\nnot consider the characteristics of multi-coil acquisition of MRI data.\nTherefore, we give a new diffusion model, called SPIRiT-Diffusion, based on the\nSPIRiT iterative reconstruction algorithm. Specifically, SPIRiT-Diffusion\ncharacterizes the prior distribution of coil-by-coil images by score matching\nand characterizes the k-space redundant prior between coils based on\nself-consistency. With sufficient prior constraint utilized, we achieve\nsuperior reconstruction results on the joint Intracranial and Carotid Vessel\nWall imaging dataset.\n", "date": "2022-12-14", "categories": []}
{"id": "2207.00698", "title": "Uncertainty Quantification for Deep Unrolling-Based Computational\n  Imaging", "abstract": "  Deep unrolling is an emerging deep learning-based image reconstruction\nmethodology that bridges the gap between model-based and purely deep\nlearning-based image reconstruction methods. Although deep unrolling methods\nachieve state-of-the-art performance for imaging problems and allow the\nincorporation of the observation model into the reconstruction process, they do\nnot provide any uncertainty information about the reconstructed image, which\nseverely limits their use in practice, especially for safety-critical imaging\napplications. In this paper, we propose a learning-based image reconstruction\nframework that incorporates the observation model into the reconstruction task\nand that is capable of quantifying epistemic and aleatoric uncertainties, based\non deep unrolling and Bayesian neural networks. We demonstrate the uncertainty\ncharacterization capability of the proposed framework on magnetic resonance\nimaging and computed tomography reconstruction problems. We investigate the\ncharacteristics of the epistemic and aleatoric uncertainty information provided\nby the proposed framework to motivate future research on utilizing uncertainty\ninformation to develop more accurate, robust, trustworthy, uncertainty-aware,\nlearning-based image reconstruction and analysis methods for imaging problems.\nWe show that the proposed framework can provide uncertainty information while\nachieving comparable reconstruction performance to state-of-the-art deep\nunrolling methods.\n", "date": "2022-07-02", "categories": []}
{"id": "2212.10167", "title": "Seafloor-Invariant Caustics Removal from Underwater Imagery", "abstract": "  Mapping the seafloor with underwater imaging cameras is of significant\nimportance for various applications including marine engineering, geology,\ngeomorphology, archaeology and biology. For shallow waters, among the\nunderwater imaging challenges, caustics i.e., the complex physical phenomena\nresulting from the projection of light rays being refracted by the wavy\nsurface, is likely the most crucial one. Caustics is the main factor during\nunderwater imaging campaigns that massively degrade image quality and affect\nseverely any 2D mosaicking or 3D reconstruction of the seabed. In this work, we\npropose a novel method for correcting the radiometric effects of caustics on\nshallow underwater imagery. Contrary to the state-of-the-art, the developed\nmethod can handle seabed and riverbed of any anaglyph, correcting the images\nusing real pixel information, thus, improving image matching and 3D\nreconstruction processes. In particular, the developed method employs deep\nlearning architectures in order to classify image pixels to \"non-caustics\" and\n\"caustics\". Then, exploits the 3D geometry of the scene to achieve a pixel-wise\ncorrection, by transferring appropriate color values between the overlapping\nunderwater images. Moreover, to fill the current gap, we have collected,\nannotated and structured a real-world caustic dataset, namely R-CAUSTIC, which\nis openly available. Overall, based on the experimental results and validation\nthe developed methodology is quite promising in both detecting caustics and\nreconstructing their intensity.\n", "date": "2022-12-20", "categories": []}
{"id": "2212.09088", "title": "LR-CSNet: Low-Rank Deep Unfolding Network for Image Compressive Sensing", "abstract": "  Deep unfolding networks (DUNs) have proven to be a viable approach to\ncompressive sensing (CS). In this work, we propose a DUN called low-rank CS\nnetwork (LR-CSNet) for natural image CS. Real-world image patches are often\nwell-represented by low-rank approximations. LR-CSNet exploits this property by\nadding a low-rank prior to the CS optimization task. We derive a corresponding\niterative optimization procedure using variable splitting, which is then\ntranslated to a new DUN architecture. The architecture uses low-rank generation\nmodules (LRGMs), which learn low-rank matrix factorizations, as well as\ngradient descent and proximal mappings (GDPMs), which are proposed to extract\nhigh-frequency features to refine image details. In addition, the deep features\ngenerated at each reconstruction stage in the DUN are transferred between\nstages to boost the performance. Our extensive experiments on three widely\nconsidered datasets demonstrate the promising performance of LR-CSNet compared\nto state-of-the-art methods in natural image CS.\n", "date": "2022-12-18", "categories": []}
{"id": "2211.14821", "title": "Towards Realistic Underwater Dataset Generation and Color Restoration", "abstract": "  Recovery of true color from underwater images is an ill-posed problem. This\nis because the wide-band attenuation coefficients for the RGB color channels\ndepend on object range, reflectance, etc. which are difficult to model. Also,\nthere is backscattering due to suspended particles in water. Thus, most\nexisting deep-learning based color restoration methods, which are trained on\nsynthetic underwater datasets, do not perform well on real underwater data.\nThis can be attributed to the fact that synthetic data cannot accurately\nrepresent real conditions. To address this issue, we use an image to image\ntranslation network to bridge the gap between the synthetic and real domains by\ntranslating images from synthetic underwater domain to real underwater domain.\nUsing this multimodal domain adaptation technique, we create a dataset that can\ncapture a diverse array of underwater conditions. We then train a simple but\neffective CNN based network on our domain adapted dataset to perform color\nrestoration. Code and pre-trained models can be accessed at\nhttps://github.com/nehamjain10/TRUDGCR\n", "date": "2022-11-27", "categories": []}
{"id": "2212.07599", "title": "Universal Generative Modeling in Dual-domain for Dynamic MR Imaging", "abstract": "  Dynamic magnetic resonance image reconstruction from incomplete k-space data\nhas generated great research interest due to its capability to reduce scan\ntime. Never-theless, the reconstruction problem is still challenging due to its\nill-posed nature. Recently, diffusion models espe-cially score-based generative\nmodels have exhibited great potential in algorithm robustness and usage\nflexi-bility. Moreover, the unified framework through the variance exploding\nstochastic differential equation (VE-SDE) is proposed to enable new sampling\nmethods and further extend the capabilities of score-based gener-ative models.\nTherefore, by taking advantage of the uni-fied framework, we proposed a k-space\nand image Du-al-Domain collaborative Universal Generative Model (DD-UGM) which\ncombines the score-based prior with low-rank regularization penalty to\nreconstruct highly under-sampled measurements. More precisely, we extract prior\ncomponents from both image and k-space domains via a universal generative model\nand adaptively handle these prior components for faster processing while\nmaintaining good generation quality. Experimental comparisons demonstrated the\nnoise reduction and detail preservation abilities of the proposed method. Much\nmore than that, DD-UGM can reconstruct data of differ-ent frames by only\ntraining a single frame image, which reflects the flexibility of the proposed\nmodel.\n", "date": "2022-12-15", "categories": []}
{"id": "2212.06345", "title": "Foveated Thermal Computational Imaging in the Wild Using All-Silicon\n  Meta-Optics", "abstract": "  Foveated imaging provides a better tradeoff between situational awareness\n(field of view) and resolution and is critical in long-wavelength infrared\nregimes because of the size, weight, power, and cost of thermal sensors. We\ndemonstrate computational foveated imaging by exploiting the ability of a\nmeta-optical frontend to discriminate between different polarization states and\na computational backend to reconstruct the captured image/video. The frontend\nis a three-element optic: the first element which we call the \"foveal\" element\nis a metalens that focuses s-polarized light at a distance of $f_1$ without\naffecting the p-polarized light; the second element which we call the\n\"perifoveal\" element is another metalens that focuses p-polarized light at a\ndistance of $f_2$ without affecting the s-polarized light. The third element is\na freely rotating polarizer that dynamically changes the mixing ratios between\nthe two polarization states. Both the foveal element (focal length = 150mm;\ndiameter = 75mm), and the perifoveal element (focal length = 25mm; diameter =\n25mm) were fabricated as polarization-sensitive, all-silicon, meta surfaces\nresulting in a large-aperture, 1:6 foveal expansion, thermal imaging\ncapability. A computational backend then utilizes a deep image prior to\nseparate the resultant multiplexed image or video into a foveated image\nconsisting of a high-resolution center and a lower-resolution large field of\nview context. We build a first-of-its-kind prototype system and demonstrate 12\nframes per second real-time, thermal, foveated image, and video capture in the\nwild.\n", "date": "2022-12-13", "categories": []}
{"id": "2208.10171", "title": "Noise-Adaptive Intelligent Programmable Meta-Imager", "abstract": "  We present an intelligent programmable computational meta-imager that tailors\nits sequence of coherent scene illuminations not only to a specific\ninformation-extraction task (e.g., object recognition) but also adapts to\ndifferent types and levels of noise. We systematically study how the learned\nillumination patterns depend on the noise, and we discover that trends in\nintensity and overlap of the learned illumination patterns can be understood\nintuitively. We conduct our analysis based on an analytical coupled-dipole\nforward model of a microwave dynamic metasurface antenna (DMA); we formulate a\ndifferentiable end-to-end information-flow pipeline comprising the programmable\nphysical measurement process including noise as well as the subsequent digital\nprocessing layers. This pipeline allows us to jointly inverse-design the\nprogrammable physical weights (DMA configurations that determine the coherent\nscene illuminations) and the trainable digital weights. Our noise-adaptive\nintelligent meta-imager outperforms the conventional use of pseudo-random\nillumination patterns most clearly under conditions that make the extraction of\nsufficient task-relevant information challenging: latency constraints (limiting\nthe number of allowed measurements) and strong noise. Programmable microwave\nmeta-imagers in indoor surveillance and earth observation will be confronted\nwith these conditions.\n", "date": "2022-08-22", "categories": []}
{"id": "2212.05503", "title": "Low-rank Tensor Assisted K-space Generative Model for Parallel Imaging\n  Reconstruction", "abstract": "  Although recent deep learning methods, especially generative models, have\nshown good performance in fast magnetic resonance imaging, there is still much\nroom for improvement in high-dimensional generation. Considering that internal\ndimensions in score-based generative models have a critical impact on\nestimating the gradient of the data distribution, we present a new idea,\nlow-rank tensor assisted k-space generative model (LR-KGM), for parallel\nimaging reconstruction. This means that we transform original prior information\ninto high-dimensional prior information for learning. More specifically, the\nmulti-channel data is constructed into a large Hankel matrix and the matrix is\nsubsequently folded into tensor for prior learning. In the testing phase, the\nlow-rank rotation strategy is utilized to impose low-rank constraints on tensor\noutput of the generative network. Furthermore, we alternately use traditional\ngenerative iterations and low-rank high-dimensional tensor iterations for\nreconstruction. Experimental comparisons with the state-of-the-arts\ndemonstrated that the proposed LR-KGM method achieved better performance.\n", "date": "2022-12-11", "categories": []}
{"id": "2212.04309", "title": "Deep Variational Inverse Scattering", "abstract": "  Inverse medium scattering solvers generally reconstruct a single solution\nwithout an associated measure of uncertainty. This is true both for the\nclassical iterative solvers and for the emerging deep learning methods. But\nill-posedness and noise can make this single estimate inaccurate or misleading.\nWhile deep networks such as conditional normalizing flows can be used to sample\nposteriors in inverse problems, they often yield low-quality samples and\nuncertainty estimates. In this paper, we propose U-Flow, a Bayesian U-Net based\non conditional normalizing flows, which generates high-quality posterior\nsamples and estimates physically-meaningful uncertainty. We show that the\nproposed model significantly outperforms the recent normalizing flows in terms\nof posterior sample quality while having comparable performance with the U-Net\nin point estimation.\n", "date": "2022-12-08", "categories": []}
{"id": "2208.07181", "title": "One-shot Generative Prior in Hankel-k-space for Parallel Imaging\n  Reconstruction", "abstract": "  Magnetic resonance imaging serves as an essential tool for clinical\ndiagnosis. However, it suffers from a long acquisition time. The utilization of\ndeep learning, especially the deep generative models, offers aggressive\nacceleration and better reconstruction in magnetic resonance imaging.\nNevertheless, learning the data distribution as prior knowledge and\nreconstructing the image from limited data remains challenging. In this work,\nwe propose a novel Hankel-k-space generative model (HKGM), which can generate\nsamples from a training set of as little as one k-space data. At the prior\nlearning stage, we first construct a large Hankel matrix from k-space data,\nthen extract multiple structured k-space patches from the large Hankel matrix\nto capture the internal distribution among different patches. Extracting\npatches from a Hankel matrix enables the generative model to be learned from\nredundant and low-rank data space. At the iterative reconstruction stage, it is\nobserved that the desired solution obeys the learned prior knowledge. The\nintermediate reconstruction solution is updated by taking it as the input of\nthe generative model. The updated result is then alternatively operated by\nimposing low-rank penalty on its Hankel matrix and data consistency con-strain\non the measurement data. Experimental results confirmed that the internal\nstatistics of patches within a single k-space data carry enough information for\nlearning a powerful generative model and provide state-of-the-art\nreconstruction.\n", "date": "2022-08-15", "categories": []}
{"id": "2212.04264", "title": "ChromaCorrect: Prescription Correction in Virtual Reality Headsets\n  through Perceptual Guidance", "abstract": "  A large portion of today's world population suffer from vision impairments\nand wear prescription eyeglasses. However, eyeglasses causes additional bulk\nand discomfort when used with augmented and virtual reality headsets, thereby\nnegatively impacting the viewer's visual experience. In this work, we remedy\nthe usage of prescription eyeglasses in Virtual Reality (VR) headsets by\nshifting the optical complexity completely into software and propose a\nprescription-aware rendering approach for providing sharper and immersive VR\nimagery. To this end, we develop a differentiable display and visual perception\nmodel encapsulating display-specific parameters, color and visual acuity of\nhuman visual system and the user-specific refractive errors. Using this\ndifferentiable visual perception model, we optimize the rendered imagery in the\ndisplay using stochastic gradient-descent solvers. This way, we provide\nprescription glasses-free sharper images for a person with vision impairments.\nWe evaluate our approach on various displays, including desktops and VR\nheadsets, and show significant quality and contrast improvements for users with\nvision impairments.\n", "date": "2022-12-08", "categories": []}
{"id": "2212.03630", "title": "One Sample Diffusion Model in Projection Domain for Low-Dose CT Imaging", "abstract": "  Low-dose computed tomography (CT) plays a significant role in reducing the\nradiation risk in clinical applications. However, lowering the radiation dose\nwill significantly degrade the image quality. With the rapid development and\nwide application of deep learning, it has brought new directions for the\ndevelopment of low-dose CT imaging algorithms. Therefore, we propose a fully\nunsupervised one sample diffusion model (OSDM)in projection domain for low-dose\nCT reconstruction. To extract sufficient prior information from single sample,\nthe Hankel matrix formulation is employed. Besides, the penalized weighted\nleast-squares and total variation are introduced to achieve superior image\nquality. Specifically, we first train a score-based generative model on one\nsinogram by extracting a great number of tensors from the structural-Hankel\nmatrix as the network input to capture prior distribution. Then, at the\ninference stage, the stochastic differential equation solver and data\nconsistency step are performed iteratively to obtain the sinogram data.\nFinally, the final image is obtained through the filtered back-projection\nalgorithm. The reconstructed results are approaching to the normal-dose\ncounterparts. The results prove that OSDM is practical and effective model for\nreducing the artifacts and preserving the image quality.\n", "date": "2022-12-07", "categories": []}
{"id": "2212.00928", "title": "Single-shot ToF sensing with sub-mm precision using conventional CMOS\n  sensors", "abstract": "  We present a novel single-shot interferometric ToF camera targeted for\nprecise 3D measurements of dynamic objects. The camera concept is based on\nSynthetic Wavelength Interferometry, a technique that allows retrieval of depth\nmaps of objects with optically rough surfaces at submillimeter depth precision.\nIn contrast to conventional ToF cameras, our device uses only off-the-shelf\nCCD/CMOS detectors and works at their native chip resolution (as of today,\ntheoretically up to 20 Mp and beyond). Moreover, we can obtain a full 3D model\nof the object in single-shot, meaning that no temporal sequence of exposures or\ntemporal illumination modulation (such as amplitude or frequency modulation) is\nnecessary, which makes our camera robust against object motion.\n  In this paper, we introduce the novel camera concept and show first\nmeasurements that demonstrate the capabilities of our system. We present 3D\nmeasurements of small (cm-sized) objects with > 2 Mp point cloud resolution\n(the resolution of our used detector) and up to sub-mm depth precision. We also\nreport a \"single-shot 3D video\" acquisition and a first single-shot\n\"Non-Line-of-Sight\" measurement. Our technique has great potential for\nhigh-precision applications with dynamic object movement, e.g., in AR/VR,\nindustrial inspection, medical imaging, and imaging through scattering media\nlike fog or human tissue.\n", "date": "2022-12-02", "categories": []}
{"id": "2206.07281", "title": "Super-resolution image display using diffractive decoders", "abstract": "  High-resolution synthesis/projection of images over a large field-of-view\n(FOV) is hindered by the restricted space-bandwidth-product (SBP) of wavefront\nmodulators. We report a deep learning-enabled diffractive display design that\nis based on a jointly-trained pair of an electronic encoder and a diffractive\noptical decoder to synthesize/project super-resolved images using\nlow-resolution wavefront modulators. The digital encoder, composed of a trained\nconvolutional neural network (CNN), rapidly pre-processes the high-resolution\nimages of interest so that their spatial information is encoded into\nlow-resolution (LR) modulation patterns, projected via a low SBP wavefront\nmodulator. The diffractive decoder processes this LR encoded information using\nthin transmissive layers that are structured using deep learning to\nall-optically synthesize and project super-resolved images at its output FOV.\nOur results indicate that this diffractive image display can achieve a\nsuper-resolution factor of ~4, demonstrating a ~16-fold increase in SBP. We\nalso experimentally validate the success of this diffractive super-resolution\ndisplay using 3D-printed diffractive decoders that operate at the THz spectrum.\nThis diffractive image decoder can be scaled to operate at visible wavelengths\nand inspire the design of large FOV and high-resolution displays that are\ncompact, low-power, and computationally efficient.\n", "date": "2022-06-15", "categories": []}
{"id": "2212.00595", "title": "Ghost-free High Dynamic Range Imaging via Hybrid CNN-Transformer and\n  Structure Tensor", "abstract": "  Eliminating ghosting artifacts due to moving objects is a challenging problem\nin high dynamic range (HDR) imaging. In this letter, we present a hybrid model\nconsisting of a convolutional encoder and a Transformer decoder to generate\nghost-free HDR images. In the encoder, a context aggregation network and\nnon-local attention block are adopted to optimize multi-scale features and\ncapture both global and local dependencies of multiple low dynamic range (LDR)\nimages. The decoder based on Swin Transformer is utilized to improve the\nreconstruction capability of the proposed model. Motivated by the phenomenal\ndifference between the presence and absence of artifacts under the field of\nstructure tensor (ST), we integrate the ST information of LDR images as\nauxiliary inputs of the network and use ST loss to further constrain artifacts.\nDifferent from previous approaches, our network is capable of processing an\narbitrary number of input LDR images. Qualitative and quantitative experiments\ndemonstrate the effectiveness of the proposed method by comparing it with\nexisting state-of-the-art HDR deghosting models. Codes are available at\nhttps://github.com/pandayuanyu/HSTHdr.\n", "date": "2022-12-01", "categories": []}
{"id": "2211.14990", "title": "Near-filed SAR Image Restoration with Deep Learning Inverse Technique: A\n  Preliminary Study", "abstract": "  Benefiting from a relatively larger aperture's angle, and in combination with\na wide transmitting bandwidth, near-field synthetic aperture radar (SAR)\nprovides a high-resolution image of a target's scattering distribution-hot\nspots. Meanwhile, imaging result suffers inevitable degradation from sidelobes,\nclutters, and noises, hindering the information retrieval of the target. To\nrestore the image, current methods make simplified assumptions; for example,\nthe point spread function (PSF) is spatially consistent, the target consists of\nsparse point scatters, etc. Thus, they achieve limited restoration performance\nin terms of the target's shape, especially for complex targets. To address\nthese issues, a preliminary study is conducted on restoration with the recent\npromising deep learning inverse technique in this work. We reformulate the\ndegradation model into a spatially variable complex-convolution model, where\nthe near-field SAR's system response is considered. Adhering to it, a\nmodel-based deep learning network is designed to restore the image. A simulated\ndegraded image dataset from multiple complex target models is constructed to\nvalidate the network. All the images are formulated using the electromagnetic\nsimulation tool. Experiments on the dataset reveal their effectiveness.\nCompared with current methods, superior performance is achieved regarding the\ntarget's shape and energy estimation.\n", "date": "2022-11-28", "categories": []}
{"id": "2211.15002", "title": "A Model-data-driven Network Embedding Multidimensional Features for\n  Tomographic SAR Imaging", "abstract": "  Deep learning (DL)-based tomographic SAR imaging algorithms are gradually\nbeing studied. Typically, they use an unfolding network to mimic the iterative\ncalculation of the classical compressive sensing (CS)-based methods and process\neach range-azimuth unit individually. However, only one-dimensional features\nare effectively utilized in this way. The correlation between adjacent\nresolution units is ignored directly. To address that, we propose a new\nmodel-data-driven network to achieve tomoSAR imaging based on multi-dimensional\nfeatures. Guided by the deep unfolding methodology, a two-dimensional deep\nunfolding imaging network is constructed. On the basis of it, we add two 2D\nprocessing modules, both convolutional encoder-decoder structures, to enhance\nmulti-dimensional features of the imaging scene effectively. Meanwhile, to\ntrain the proposed multifeature-based imaging network, we construct a tomoSAR\nsimulation dataset consisting entirely of simulation data of buildings.\nExperiments verify the effectiveness of the model. Compared with the\nconventional CS-based FISTA method and DL-based gamma-Net method, the result of\nour proposed method has better performance on completeness while having decent\nimaging accuracy.\n", "date": "2022-11-28", "categories": []}
{"id": "2211.15311", "title": "NeuralMPS: Non-Lambertian Multispectral Photometric Stereo via Spectral\n  Reflectance Decomposition", "abstract": "  Multispectral photometric stereo(MPS) aims at recovering the surface normal\nof a scene from a single-shot multispectral image captured under multispectral\nilluminations. Existing MPS methods adopt the Lambertian reflectance model to\nmake the problem tractable, but it greatly limits their application to\nreal-world surfaces. In this paper, we propose a deep neural network named\nNeuralMPS to solve the MPS problem under general non-Lambertian spectral\nreflectances. Specifically, we present a spectral reflectance\ndecomposition(SRD) model to disentangle the spectral reflectance into geometric\ncomponents and spectral components. With this decomposition, we show that the\nMPS problem for surfaces with a uniform material is equivalent to the\nconventional photometric stereo(CPS) with unknown light intensities. In this\nway, NeuralMPS reduces the difficulty of the non-Lambertian MPS problem by\nleveraging the well-studied non-Lambertian CPS methods. Experiments on both\nsynthetic and real-world scenes demonstrate the effectiveness of our method.\n", "date": "2022-11-28", "categories": []}
{"id": "2211.15367", "title": "Few-shot Non-line-of-sight Imaging with Signal-surface Collaborative\n  Regularization", "abstract": "  The non-line-of-sight imaging technique aims to reconstruct targets from\nmultiply reflected light. For most existing methods, dense points on the relay\nsurface are raster scanned to obtain high-quality reconstructions, which\nrequires a long acquisition time. In this work, we propose a signal-surface\ncollaborative regularization (SSCR) framework that provides noise-robust\nreconstructions with a minimal number of measurements. Using Bayesian\ninference, we design joint regularizations of the estimated signal, the 3D\nvoxel-based representation of the objects, and the 2D surface-based description\nof the targets. To our best knowledge, this is the first work that combines\nregularizations in mixed dimensions for hidden targets. Experiments on\nsynthetic and experimental datasets illustrated the efficiency and robustness\nof the proposed method under both confocal and non-confocal settings. We report\nthe reconstruction of the hidden targets with complex geometric structures with\nonly $5 \\times 5$ confocal measurements from public datasets, indicating an\nacceleration of the conventional measurement process by a factor of 10000.\nBesides, the proposed method enjoys low time and memory complexities with\nsparse measurements. Our approach has great potential in real-time\nnon-line-of-sight imaging applications such as rescue operations and autonomous\ndriving.\n", "date": "2022-11-21", "categories": []}
{"id": "2211.15435", "title": "Hyperspectral Demosaicing of Snapshot Camera Images Using Deep Learning", "abstract": "  Spectral imaging technologies have rapidly evolved during the past decades.\nThe recent development of single-camera-one-shot techniques for hyperspectral\nimaging allows multiple spectral bands to be captured simultaneously (3x3, 4x4\nor 5x5 mosaic), opening up a wide range of applications. Examples include\nintraoperative imaging, agricultural field inspection and food quality\nassessment. To capture images across a wide spectrum range, i.e. to achieve\nhigh spectral resolution, the sensor design sacrifices spatial resolution. With\nincreasing mosaic size, this effect becomes increasingly detrimental.\nFurthermore, demosaicing is challenging. Without incorporating edge, shape, and\nobject information during interpolation, chromatic artifacts are likely to\nappear in the obtained images. Recent approaches use neural networks for\ndemosaicing, enabling direct information extraction from image data. However,\nobtaining training data for these approaches poses a challenge as well. This\nwork proposes a parallel neural network based demosaicing procedure trained on\na new ground truth dataset captured in a controlled environment by a\nhyperspectral snapshot camera with a 4x4 mosaic pattern. The dataset is a\ncombination of real captured scenes with images from publicly available data\nadapted to the 4x4 mosaic pattern. To obtain real world ground-truth data, we\nperformed multiple camera captures with 1-pixel shifts in order to compose the\nentire data cube. Experiments show that the proposed network outperforms\nstate-of-art networks.\n", "date": "2022-11-21", "categories": []}
{"id": "2211.15456", "title": "Noise-resilient approach for deep tomographic imaging", "abstract": "  We propose a noise-resilient deep reconstruction algorithm for X-ray\ntomography. Our approach shows strong noise resilience without obtaining noisy\ntraining examples. The advantages of our framework may further enable\nlow-photon tomographic imaging.\n", "date": "2022-11-22", "categories": []}
{"id": "2211.13452", "title": "Deep unfolding as iterative regularization for imaging inverse problems", "abstract": "  Recently, deep unfolding methods that guide the design of deep neural\nnetworks (DNNs) through iterative algorithms have received increasing attention\nin the field of inverse problems. Unlike general end-to-end DNNs, unfolding\nmethods have better interpretability and performance. However, to our\nknowledge, their accuracy and stability in solving inverse problems cannot be\nfully guaranteed. To bridge this gap, we modified the training procedure and\nproved that the unfolding method is an iterative regularization method. More\nprecisely, we jointly learn a convex penalty function adversarially by an\ninput-convex neural network (ICNN) to characterize the distance to a real data\nmanifold and train a DNN unfolded from the proximal gradient descent algorithm\nwith this learned penalty. Suppose the real data manifold intersects the\ninverse problem solutions with only the unique real solution. We prove that the\nunfolded DNN will converge to it stably. Furthermore, we demonstrate with an\nexample of MRI reconstruction that the proposed method outperforms conventional\nunfolding methods and traditional regularization methods in terms of\nreconstruction quality, stability and convergence speed.\n", "date": "2022-11-24", "categories": []}
{"id": "2211.13524", "title": "GAN Prior based Null-Space Learning for Consistent Super-Resolution", "abstract": "  Consistency and realness have always been the two critical issues of image\nsuper-resolution. While the realness has been dramatically improved with the\nuse of GAN prior, the state-of-the-art methods still suffer inconsistencies in\nlocal structures and colors (e.g., tooth and eyes). In this paper, we show that\nthese inconsistencies can be analytically eliminated by learning only the\nnull-space component while fixing the range-space part. Further, we design a\npooling-based decomposition (PD), a universal range-null space decomposition\nfor super-resolution tasks, which is concise, fast, and parameter-free. PD can\nbe easily applied to state-of-the-art GAN Prior based SR methods to eliminate\ntheir inconsistencies, neither compromising the realness nor bringing extra\nparameters or computational costs. Besides, our ablation studies reveal that PD\ncan replace pixel-wise losses for training and achieve better generalization\nperformance when facing unseen downsamplings or even real-world degradation.\nExperiments show that the use of PD refreshes state-of-the-art SR performance\nand speeds up the convergence of training up to 2~10 times.\n", "date": "2022-11-24", "categories": []}
{"id": "2211.13732", "title": "Deep Demosaicing for Polarimetric Filter Array Cameras", "abstract": "  Polarisation Filter Array (PFA) cameras allow the analysis of light\npolarisation state in a simple and cost-effective manner. Such filter arrays\nwork as the Bayer pattern for colour cameras, sharing similar advantages and\ndrawbacks. Among the others, the raw image must be demosaiced considering the\nlocal variations of the PFA and the characteristics of the imaged scene.\nNon-linear effects, like the cross-talk among neighbouring pixels, are\ndifficult to explicitly model and suggest the potential advantage of a\ndata-driven learning approach. However, the PFA cannot be removed from the\nsensor, making it difficult to acquire the ground-truth polarization state for\ntraining. In this work we propose a novel CNN-based model which directly\ndemosaics the raw camera image to a per-pixel Stokes vector. Our contribution\nis twofold. First, we propose a network architecture composed by a sequence of\nMosaiced Convolutions operating coherently with the local arrangement of the\ndifferent filters. Second, we introduce a new method, employing a consumer LCD\nscreen, to effectively acquire real-world data for training. The process is\ndesigned to be invariant by monitor gamma and external lighting conditions. We\nextensively compared our method against algorithmic and learning-based\ndemosaicing techniques, obtaining a consistently lower error especially in\nterms of polarisation angle.\n", "date": "2022-11-24", "categories": []}
{"id": "2211.13926", "title": "Generative Modeling in Sinogram Domain for Sparse-view CT Reconstruction", "abstract": "  The radiation dose in computed tomography (CT) examinations is harmful for\npatients but can be significantly reduced by intuitively decreasing the number\nof projection views. Reducing projection views usually leads to severe aliasing\nartifacts in reconstructed images. Previous deep learning (DL) techniques with\nsparse-view data require sparse-view/full-view CT image pairs to train the\nnetwork with supervised manners. When the number of projection view changes,\nthe DL network should be retrained with updated sparse-view/full-view CT image\npairs. To relieve this limitation, we present a fully unsupervised score-based\ngenerative model in sinogram domain for sparse-view CT reconstruction.\nSpecifically, we first train a score-based generative model on full-view\nsinogram data and use multi-channel strategy to form highdimensional tensor as\nthe network input to capture their prior distribution. Then, at the inference\nstage, the stochastic differential equation (SDE) solver and data-consistency\nstep were performed iteratively to achieve fullview projection. Filtered\nback-projection (FBP) algorithm was used to achieve the final image\nreconstruction. Qualitative and quantitative studies were implemented to\nevaluate the presented method with several CT data. Experimental results\ndemonstrated that our method achieved comparable or better performance than the\nsupervised learning counterparts.\n", "date": "2022-11-25", "categories": []}
{"id": "2211.12786", "title": "Nonlinear Equivariant Imaging: Learning Multi-Parametric Tissue Mapping\n  without Ground Truth for Compressive Quantitative MRI", "abstract": "  Current state-of-the-art reconstruction for quantitative tissue maps from\nfast, compressive, Magnetic Resonance Fingerprinting (MRF), use supervised deep\nlearning, with the drawback of requiring high-fidelity ground truth tissue map\ntraining data which is limited. This paper proposes NonLinear Equivariant\nImaging (NLEI), a self-supervised learning approach to eliminate the need for\nground truth for deep MRF image reconstruction. NLEI extends the recent\nEquivariant Imaging framework to nonlinear inverse problems such as MRF. Only\nfast, compressed-sampled MRF scans are used for training. NLEI learns tissue\nmapping using spatiotemporal priors: spatial priors are obtained from the\ninvariance of MRF data to a group of geometric image transformations, while\ntemporal priors are obtained from a nonlinear Bloch response model approximated\nby a pre-trained neural network. Tested retrospectively on two acquisition\nsettings, we observe that NLEI (self-supervised learning) closely approaches\nthe performance of supervised learning, despite not using ground truth during\ntraining.\n", "date": "2022-11-23", "categories": []}
{"id": "2211.12180", "title": "SRTGAN: Triplet Loss based Generative Adversarial Network for Real-World\n  Super-Resolution", "abstract": "  Many applications such as forensics, surveillance, satellite imaging, medical\nimaging, etc., demand High-Resolution (HR) images. However, obtaining an HR\nimage is not always possible due to the limitations of optical sensors and\ntheir costs. An alternative solution called Single Image Super-Resolution\n(SISR) is a software-driven approach that aims to take a Low-Resolution (LR)\nimage and obtain the HR image. Most supervised SISR solutions use ground truth\nHR image as a target and do not include the information provided in the LR\nimage, which could be valuable. In this work, we introduce Triplet Loss-based\nGenerative Adversarial Network hereafter referred as SRTGAN for Image\nSuper-Resolution problem on real-world degradation. We introduce a new\ntriplet-based adversarial loss function that exploits the information provided\nin the LR image by using it as a negative sample. Allowing the patch-based\ndiscriminator with access to both HR and LR images optimizes to better\ndifferentiate between HR and LR images; hence, improving the adversary.\nFurther, we propose to fuse the adversarial loss, content loss, perceptual\nloss, and quality loss to obtain Super-Resolution (SR) image with high\nperceptual fidelity. We validate the superior performance of the proposed\nmethod over the other existing methods on the RealSR dataset in terms of\nquantitative and qualitative metrics.\n", "date": "2022-11-22", "categories": []}
{"id": "2211.12340", "title": "DOLCE: A Model-Based Probabilistic Diffusion Framework for Limited-Angle\n  CT Reconstruction", "abstract": "  Limited-Angle Computed Tomography (LACT) is a non-destructive evaluation\ntechnique used in a variety of applications ranging from security to medicine.\nThe limited angle coverage in LACT is often a dominant source of severe\nartifacts in the reconstructed images, making it a challenging inverse problem.\nWe present DOLCE, a new deep model-based framework for LACT that uses a\nconditional diffusion model as an image prior. Diffusion models are a recent\nclass of deep generative models that are relatively easy to train due to their\nimplementation as image denoisers. DOLCE can form high-quality images from\nseverely under-sampled data by integrating data-consistency updates with the\nsampling updates of a diffusion model, which is conditioned on the transformed\nlimited-angle data. We show through extensive experimentation on several\nchallenging real LACT datasets that, the same pre-trained DOLCE model achieves\nthe SOTA performance on drastically different types of images. Additionally, we\nshow that, unlike standard LACT reconstruction methods, DOLCE naturally enables\nthe quantification of the reconstruction uncertainty by generating multiple\nsamples consistent with the measured data.\n", "date": "2022-11-22", "categories": []}
{"id": "2209.07911", "title": "Estimation of Optical Aberrations in 3D Microscopic Bioimages", "abstract": "  The quality of microscopy images often suffers from optical aberrations.\nThese aberrations and their associated point spread functions have to be\nquantitatively estimated to restore aberrated images. The recent\nstate-of-the-art method PhaseNet, based on a convolutional neural network, can\nquantify aberrations accurately but is limited to images of point light\nsources, e.g. fluorescent beads. In this research, we describe an extension of\nPhaseNet enabling its use on 3D images of biological samples. To this end, our\nmethod incorporates object-specific information into the simulated images used\nfor training the network. Further, we add a Python-based restoration of images\nvia Richardson-Lucy deconvolution. We demonstrate that the deconvolution with\nthe predicted PSF can not only remove the simulated aberrations but also\nimprove the quality of the real raw microscopic images with unknown residual\nPSF. We provide code for fast and convenient prediction and correction of\naberrations.\n", "date": "2022-09-16", "categories": []}
{"id": "2208.06787", "title": "HDR-Plenoxels: Self-Calibrating High Dynamic Range Radiance Fields", "abstract": "  We propose high dynamic range (HDR) radiance fields, HDR-Plenoxels, that\nlearn a plenoptic function of 3D HDR radiance fields, geometry information, and\nvarying camera settings inherent in 2D low dynamic range (LDR) images. Our\nvoxel-based volume rendering pipeline reconstructs HDR radiance fields with\nonly multi-view LDR images taken from varying camera settings in an end-to-end\nmanner and has a fast convergence speed. To deal with various cameras in\nreal-world scenarios, we introduce a tone mapping module that models the\ndigital in-camera imaging pipeline (ISP) and disentangles radiometric settings.\nOur tone mapping module allows us to render by controlling the radiometric\nsettings of each novel view. Finally, we build a multi-view dataset with\nvarying camera conditions, which fits our problem setting. Our experiments show\nthat HDR-Plenoxels can express detail and high-quality HDR novel views from\nonly LDR images with various cameras.\n", "date": "2022-08-14", "categories": []}
{"id": "2211.10026", "title": "DGD-cGAN: A Dual Generator for Image Dewatering and Restoration", "abstract": "  Underwater images are usually covered with a blue-greenish colour cast,\nmaking them distorted, blurry or low in contrast. This phenomenon occurs due to\nthe light attenuation given by the scattering and absorption in the water\ncolumn. In this paper, we present an image enhancement approach for dewatering\nwhich employs a conditional generative adversarial network (cGAN) with two\ngenerators. Our Dual Generator Dewatering cGAN (DGD-cGAN) removes the haze and\ncolour cast induced by the water column and restores the true colours of\nunderwater scenes whereby the effects of various attenuation and scattering\nphenomena that occur in underwater images are tackled by the two generators.\nThe first generator takes at input the underwater image and predicts the\ndewatered scene, while the second generator learns the underwater image\nformation process by implementing a custom loss function based upon the\ntransmission and the veiling light components of the image formation model. Our\nexperiments show that DGD-cGAN consistently delivers a margin of improvement as\ncompared with the state-of-the-art methods on several widely available\ndatasets.\n", "date": "2022-11-18", "categories": []}
{"id": "2211.07390", "title": "StereoISP: Rethinking Image Signal Processing for Dual Camera Systems", "abstract": "  Conventional image signal processing (ISP) frameworks are designed to\nreconstruct an RGB image from a single raw measurement. As multi-camera systems\nbecome increasingly popular these days, it is worth exploring improvements in\nISP frameworks by incorporating raw measurements from multiple cameras. This\nmanuscript is an intermediate progress report of a new ISP framework that is\nunder development, StereoISP. It employs raw measurements from a stereo camera\npair to generate a demosaicked, denoised RGB image by utilizing disparity\nestimated between the two views. We investigate StereoISP by testing the\nperformance on raw image pairs synthesized from stereo datasets. Our\npreliminary results show an improvement in the PSNR of the reconstructed RGB\nimage by at least 2dB on KITTI 2015 and drivingStereo datasets using ground\ntruth sparse disparity maps.\n", "date": "2022-11-11", "categories": []}
{"id": "2211.07871", "title": "DINER: Disorder-Invariant Implicit Neural Representation", "abstract": "  Implicit neural representation (INR) characterizes the attributes of a signal\nas a function of corresponding coordinates which emerges as a sharp weapon for\nsolving inverse problems. However, the capacity of INR is limited by the\nspectral bias in the network training. In this paper, we find that such a\nfrequency-related problem could be largely solved by re-arranging the\ncoordinates of the input signal, for which we propose the disorder-invariant\nimplicit neural representation (DINER) by augmenting a hash-table to a\ntraditional INR backbone. Given discrete signals sharing the same histogram of\nattributes and different arrangement orders, the hash-table could project the\ncoordinates into the same distribution for which the mapped signal can be\nbetter modeled using the subsequent INR network, leading to significantly\nalleviated spectral bias. Experiments not only reveal the generalization of the\nDINER for different INR backbones (MLP vs. SIREN) and various tasks\n(image/video representation, phase retrieval, and refractive index recovery)\nbut also show the superiority over the state-of-the-art algorithms both in\nquality and speed.\n", "date": "2022-11-15", "categories": []}
{"id": "2211.06502", "title": "Self-Supervised Isotropic Superresolution Fetal Brain MRI", "abstract": "  Superresolution T2-weighted fetal-brain magnetic-resonance imaging (FBMRI)\ntraditionally relies on the availability of several orthogonal low-resolution\nseries of 2-dimensional thick slices (volumes). In practice, only a few\nlow-resolution volumes are acquired. Thus, optimization-based\nimage-reconstruction methods require strong regularization using hand-crafted\nregularizers (e.g., TV). Yet, due to in utero fetal motion and the rapidly\nchanging fetal brain anatomy, the acquisition of the high-resolution images\nthat are required to train supervised learning methods is difficult. In this\npaper, we sidestep this difficulty by providing a proof of concept of a\nself-supervised single-volume superresolution framework for T2-weighted FBMRI\n(SAIR). We validate SAIR quantitatively in a motion-free simulated environment.\nOur results for different noise levels and resolution ratios suggest that SAIR\nis comparable to multiple-volume superresolution reconstruction methods. We\nalso evaluate SAIR qualitatively on clinical FBMRI data. The results suggest\nSAIR could be incorporated into current reconstruction pipelines.\n", "date": "2022-11-11", "categories": []}
{"id": "2211.06770", "title": "MicroISP: Processing 32MP Photos on Mobile Devices with Deep Learning", "abstract": "  While neural networks-based photo processing solutions can provide a better\nimage quality compared to the traditional ISP systems, their application to\nmobile devices is still very limited due to their very high computational\ncomplexity. In this paper, we present a novel MicroISP model designed\nspecifically for edge devices, taking into account their computational and\nmemory limitations. The proposed solution is capable of processing up to 32MP\nphotos on recent smartphones using the standard mobile ML libraries and\nrequiring less than 1 second to perform the inference, while for FullHD images\nit achieves real-time performance. The architecture of the model is flexible,\nallowing to adjust its complexity to devices of different computational power.\nTo evaluate the performance of the model, we collected a novel Fujifilm\nUltraISP dataset consisting of thousands of paired photos captured with a\nnormal mobile camera sensor and a professional 102MP medium-format FujiFilm\nGFX100 camera. The experiments demonstrated that, despite its compact size, the\nMicroISP model is able to provide comparable or better visual results than the\ntraditional mobile ISP systems, while outperforming the previously proposed\nefficient deep learning based solutions. Finally, this model is also compatible\nwith the latest mobile AI accelerators, achieving good runtime and low power\nconsumption on smartphone NPUs and APUs. The code, dataset and pre-trained\nmodels are available on the project website:\nhttps://people.ee.ethz.ch/~ihnatova/microisp.html\n", "date": "2022-11-08", "categories": []}
{"id": "2207.11876", "title": "nLMVS-Net: Deep Non-Lambertian Multi-View Stereo", "abstract": "  We introduce a novel multi-view stereo (MVS) method that can simultaneously\nrecover not just per-pixel depth but also surface normals, together with the\nreflectance of textureless, complex non-Lambertian surfaces captured under\nknown but natural illumination. Our key idea is to formulate MVS as an\nend-to-end learnable network, which we refer to as nLMVS-Net, that seamlessly\nintegrates radiometric cues to leverage surface normals as view-independent\nsurface features for learned cost volume construction and filtering. It first\nestimates surface normals as pixel-wise probability densities for each view\nwith a novel shape-from-shading network. These per-pixel surface normal\ndensities and the input multi-view images are then input to a novel cost volume\nfiltering network that learns to recover per-pixel depth and surface normal.\nThe reflectance is also explicitly estimated by alternating with geometry\nreconstruction. Extensive quantitative evaluations on newly established\nsynthetic and real-world datasets show that nLMVS-Net can robustly and\naccurately recover the shape and reflectance of complex objects in natural\nsettings.\n", "date": "2022-07-25", "categories": []}
{"id": "2208.06885", "title": "Global Priors Guided Modulation Network for Joint Super-Resolution and\n  Inverse Tone-Mapping", "abstract": "  Joint super-resolution and inverse tone-mapping (SR-ITM) aims to enhance the\nvisual quality of videos that have quality deficiencies in resolution and\ndynamic range. This problem arises when using 4K high dynamic range (HDR) TVs\nto watch a low-resolution standard dynamic range (LR SDR) video. Previous\nmethods that rely on learning local information typically cannot do well in\npreserving color conformity and long-range structural similarity, resulting in\nunnatural color transition and texture artifacts. In order to tackle these\nchallenges, we propose a global priors guided modulation network (GPGMNet) for\njoint SR-ITM. In particular, we design a global priors extraction module (GPEM)\nto extract color conformity prior and structural similarity prior that are\nbeneficial for ITM and SR tasks, respectively. To further exploit the global\npriors and preserve spatial information, we devise multiple global priors\nguided spatial-wise modulation blocks (GSMBs) with a few parameters for\nintermediate feature modulation, in which the modulation parameters are\ngenerated by the shared global priors and the spatial features map from the\nspatial pyramid convolution block (SPCB). With these elaborate designs, the\nGPGMNet can achieve higher visual quality with lower computational complexity.\nExtensive experiments demonstrate that our proposed GPGMNet is superior to the\nstate-of-the-art methods. Specifically, our proposed model exceeds the\nstate-of-the-art by 0.64 dB in PSNR, with 69$\\%$ fewer parameters and\n3.1$\\times$ speedup. The code will be released soon.\n", "date": "2022-08-14", "categories": []}
{"id": "2211.05094", "title": "3D Scene Inference from Transient Histograms", "abstract": "  Time-resolved image sensors that capture light at pico-to-nanosecond\ntimescales were once limited to niche applications but are now rapidly becoming\nmainstream in consumer devices. We propose low-cost and low-power imaging\nmodalities that capture scene information from minimal time-resolved image\nsensors with as few as one pixel. The key idea is to flood illuminate large\nscene patches (or the entire scene) with a pulsed light source and measure the\ntime-resolved reflected light by integrating over the entire illuminated area.\nThe one-dimensional measured temporal waveform, called \\emph{transient},\nencodes both distances and albedoes at all visible scene points and as such is\nan aggregate proxy for the scene's 3D geometry. We explore the viability and\nlimitations of the transient waveforms by themselves for recovering scene\ninformation, and also when combined with traditional RGB cameras. We show that\nplane estimation can be performed from a single transient and that using only a\nfew more it is possible to recover a depth map of the whole scene. We also show\ntwo proof-of-concept hardware prototypes that demonstrate the feasibility of\nour approach for compact, mobile, and budget-limited applications.\n", "date": "2022-11-09", "categories": []}
{"id": "2211.02973", "title": "Mixture-Net: Low-Rank Deep Image Prior Inspired by Mixture Models for\n  Spectral Image Recovery", "abstract": "  This paper proposes a non-data-driven deep neural network for spectral image\nrecovery problems such as denoising, single hyperspectral image\nsuper-resolution, and compressive spectral imaging reconstruction. Unlike\nprevious methods, the proposed approach, dubbed Mixture-Net, implicitly learns\nthe prior information through the network. Mixture-Net consists of a deep\ngenerative model whose layers are inspired by the linear and non-linear\nlow-rank mixture models, where the recovered image is composed of a weighted\nsum between the linear and non-linear decomposition. Mixture-Net also provides\na low-rank decomposition interpreted as the spectral image abundances and\nendmembers, helpful in achieving remote sensing tasks without running\nadditional routines. The experiments show the MixtureNet effectiveness\noutperforming state-of-the-art methods in recovery quality with the advantage\nof architecture interpretability.\n", "date": "2022-11-05", "categories": []}
{"id": "2211.03494", "title": "A Targeted Sampling Strategy for Compressive Cryo Focused Ion Beam\n  Scanning Electron Microscopy", "abstract": "  Cryo Focused Ion-Beam Scanning Electron Microscopy (cryo FIB-SEM) enables\nthree-dimensional and nanoscale imaging of biological specimens via a slice and\nview mechanism. The FIB-SEM experiments are, however, limited by a slow\n(typically, several hours) acquisition process and the high electron doses\nimposed on the beam sensitive specimen can cause damage. In this work, we\npresent a compressive sensing variant of cryo FIB-SEM capable of reducing the\noperational electron dose and increasing speed. We propose two Targeted\nSampling (TS) strategies that leverage the reconstructed image of the previous\nsample layer as a prior for designing the next subsampling mask. Our image\nrecovery is based on a blind Bayesian dictionary learning approach, i.e., Beta\nProcess Factor Analysis (BPFA). This method is experimentally viable due to our\nultra-fast GPU-based implementation of BPFA. Simulations on artificial\ncompressive FIB-SEM measurements validate the success of proposed methods: the\noperational electron dose can be reduced by up to 20 times. These methods have\nlarge implications for the cryo FIB-SEM community, in which the imaging of beam\nsensitive biological materials without beam damage is crucial.\n", "date": "2022-11-07", "categories": []}
{"id": "2211.01373", "title": "Interpretable Modeling and Reduction of Unknown Errors in Mechanistic\n  Operators", "abstract": "  Prior knowledge about the imaging physics provides a mechanistic forward\noperator that plays an important role in image reconstruction, although myriad\nsources of possible errors in the operator could negatively impact the\nreconstruction solutions. In this work, we propose to embed the traditional\nmechanistic forward operator inside a neural function, and focus on modeling\nand correcting its unknown errors in an interpretable manner. This is achieved\nby a conditional generative model that transforms a given mechanistic operator\nwith unknown errors, arising from a latent space of self-organizing clusters of\npotential sources of error generation. Once learned, the generative model can\nbe used in place of a fixed forward operator in any traditional\noptimization-based reconstruction process where, together with the inverse\nsolution, the error in prior mechanistic forward operator can be minimized and\nthe potential source of error uncovered. We apply the presented method to the\nreconstruction of heart electrical potential from body surface potential. In\ncontrolled simulation experiments and in-vivo real data experiments, we\ndemonstrate that the presented method allowed reduction of errors in the\nphysics-based forward operator and thereby delivered inverse reconstruction of\nheart-surface potential with increased accuracy.\n", "date": "2022-11-02", "categories": []}
{"id": "2211.00529", "title": "DOLPH: Diffusion Models for Phase Retrieval", "abstract": "  Phase retrieval refers to the problem of recovering an image from the\nmagnitudes of its complex-valued linear measurements. Since the problem is\nill-posed, the recovery requires prior knowledge on the unknown image. We\npresent DOLPH as a new deep model-based architecture for phase retrieval that\nintegrates an image prior specified using a diffusion model with a nonconvex\ndata-fidelity term for phase retrieval. Diffusion models are a recent class of\ndeep generative models that are relatively easy to train due to their\nimplementation as image denoisers. DOLPH reconstructs high-quality solutions by\nalternating data-consistency updates with the sampling step of a diffusion\nmodel. Our numerical results show the robustness of DOLPH to noise and its\nability to generate several candidate solutions given a set of measurements.\n", "date": "2022-11-01", "categories": []}
{"id": "2211.00725", "title": "LARO: Learned Acquisition and Reconstruction Optimization to accelerate\n  Quantitative Susceptibility Mapping", "abstract": "  Quantitative susceptibility mapping (QSM) involves acquisition and\nreconstruction of a series of images at multi-echo time points to estimate\ntissue field, which prolongs scan time and requires specific reconstruction\ntechnique. In this paper, we present our new framework, called Learned\nAcquisition and Reconstruction Optimization (LARO), which aims to accelerate\nthe multi-echo gradient echo (mGRE) pulse sequence for QSM. Our approach\ninvolves optimizing a Cartesian multi-echo k-space sampling pattern with a deep\nreconstruction network. Next, this optimized sampling pattern was implemented\nin an mGRE sequence using Cartesian fan-beam k-space segmenting and ordering\nfor prospective scans. Furthermore, we propose to insert a recurrent temporal\nfeature fusion module into the reconstruction network to capture signal\nredundancies along echo time. Our ablation studies show that both the optimized\nsampling pattern and proposed reconstruction strategy help improve the quality\nof the multi-echo image reconstructions. Generalization experiments show that\nLARO is robust on the test data with new pathologies and different sequence\nparameters. Our code is available at https://github.com/Jinwei1209/LARO.git.\n", "date": "2022-11-01", "categories": []}
{"id": "2211.00799", "title": "Practical Phase Retrieval Using Double Deep Image Priors", "abstract": "  Phase retrieval (PR) concerns the recovery of complex phases from complex\nmagnitudes. We identify the connection between the difficulty level and the\nnumber and variety of symmetries in PR problems. We focus on the most difficult\nfar-field PR (FFPR), and propose a novel method using double deep image priors.\nIn realistic evaluation, our method outperforms all competing methods by large\nmargins. As a single-instance method, our method requires no training data and\nminimal hyperparameter tuning, and hence enjoys good practicality.\n", "date": "2022-11-02", "categories": []}
{"id": "2211.00002", "title": "A Self-Supervised Approach to Reconstruction in Sparse X-Ray Computed\n  Tomography", "abstract": "  Computed tomography has propelled scientific advances in fields from biology\nto materials science. This technology allows for the elucidation of\n3-dimensional internal structure by the attenuation of x-rays through an object\nat different rotations relative to the beam. By imaging 2-dimensional\nprojections, a 3-dimensional object can be reconstructed through a\ncomputational algorithm. Imaging at a greater number of rotation angles allows\nfor improved reconstruction. However, taking more measurements increases the\nx-ray dose and may cause sample damage. Deep neural networks have been used to\ntransform sparse 2-D projection measurements to a 3-D reconstruction by\ntraining on a dataset of known similar objects. However, obtaining high-quality\nobject reconstructions for the training dataset requires high x-ray dose\nmeasurements that can destroy or alter the specimen before imaging is complete.\nThis becomes a chicken-and-egg problem: high-quality reconstructions cannot be\ngenerated without deep learning, and the deep neural network cannot be learned\nwithout the reconstructions. This work develops and validates a self-supervised\nprobabilistic deep learning technique, the physics-informed variational\nautoencoder, to solve this problem. A dataset consisting solely of sparse\nprojection measurements from each object is used to jointly reconstruct all\nobjects of the set. This approach has the potential to allow visualization of\nfragile samples with x-ray computed tomography. We release our code for\nreproducing our results at: https://github.com/vganapati/CT_PVAE .\n", "date": "2022-10-30", "categories": []}
{"id": "2211.00531", "title": "Robustness of Deep Equilibrium Architectures to Changes in the\n  Measurement Model", "abstract": "  Deep model-based architectures (DMBAs) are widely used in imaging inverse\nproblems to integrate physical measurement models and learned image priors.\nPlug-and-play priors (PnP) and deep equilibrium models (DEQ) are two DMBA\nframeworks that have received significant attention. The key difference between\nthe two is that the image prior in DEQ is trained by using a specific\nmeasurement model, while that in PnP is trained as a general image denoiser.\nThis difference is behind a common assumption that PnP is more robust to\nchanges in the measurement models compared to DEQ. This paper investigates the\nrobustness of DEQ priors to changes in the measurement models. Our results on\ntwo imaging inverse problems suggest that DEQ priors trained under mismatched\nmeasurement models outperform image denoisers.\n", "date": "2022-11-01", "categories": []}
{"id": "2210.16847", "title": "1st Place Solutions for UG2+ Challenge 2022 ATMOSPHERIC TURBULENCE\n  MITIGATION", "abstract": "  In this technical report, we briefly introduce the solution of our team\n''summer'' for Atomospheric Turbulence Mitigation in UG$^2$+ Challenge in CVPR\n2022. In this task, we propose a unified end-to-end framework to reconstruct a\nhigh quality image from distorted frames, which is mainly consists of a\nRestormer-based image reconstruction module and a NIMA-based image quality\nassessment module. Our framework is efficient and generic, which is adapted to\nboth hot-air image and text pattern. Moreover, we elaborately synthesize more\nthan 10 thousands of images to simulate atmospheric turbulence. And these\nimages improve the robustness of the model. Finally, we achieve the average\naccuracy of 98.53\\% on the reconstruction result of the text patterns, ranking\n1st on the final leaderboard.\n", "date": "2022-10-30", "categories": []}
{"id": "2210.15897", "title": "Single-Image HDR Reconstruction by Multi-Exposure Generation", "abstract": "  High dynamic range (HDR) imaging is an indispensable technique in modern\nphotography. Traditional methods focus on HDR reconstruction from multiple\nimages, solving the core problems of image alignment, fusion, and tone mapping,\nyet having a perfect solution due to ghosting and other visual artifacts in the\nreconstruction. Recent attempts at single-image HDR reconstruction show a\npromising alternative: by learning to map pixel values to their irradiance\nusing a neural network, one can bypass the align-and-merge pipeline completely\nyet still obtain a high-quality HDR image. In this work, we propose a weakly\nsupervised learning method that inverts the physical image formation process\nfor HDR reconstruction via learning to generate multiple exposures from a\nsingle image. Our neural network can invert the camera response to reconstruct\npixel irradiance before synthesizing multiple exposures and hallucinating\ndetails in under- and over-exposed regions from a single input image. To train\nthe network, we propose a representation loss, a reconstruction loss, and a\nperceptual loss applied on pairs of under- and over-exposure images and thus do\nnot require HDR images for training. Our experiments show that our proposed\nmodel can effectively reconstruct HDR images. Our qualitative and quantitative\nresults show that our method achieves state-of-the-art performance on the DrTMO\ndataset. Our code is available at\nhttps://github.com/VinAIResearch/single_image_hdr.\n", "date": "2022-10-28", "categories": []}
{"id": "2210.12548", "title": "JoJoNet: Joint-contrast and Joint-sampling-and-reconstruction Network\n  for Multi-contrast MRI", "abstract": "  Multi-contrast Magnetic Resonance Imaging (MRI) generates multiple medical\nimages with rich and complementary information for routine clinical use;\nhowever, it suffers from a long acquisition time. Recent works for accelerating\nMRI, mainly designed for single contrast, may not be optimal for multi-contrast\nscenario since the inherent correlations among the multi-contrast images are\nnot exploited. In addition, independent reconstruction of each contrast usually\ndoes not translate to optimal performance of downstream tasks. Motivated by\nthese aspects, in this paper we design an end-to-end framework for accelerating\nmulti-contrast MRI which simultaneously optimizes the entire MR imaging\nworkflow including sampling, reconstruction and downstream tasks to achieve the\nbest overall outcomes. The proposed framework consists of a sampling mask\ngenerator for each image contrast and a reconstructor exploiting the\ninter-contrast correlations with a recurrent structure which enables the\ninformation sharing in a holistic way. The sampling mask generator and the\nreconstructor are trained jointly across the multiple image contrasts. The\nacceleration ratio of each image contrast is also learnable and can be driven\nby a downstream task performance. We validate our approach on a multi-contrast\nbrain dataset and a multi-contrast knee dataset. Experiments show that (1) our\nframework consistently outperforms the baselines designed for single contrast\non both datasets; (2) our newly designed recurrent reconstruction network\neffectively improves the reconstruction quality for multi-contrast images; (3)\nthe learnable acceleration ratio improves the downstream task performance\nsignificantly. Overall, this work has potentials to open up new avenues for\noptimizing the entire multi-contrast MR imaging workflow.\n", "date": "2022-10-22", "categories": []}
{"id": "2209.07052", "title": "MIPI 2022 Challenge on Under-Display Camera Image Restoration: Methods\n  and Results", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for in-depth exchange of views from\nindustry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). To bridge the gap, we introduce the first MIPI\nchallenge including five tracks focusing on novel image sensors and imaging\nalgorithms. In this paper, we summarize and review the Under-Display Camera\n(UDC) Image Restoration track on MIPI 2022. In total, 167 participants were\nsuccessfully registered, and 19 teams submitted results in the final testing\nphase. The developed solutions in this challenge achieved state-of-the-art\nperformance on Under-Display Camera Image Restoration. A detailed description\nof all models developed in this challenge is provided in this paper. More\ndetails of this challenge and the link to the dataset can be found at\nhttps://github.com/mipi-challenge/MIPI2022.\n", "date": "2022-09-15", "categories": []}
{"id": "2210.13415", "title": "Deep Learning Approach for Dynamic Sampling for Multichannel Mass\n  Spectrometry Imaging", "abstract": "  Mass Spectrometry Imaging (MSI), using traditional rectilinear scanning,\ntakes hours to days for high spatial resolution acquisitions. Given that most\npixels within a sample's field of view are often neither relevant to underlying\nbiological structures nor chemically informative, MSI presents as a prime\ncandidate for integration with sparse and dynamic sampling algorithms. During a\nscan, stochastic models determine which locations probabilistically contain\ninformation critical to the generation of low-error reconstructions. Decreasing\nthe number of required physical measurements thereby minimizes overall\nacquisition times. A Deep Learning Approach for Dynamic Sampling (DLADS),\nutilizing a Convolutional Neural Network (CNN) and encapsulating molecular mass\nintensity distributions within a third dimension, demonstrates a simulated 70%\nthroughput improvement for Nanospray Desorption Electrospray Ionization\n(nano-DESI) MSI tissues. Evaluations are conducted between DLADS and a\nSupervised Learning Approach for Dynamic Sampling, with Least-Squares\nregression (SLADS-LS) and a Multi-Layer Perceptron (MLP) network (SLADS-Net).\nWhen compared with SLADS-LS, limited to a single m/z channel, as well as\nmultichannel SLADS-LS and SLADS-Net, DLADS respectively improves regression\nperformance by 36.7%, 7.0%, and 6.2%, resulting in gains to reconstruction\nquality of 6.0%, 2.1%, and 3.4% for acquisition of targeted m/z.\n", "date": "2022-10-24", "categories": []}
{"id": "2210.11511", "title": "Overexposure Mask Fusion: Generalizable Reverse ISP Multi-Step\n  Refinement", "abstract": "  With the advent of deep learning methods replacing the ISP in transforming\nsensor RAW readings into RGB images, numerous methodologies solidified into\nreal-life applications. Equally potent is the task of inverting this process\nwhich will have applications in enhancing computational photography tasks that\nare conducted in the RAW domain, addressing lack of available RAW data while\nreaping from the benefits of performing tasks directly on sensor readings. This\npaper's proposed methodology is a state-of-the-art solution to the task of RAW\nreconstruction, and the multi-step refinement process integrating an\noverexposure mask is novel in three ways: instead of from RGB to bayer, the\npipeline trains from RGB to demosaiced RAW allowing use of perceptual loss\nfunctions; the multi-step processes has greatly enhanced the performance of the\nbaseline U-Net from start to end; the pipeline is a generalizable process of\nrefinement that can enhance other high performance methodologies that support\nend-to-end learning.\n", "date": "2022-10-20", "categories": []}
{"id": "2210.11421", "title": "Application of artificial neural network to determine the thickness\n  profile of thin film", "abstract": "  In this paper, we introduce a novel artificial neural network (ANN) based\nscheme to estimate the thickness of thin films deposited on a given substrate.\nHere we consider the visible interference pattern between a plane wave and a\ndiverging wave reflected from the thin film surface that records the thickness\ninformation of the thin film. We assume a uniform thickness profile of the\nfilm. However, the thickness increases as the deposition takes place. We\nextract the intensity data along a line through the center of the interference\npattern. We train our network by using a number of such line information of\nknown thickness profiles. The performance of the trained network is then tested\nby estimating the thickness of unknown surfaces. The numerical simulation\nresults show that the proposed technique can be very much useful for automated\nmeasurement of thickness, quickly and in real time, during deposition\n", "date": "2022-09-24", "categories": []}
{"id": "2210.08036", "title": "Meta Transferring for Deblurring", "abstract": "  Most previous deblurring methods were built with a generic model trained on\nblurred images and their sharp counterparts. However, these approaches might\nhave sub-optimal deblurring results due to the domain gap between the training\nand test sets. This paper proposes a reblur-deblur meta-transferring scheme to\nrealize test-time adaptation without using ground truth for dynamic scene\ndeblurring. Since the ground truth is usually unavailable at inference time in\na real-world scenario, we leverage the blurred input video to find and use\nrelatively sharp patches as the pseudo ground truth. Furthermore, we propose a\nreblurring model to extract the homogenous blur from the blurred input and\ntransfer it to the pseudo-sharps to obtain the corresponding pseudo-blurred\npatches for meta-learning and test-time adaptation with only a few gradient\nupdates. Extensive experimental results show that our reblur-deblur\nmeta-learning scheme can improve state-of-the-art deblurring models on the DVD,\nREDS, and RealBlur benchmark datasets.\n", "date": "2022-10-14", "categories": []}
{"id": "2210.09074", "title": "Reversing Image Signal Processors by Reverse Style Transferring", "abstract": "  RAW image datasets are more suitable than the standard RGB image datasets for\nthe ill-posed inverse problems in low-level vision, but not common in the\nliterature. There are also a few studies to focus on mapping sRGB images to RAW\nformat. Mapping from sRGB to RAW format could be a relevant domain for reverse\nstyle transferring since the task is an ill-posed reversing problem. In this\nstudy, we seek an answer to the question: Can the ISP operations be modeled as\nthe style factor in an end-to-end learning pipeline? To investigate this idea,\nwe propose a novel architecture, namely RST-ISP-Net, for learning to reverse\nthe ISP operations with the help of adaptive feature normalization. We\nformulate this problem as a reverse style transferring and mostly follow the\npractice used in the prior work. We have participated in the AIM Reversed ISP\nchallenge with our proposed architecture. Results indicate that the idea of\nmodeling disruptive or modifying factors as style is still valid, but further\nimprovements are required to be competitive in such a challenge.\n", "date": "2022-10-17", "categories": []}
{"id": "2210.09293", "title": "Learning Texture Transformer Network for Light Field Super-Resolution", "abstract": "  Hand-held light field cameras suffer from low spatial resolution due to the\ninherent spatio-angular tradeoff. In this paper, we propose a method to improve\nthe spatial resolution of light field images with the aid of the Texture\nTransformer Network (TTSR). The proposed method consists of three modules: the\nfirst module produces an all-in focus high-resolution perspective image which\nserves as a reference image for the second module, i.e. TTSR, which in turn\nproduces a high-resolution light field. The last module refines the spatial\nresolution by imposing a light field prior. The results demonstrate around 4 dB\nto 6 dB PSNR gain over a bicubically resized light field image\n", "date": "2022-10-09", "categories": []}
{"id": "2210.07387", "title": "Amortized Inference for Heterogeneous Reconstruction in Cryo-EM", "abstract": "  Cryo-electron microscopy (cryo-EM) is an imaging modality that provides\nunique insights into the dynamics of proteins and other building blocks of\nlife. The algorithmic challenge of jointly estimating the poses, 3D structure,\nand conformational heterogeneity of a biomolecule from millions of noisy and\nrandomly oriented 2D projections in a computationally efficient manner,\nhowever, remains unsolved. Our method, cryoFIRE, performs ab initio\nheterogeneous reconstruction with unknown poses in an amortized framework,\nthereby avoiding the computationally expensive step of pose search while\nenabling the analysis of conformational heterogeneity. Poses and conformation\nare jointly estimated by an encoder while a physics-based decoder aggregates\nthe images into an implicit neural representation of the conformational space.\nWe show that our method can provide one order of magnitude speedup on datasets\ncontaining millions of images without any loss of accuracy. We validate that\nthe joint estimation of poses and conformations can be amortized over the size\nof the dataset. For the first time, we prove that an amortized method can\nextract interpretable dynamic information from experimental datasets.\n", "date": "2022-10-13", "categories": []}
{"id": "2210.05644", "title": "Simulating single-photon detector array sensors for depth imaging", "abstract": "  Single-Photon Avalanche Detector (SPAD) arrays are a rapidly emerging\ntechnology. These multi-pixel sensors have single-photon sensitivities and\npico-second temporal resolutions thus they can rapidly generate depth images\nwith millimeter precision. Such sensors are a key enabling technology for\nfuture autonomous systems as they provide guidance and situational awareness.\nHowever, to fully exploit the capabilities of SPAD array sensors, it is crucial\nto establish the quality of depth images they are able to generate in a wide\nrange of scenarios. Given a particular optical system and a finite image\nacquisition time, what is the best-case depth resolution and what are realistic\nimages generated by SPAD arrays? In this work, we establish a robust yet simple\nnumerical procedure that rapidly establishes the fundamental limits to depth\nimaging with SPAD arrays under real world conditions. Our approach accurately\ngenerates realistic depth images in a wide range of scenarios, allowing the\nperformance of an optical depth imaging system to be established without the\nneed for costly and laborious field testing. This procedure has applications in\nobject detection and tracking for autonomous systems and could be easily\nextended to systems for underwater imaging or for imaging around corners.\n", "date": "2022-10-07", "categories": []}
{"id": "2210.03837", "title": "Self-Supervised Deep Equilibrium Models for Inverse Problems with\n  Theoretical Guarantees", "abstract": "  Deep equilibrium models (DEQ) have emerged as a powerful alternative to deep\nunfolding (DU) for image reconstruction. DEQ models-implicit neural networks\nwith effectively infinite number of layers-were shown to achieve\nstate-of-the-art image reconstruction without the memory complexity associated\nwith DU. While the performance of DEQ has been widely investigated, the\nexisting work has primarily focused on the settings where groundtruth data is\navailable for training. We present self-supervised deep equilibrium model\n(SelfDEQ) as the first self-supervised reconstruction framework for training\nmodel-based implicit networks from undersampled and noisy MRI measurements. Our\ntheoretical results show that SelfDEQ can compensate for unbalanced sampling\nacross multiple acquisitions and match the performance of fully supervised DEQ.\nOur numerical results on in-vivo MRI data show that SelfDEQ leads to\nstate-of-the-art performance using only undersampled and noisy training data.\n", "date": "2022-10-07", "categories": []}
{"id": "2210.03904", "title": "LW-ISP: A Lightweight Model with ISP and Deep Learning", "abstract": "  The deep learning (DL)-based methods of low-level tasks have many advantages\nover the traditional camera in terms of hardware prospects, error accumulation\nand imaging effects. Recently, the application of deep learning to replace the\nimage signal processing (ISP) pipeline has appeared one after another; however,\nthere is still a long way to go towards real landing. In this paper, we show\nthe possibility of learning-based method to achieve real-time high-performance\nprocessing in the ISP pipeline. We propose LW-ISP, a novel architecture\ndesigned to implicitly learn the image mapping from RAW data to RGB image.\nBased on U-Net architecture, we propose the fine-grained attention module and a\nplug-and-play upsampling block suitable for low-level tasks. In particular, we\ndesign a heterogeneous distillation algorithm to distill the implicit features\nand reconstruction information of the clean image, so as to guide the learning\nof the student model. Our experiments demonstrate that LW-ISP has achieved a\n0.38 dB improvement in PSNR compared to the previous best method, while the\nmodel parameters and calculation have been reduced by 23 times and 81 times.\nThe inference efficiency has been accelerated by at least 15 times. Without\nbells and whistles, LW-ISP has achieved quite competitive results in ISP\nsubtasks including image denoising and enhancement.\n", "date": "2022-10-08", "categories": []}
{"id": "2210.04429", "title": "DeepHS-HDRVideo: Deep High Speed High Dynamic Range Video Reconstruction", "abstract": "  Due to hardware constraints, standard off-the-shelf digital cameras suffers\nfrom low dynamic range (LDR) and low frame per second (FPS) outputs. Previous\nworks in high dynamic range (HDR) video reconstruction uses sequence of\nalternating exposure LDR frames as input, and align the neighbouring frames\nusing optical flow based networks. However, these methods often result in\nmotion artifacts in challenging situations. This is because, the alternate\nexposure frames have to be exposure matched in order to apply alignment using\noptical flow. Hence, over-saturation and noise in the LDR frames results in\ninaccurate alignment. To this end, we propose to align the input LDR frames\nusing a pre-trained video frame interpolation network. This results in better\nalignment of LDR frames, since we circumvent the error-prone exposure matching\nstep, and directly generate intermediate missing frames from the same exposure\ninputs. Furthermore, it allows us to generate high FPS HDR videos by\nrecursively interpolating the intermediate frames. Through this work, we\npropose to use video frame interpolation for HDR video reconstruction, and\npresent the first method to generate high FPS HDR videos. Experimental results\ndemonstrate the efficacy of the proposed framework against optical flow based\nalignment methods, with an absolute improvement of 2.4 PSNR value on standard\nHDR video datasets [1], [2] and further benchmark our method for high FPS HDR\nvideo generation.\n", "date": "2022-10-10", "categories": []}
{"id": "2210.04553", "title": "SiNeRF: Sinusoidal Neural Radiance Fields for Joint Pose Estimation and\n  Scene Reconstruction", "abstract": "  NeRFmm is the Neural Radiance Fields (NeRF) that deal with Joint Optimization\ntasks, i.e., reconstructing real-world scenes and registering camera parameters\nsimultaneously. Despite NeRFmm producing precise scene synthesis and pose\nestimations, it still struggles to outperform the full-annotated baseline on\nchallenging scenes. In this work, we identify that there exists a systematic\nsub-optimality in joint optimization and further identify multiple potential\nsources for it. To diminish the impacts of potential sources, we propose\nSinusoidal Neural Radiance Fields (SiNeRF) that leverage sinusoidal activations\nfor radiance mapping and a novel Mixed Region Sampling (MRS) for selecting ray\nbatch efficiently. Quantitative and qualitative results show that compared to\nNeRFmm, SiNeRF achieves comprehensive significant improvements in image\nsynthesis quality and pose estimation accuracy. Codes are available at\nhttps://github.com/yitongx/sinerf.\n", "date": "2022-10-10", "categories": []}
{"id": "2210.03400", "title": "Computational imaging with the human brain", "abstract": "  Brain-computer interfaces (BCIs) are enabling a range of new possibilities\nand routes for augmenting human capability. Here, we propose BCIs as a route\ntowards forms of computation, i.e. computational imaging, that blend the brain\nwith external silicon processing. We demonstrate ghost imaging of a hidden\nscene using the human visual system that is combined with an adaptive\ncomputational imaging scheme. This is achieved through a projection pattern\n`carving' technique that relies on real-time feedback from the brain to modify\npatterns at the light projector, thus enabling more efficient and higher\nresolution imaging. This brain-computer connectivity demonstrates a form of\naugmented human computation that could in the future extend the sensing range\nof human vision and provide new approaches to the study of the neurophysics of\nhuman perception. As an example, we illustrate a simple experiment whereby\nimage reconstruction quality is affected by simultaneous conscious processing\nand readout of the perceived light intensities.\n", "date": "2022-10-07", "categories": []}
{"id": "2210.03417", "title": "A Simple Plugin for Transforming Images to Arbitrary Scales", "abstract": "  Existing models on super-resolution often specialized for one scale,\nfundamentally limiting their use in practical scenarios. In this paper, we aim\nto develop a general plugin that can be inserted into existing super-resolution\nmodels, conveniently augmenting their ability towards Arbitrary Resolution\nImage Scaling, thus termed ARIS. We make the following contributions: (i) we\npropose a transformer-based plugin module, which uses spatial coordinates as\nquery, iteratively attend the low-resolution image feature through\ncross-attention, and output visual feature for the queried spatial location,\nresembling an implicit representation for images; (ii) we introduce a novel\nself-supervised training scheme, that exploits consistency constraints to\neffectively augment the model's ability for upsampling images towards unseen\nscales, i.e. ground-truth high-resolution images are not available; (iii)\nwithout loss of generality, we inject the proposed ARIS plugin module into\nseveral existing models, namely, IPT, SwinIR, and HAT, showing that the\nresulting models can not only maintain their original performance on fixed\nscale factor but also extrapolate to unseen scales, substantially outperforming\nexisting any-scale super-resolution models on standard benchmarks, e.g.\nUrban100, DIV2K, etc.\n", "date": "2022-10-07", "categories": []}
{"id": "2210.03040", "title": "Rolling Shutter Inversion: Bring Rolling Shutter Images to High\n  Framerate Global Shutter Video", "abstract": "  A single rolling-shutter (RS) image may be viewed as a row-wise combination\nof a sequence of global-shutter (GS) images captured by a (virtual) moving GS\ncamera within the exposure duration. Although RS cameras are widely used, the\nRS effect causes obvious image distortion especially in the presence of fast\ncamera motion, hindering downstream computer vision tasks. In this paper, we\npropose to invert the RS image capture mechanism, i.e., recovering a continuous\nhigh framerate GS video from two time-consecutive RS frames. We call this task\nthe RS temporal super-resolution (RSSR) problem. The RSSR is a very challenging\ntask, and to our knowledge, no practical solution exists to date. This paper\npresents a novel deep-learning based solution. By leveraging the multi-view\ngeometry relationship of the RS imaging process, our learning-based framework\nsuccessfully achieves high framerate GS generation. Specifically, three novel\ncontributions can be identified: (i) novel formulations for bidirectional RS\nundistortion flows under constant velocity as well as constant acceleration\nmotion model. (ii) a simple linear scaling operation, which bridges the RS\nundistortion flow and regular optical flow. (iii) a new mutual conversion\nscheme between varying RS undistortion flows that correspond to different\nscanlines. Our method also exploits the underlying spatial-temporal geometric\nrelationships within a deep learning framework, where no additional supervision\nis required beyond the necessary middle-scanline GS image. Building upon these\ncontributions, we represent the very first rolling-shutter temporal\nsuper-resolution deep-network that is able to recover high framerate GS videos\nfrom just two RS frames. Extensive experimental results on both synthetic and\nreal data show that our proposed method can produce high-quality GS image\nsequences with rich details, outperforming the state-of-the-art methods.\n", "date": "2022-10-06", "categories": []}
{"id": "2207.11050", "title": "Graph Spatio-Spectral Total Variation Model for Hyperspectral Image\n  Denoising", "abstract": "  The spatio-spectral total variation (SSTV) model has been widely used as an\neffective regularization of hyperspectral images (HSI) for various applications\nsuch as mixed noise removal. However, since SSTV computes local spatial\ndifferences uniformly, it is difficult to remove noise while preserving complex\nspatial structures with fine edges and textures, especially in situations of\nhigh noise intensity. To solve this problem, we propose a new TV-type\nregularization called Graph-SSTV (GSSTV), which generates a graph explicitly\nreflecting the spatial structure of the target HSI from noisy HSIs and\nincorporates a weighted spatial difference operator designed based on this\ngraph. Furthermore, we formulate the mixed noise removal problem as a convex\noptimization problem involving GSSTV and develop an efficient algorithm based\non the primal-dual splitting method to solve this problem. Finally, we\ndemonstrate the effectiveness of GSSTV compared with existing HSI\nregularization models through experiments on mixed noise removal. The source\ncode will be available at https://www.mdi.c.titech.ac.jp/publications/gsstv.\n", "date": "2022-07-22", "categories": []}
{"id": "2208.03042", "title": "Low-Light Hyperspectral Image Enhancement", "abstract": "  Due to inadequate energy captured by the hyperspectral camera sensor in poor\nillumination conditions, low-light hyperspectral images (HSIs) usually suffer\nfrom low visibility, spectral distortion, and various noises. A range of HSI\nrestoration methods have been developed, yet their effectiveness in enhancing\nlow-light HSIs is constrained. This work focuses on the low-light HSI\nenhancement task, which aims to reveal the spatial-spectral information hidden\nin darkened areas. To facilitate the development of low-light HSI processing,\nwe collect a low-light HSI (LHSI) dataset of both indoor and outdoor scenes.\nBased on Laplacian pyramid decomposition and reconstruction, we developed an\nend-to-end data-driven low-light HSI enhancement (HSIE) approach trained on the\nLHSI dataset. With the observation that illumination is related to the\nlow-frequency component of HSI, while textural details are closely correlated\nto the high-frequency component, the proposed HSIE is designed to have two\nbranches. The illumination enhancement branch is adopted to enlighten the\nlow-frequency component with reduced resolution. The high-frequency refinement\nbranch is utilized for refining the high-frequency component via a predicted\nmask. In addition, to improve information flow and boost performance, we\nintroduce an effective channel attention block (CAB) with residual dense\nconnection, which served as the basic block of the illumination enhancement\nbranch. The effectiveness and efficiency of HSIE both in quantitative\nassessment measures and visual effects are demonstrated by experimental results\non the LHSI dataset. According to the classification performance on the remote\nsensing Indian Pines dataset, downstream tasks benefit from the enhanced HSI.\nDatasets and codes are available:\n\\href{https://github.com/guanguanboy/HSIE}{https://github.com/guanguanboy/HSIE}.\n", "date": "2022-08-05", "categories": []}
{"id": "2210.01436", "title": "Non-learning Stereo-aided Depth Completion under Mis-projection via\n  Selective Stereo Matching", "abstract": "  We propose a non-learning depth completion method for a sparse depth map\ncaptured using a light detection and ranging (LiDAR) sensor guided by a pair of\nstereo images. Generally, conventional stereo-aided depth completion methods\nhave two limiations. (i) They assume the given sparse depth map is accurately\naligned to the input image, whereas the alignment is difficult to achieve in\npractice. (ii) They have limited accuracy in the long range because the depth\nis estimated by pixel disparity. To solve the abovementioned limitations, we\npropose selective stereo matching (SSM) that searches the most appropriate\ndepth value for each image pixel from its neighborly projected LiDAR points\nbased on an energy minimization framework. This depth selection approach can\nhandle any type of mis-projection. Moreover, SSM has an advantage in terms of\nlong-range depth accuracy because it directly uses the LiDAR measurement rather\nthan the depth acquired from the stereo. SSM is a discrete process; thus, we\napply variational smoothing with binary anisotropic diffusion tensor (B-ADT) to\ngenerate a continuous depth map while preserving depth discontinuity across\nobject boundaries. Experimentally, compared with the previous state-of-the-art\nstereo-aided depth completion, the proposed method reduced the mean absolute\nerror (MAE) of the depth estimation to 0.65 times and demonstrated\napproximately twice more accurate estimation in the long range. Moreover, under\nvarious LiDAR-camera calibration errors, the proposed method reduced the depth\nestimation MAE to 0.34-0.93 times from previous depth completion methods.\n", "date": "2022-10-04", "categories": []}
{"id": "2210.01543", "title": "Amortized Bayesian Inference of GISAXS Data with Normalizing Flows", "abstract": "  Grazing-Incidence Small-Angle X-ray Scattering (GISAXS) is a modern imaging\ntechnique used in material research to study nanoscale materials.\nReconstruction of the parameters of an imaged object imposes an ill-posed\ninverse problem that is further complicated when only an in-plane GISAXS signal\nis available. Traditionally used inference algorithms such as Approximate\nBayesian Computation (ABC) rely on computationally expensive scattering\nsimulation software, rendering analysis highly time-consuming. We propose a\nsimulation-based framework that combines variational auto-encoders and\nnormalizing flows to estimate the posterior distribution of object parameters\ngiven its GISAXS data. We apply the inference pipeline to experimental data and\ndemonstrate that our method reduces the inference cost by orders of magnitude\nwhile producing consistent results with ABC.\n", "date": "2022-10-04", "categories": []}
{"id": "2209.12394", "title": "Multi-stage image denoising with the wavelet transform", "abstract": "  Deep convolutional neural networks (CNNs) are used for image denoising via\nautomatically mining accurate structure information. However, most of existing\nCNNs depend on enlarging depth of designed networks to obtain better denoising\nperformance, which may cause training difficulty. In this paper, we propose a\nmulti-stage image denoising CNN with the wavelet transform (MWDCNN) via three\nstages, i.e., a dynamic convolutional block (DCB), two cascaded wavelet\ntransform and enhancement blocks (WEBs) and a residual block (RB). DCB uses a\ndynamic convolution to dynamically adjust parameters of several convolutions\nfor making a tradeoff between denoising performance and computational costs.\nWEB uses a combination of signal processing technique (i.e., wavelet\ntransformation) and discriminative learning to suppress noise for recovering\nmore detailed information in image denoising. To further remove redundant\nfeatures, RB is used to refine obtained features for improving denoising\neffects and reconstruct clean images via improved residual dense architectures.\nExperimental results show that the proposed MWDCNN outperforms some popular\ndenoising methods in terms of quantitative and qualitative analysis. Codes are\navailable at https://github.com/hellloxiaotian/MWDCNN.\n", "date": "2022-09-26", "categories": []}
{"id": "2209.14973", "title": "Deep Unfolding for Iterative Stripe Noise Removal", "abstract": "  The non-uniform photoelectric response of infrared imaging systems results in\nfixed-pattern stripe noise being superimposed on infrared images, which\nseverely reduces image quality. As the applications of degraded infrared images\nare limited, it is crucial to effectively preserve original details. Existing\nimage destriping methods struggle to concurrently remove all stripe noise\nartifacts, preserve image details and structures, and balance real-time\nperformance. In this paper we propose a novel algorithm for destriping degraded\nimages, which takes advantage of neighbouring column signal correlation to\nremove independent column stripe noise. This is achieved through an iterative\ndeep unfolding algorithm where the estimated noise of one network iteration is\nused as input to the next iteration. This progression substantially reduces the\nsearch space of possible function approximations, allowing for efficient\ntraining on larger datasets. The proposed method allows for a more precise\nestimation of stripe noise to preserve scene details more accurately. Extensive\nexperimental results demonstrate that the proposed model outperforms existing\ndestriping methods on artificially corrupted images on both quantitative and\nqualitative assessments.\n", "date": "2022-09-27", "categories": []}
{"id": "2209.13875", "title": "A General Scattering Phase Function for Inverse Rendering", "abstract": "  We tackle the problem of modeling light scattering in homogeneous translucent\nmaterial and estimating its scattering parameters. A scattering phase function\nis one of such parameters which affects the distribution of scattered\nradiation. It is the most complex and challenging parameter to be modeled in\npractice, and empirical phase functions are usually used. Empirical phase\nfunctions (such as Henyey-Greenstein (HG) phase function or its modified ones)\nare usually presented and limited to a specific range of scattering materials.\nThis limitation raises concern for an inverse rendering problem where the\ntarget material is generally unknown. In such a situation, a more general phase\nfunction is preferred. Although there exists such a general phase function in\nthe polynomial form using a basis such as Legendre polynomials\n\\cite{Fowler1983}, inverse rendering with this phase function is not\nstraightforward. This is because the base polynomials may be negative\nsomewhere, while a phase function cannot. This research presents a novel\ngeneral phase function that can avoid this issue and an inverse rendering\napplication using this phase function. The proposed phase function was\npositively evaluated with a wide range of materials modeled with Mie scattering\ntheory. The scattering parameters estimation with the proposed phase function\nwas evaluated with simulation and real-world experiments.\n", "date": "2022-09-28", "categories": []}
{"id": "2209.07530", "title": "MIPI 2022 Challenge on RGBW Sensor Fusion: Dataset and Report", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for in-depth exchange of views from\nindustry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). To bridge the gap, we introduce the first MIPI\nchallenge, including five tracks focusing on novel image sensors and imaging\nalgorithms. In this paper, RGBW Joint Fusion and Denoise, one of the five\ntracks, working on the fusion of binning-mode RGBW to Bayer, is introduced. The\nparticipants were provided with a new dataset including 70 (training) and 15\n(validation) scenes of high-quality RGBW and Bayer pairs. In addition, for each\nscene, RGBW of different noise levels was provided at 24dB and 42dB. All the\ndata were captured using an RGBW sensor in both outdoor and indoor conditions.\nThe final results are evaluated using objective metrics, including PSNR, SSIM},\nLPIPS, and KLD. A detailed description of all models developed in this\nchallenge is provided in this paper. More details of this challenge and the\nlink to the dataset can be found at https://github.com/mipi-challenge/MIPI2022.\n", "date": "2022-09-15", "categories": []}
{"id": "2209.13106", "title": "Simultaneous Acquisition of High Quality RGB Image and Polarization\n  Information using a Sparse Polarization Sensor", "abstract": "  This paper proposes a novel polarization sensor structure and network\narchitecture to obtain a high-quality RGB image and polarization information.\nConventional polarization sensors can simultaneously acquire RGB images and\npolarization information, but the polarizers on the sensor degrade the quality\nof the RGB images. There is a trade-off between the quality of the RGB image\nand polarization information as fewer polarization pixels reduce the\ndegradation of the RGB image but decrease the resolution of polarization\ninformation. Therefore, we propose an approach that resolves the trade-off by\nsparsely arranging polarization pixels on the sensor and compensating for\nlow-resolution polarization information with higher resolution using the RGB\nimage as a guide. Our proposed network architecture consists of an RGB image\nrefinement network and a polarization information compensation network. We\nconfirmed the superiority of our proposed network in compensating the\ndifferential component of polarization intensity by comparing its performance\nwith state-of-the-art methods for similar tasks: depth completion. Furthermore,\nwe confirmed that our approach could simultaneously acquire higher quality RGB\nimages and polarization information than conventional polarization sensors,\nresolving the trade-off between the quality of RGB images and polarization\ninformation. The baseline code and newly generated real and synthetic\nlarge-scale polarization image datasets are available for further research and\ndevelopment.\n", "date": "2022-09-27", "categories": []}
{"id": "2209.13362", "title": "DELTAR: Depth Estimation from a Light-weight ToF Sensor and RGB Image", "abstract": "  Light-weight time-of-flight (ToF) depth sensors are small, cheap, low-energy\nand have been massively deployed on mobile devices for the purposes like\nautofocus, obstacle detection, etc. However, due to their specific measurements\n(depth distribution in a region instead of the depth value at a certain pixel)\nand extremely low resolution, they are insufficient for applications requiring\nhigh-fidelity depth such as 3D reconstruction. In this paper, we propose\nDELTAR, a novel method to empower light-weight ToF sensors with the capability\nof measuring high resolution and accurate depth by cooperating with a color\nimage. As the core of DELTAR, a feature extractor customized for depth\ndistribution and an attention-based neural architecture is proposed to fuse the\ninformation from the color and ToF domain efficiently. To evaluate our system\nin real-world scenarios, we design a data collection device and propose a new\napproach to calibrate the RGB camera and ToF sensor. Experiments show that our\nmethod produces more accurate depth than existing frameworks designed for depth\ncompletion and depth super-resolution and achieves on par performance with a\ncommodity-level RGB-D sensor. Code and data are available at\nhttps://zju3dv.github.io/deltar/.\n", "date": "2022-09-27", "categories": []}
{"id": "2209.11002", "title": "Entropic Descent Archetypal Analysis for Blind Hyperspectral Unmixing", "abstract": "  In this paper, we introduce a new algorithm based on archetypal analysis for\nblind hyperspectral unmixing, assuming linear mixing of endmembers. Archetypal\nanalysis is a natural formulation for this task. This method does not require\nthe presence of pure pixels (i.e., pixels containing a single material) but\ninstead represents endmembers as convex combinations of a few pixels present in\nthe original hyperspectral image. Our approach leverages an entropic gradient\ndescent strategy, which (i) provides better solutions for hyperspectral\nunmixing than traditional archetypal analysis algorithms, and (ii) leads to\nefficient GPU implementations. Since running a single instance of our algorithm\nis fast, we also propose an ensembling mechanism along with an appropriate\nmodel selection procedure that make our method robust to hyper-parameter\nchoices while keeping the computational complexity reasonable. By using six\nstandard real datasets, we show that our approach outperforms state-of-the-art\nmatrix factorization and recent deep learning methods. We also provide an\nopen-source PyTorch implementation: https://github.com/inria-thoth/EDAA.\n", "date": "2022-09-22", "categories": []}
{"id": "2209.11772", "title": "A direct time-of-flight image sensor with in-pixel surface detection and\n  dynamic vision", "abstract": "  3D flash LIDAR is an alternative to the traditional scanning LIDAR systems,\npromising precise depth imaging in a compact form factor, and free of moving\nparts, for applications such as self-driving cars, robotics and augmented\nreality (AR). Typically implemented using single-photon, direct time-of-flight\n(dToF) receivers in image sensor format, the operation of the devices can be\nhindered by the large number of photon events needing to be processed and\ncompressed in outdoor scenarios, limiting frame rates and scalability to larger\narrays. We here present a 64x32 pixel (256x128 SPAD) dToF imager that overcomes\nthese limitations by using pixels with embedded histogramming, which lock onto\nand track the return signal. This reduces the size of output data frames\nconsiderably, enabling maximum frame rates in the 10 kFPS range or 100 kFPS for\ndirect depth readings. The sensor offers selective readout of pixels detecting\nsurfaces, or those sensing motion, leading to reduced power consumption and\noff-chip processing requirements. We demonstrate the application of the sensor\nin mid-range LIDAR.\n", "date": "2022-09-23", "categories": []}
{"id": "2209.12155", "title": "Discriminative feature encoding for intrinsic image decomposition", "abstract": "  Intrinsic image decomposition is an important and long-standing computer\nvision problem. Given an input image, recovering the physical scene properties\nis ill-posed. Several physically motivated priors have been used to restrict\nthe solution space of the optimization problem for intrinsic image\ndecomposition. This work takes advantage of deep learning, and shows that it\ncan solve this challenging computer vision problem with high efficiency. The\nfocus lies in the feature encoding phase to extract discriminative features for\ndifferent intrinsic layers from an input image. To achieve this goal, we\nexplore the distinctive characteristics of different intrinsic components in\nthe high dimensional feature embedding space. We define feature distribution\ndivergence to efficiently separate the feature vectors of different intrinsic\ncomponents. The feature distributions are also constrained to fit the real ones\nthrough a feature distribution consistency. In addition, a data refinement\napproach is provided to remove data inconsistency from the Sintel dataset,\nmaking it more suitable for intrinsic image decomposition. Our method is also\nextended to intrinsic video decomposition based on pixel-wise correspondences\nbetween adjacent frames. Experimental results indicate that our proposed\nnetwork structure can outperform the existing state-of-the-art.\n", "date": "2022-09-25", "categories": []}
{"id": "2209.11455", "title": "Modular Degradation Simulation and Restoration for Under-Display Camera", "abstract": "  Under-display camera (UDC) provides an elegant solution for full-screen\nsmartphones. However, UDC captured images suffer from severe degradation since\nsensors lie under the display. Although this issue can be tackled by image\nrestoration networks, these networks require large-scale image pairs for\ntraining. To this end, we propose a modular network dubbed MPGNet trained using\nthe generative adversarial network (GAN) framework for simulating UDC imaging.\nSpecifically, we note that the UDC imaging degradation process contains\nbrightness attenuation, blurring, and noise corruption. Thus we model each\ndegradation with a characteristic-related modular network, and all modular\nnetworks are cascaded to form the generator. Together with a pixel-wise\ndiscriminator and supervised loss, we can train the generator to simulate the\nUDC imaging degradation process. Furthermore, we present a Transformer-style\nnetwork named DWFormer for UDC image restoration. For practical purposes, we\nuse depth-wise convolution instead of the multi-head self-attention to\naggregate local spatial information. Moreover, we propose a novel channel\nattention module to aggregate global information, which is critical for\nbrightness recovery. We conduct evaluations on the UDC benchmark, and our\nmethod surpasses the previous state-of-the-art models by 1.23 dB on the P-OLED\ntrack and 0.71 dB on the T-OLED track, respectively.\n", "date": "2022-09-23", "categories": []}
{"id": "2209.11729", "title": "Dual-Cycle: Self-Supervised Dual-View Fluorescence Microscopy Image\n  Reconstruction using CycleGAN", "abstract": "  Three-dimensional fluorescence microscopy often suffers from anisotropy,\nwhere the resolution along the axial direction is lower than that within the\nlateral imaging plane. We address this issue by presenting Dual-Cycle, a new\nframework for joint deconvolution and fusion of dual-view fluorescence images.\nInspired by the recent Neuroclear method, Dual-Cycle is designed as a\ncycle-consistent generative network trained in a self-supervised fashion by\ncombining a dual-view generator and prior-guided degradation model. We validate\nDual-Cycle on both synthetic and real data showing its state-of-the-art\nperformance without any external training data.\n", "date": "2022-09-23", "categories": []}
{"id": "2209.10305", "title": "KXNet: A Model-Driven Deep Neural Network for Blind Super-Resolution", "abstract": "  Although current deep learning-based methods have gained promising\nperformance in the blind single image super-resolution (SISR) task, most of\nthem mainly focus on heuristically constructing diverse network architectures\nand put less emphasis on the explicit embedding of the physical generation\nmechanism between blur kernels and high-resolution (HR) images. To alleviate\nthis issue, we propose a model-driven deep neural network, called KXNet, for\nblind SISR. Specifically, to solve the classical SISR model, we propose a\nsimple-yet-effective iterative algorithm. Then by unfolding the involved\niterative steps into the corresponding network module, we naturally construct\nthe KXNet. The main specificity of the proposed KXNet is that the entire\nlearning process is fully and explicitly integrated with the inherent physical\nmechanism underlying this SISR task. Thus, the learned blur kernel has clear\nphysical patterns and the mutually iterative process between blur kernel and HR\nimage can soundly guide the KXNet to be evolved in the right direction.\nExtensive experiments on synthetic and real data finely demonstrate the\nsuperior accuracy and generality of our method beyond the current\nrepresentative state-of-the-art blind SISR methods. Code is available at:\nhttps://github.com/jiahong-fu/KXNet.\n", "date": "2022-09-21", "categories": []}
{"id": "2209.09408", "title": "Deep learning at the edge enables real-time streaming ptychographic\n  imaging", "abstract": "  Coherent microscopy techniques provide an unparalleled multi-scale view of\nmaterials across scientific and technological fields, from structural materials\nto quantum devices, from integrated circuits to biological cells. Driven by the\nconstruction of brighter sources and high-rate detectors, coherent X-ray\nmicroscopy methods like ptychography are poised to revolutionize nanoscale\nmaterials characterization. However, associated significant increases in data\nand compute needs mean that conventional approaches no longer suffice for\nrecovering sample images in real-time from high-speed coherent imaging\nexperiments. Here, we demonstrate a workflow that leverages artificial\nintelligence at the edge and high-performance computing to enable real-time\ninversion on X-ray ptychography data streamed directly from a detector at up to\n2 kHz. The proposed AI-enabled workflow eliminates the sampling constraints\nimposed by traditional ptychography, allowing low dose imaging using orders of\nmagnitude less data than required by traditional methods.\n", "date": "2022-09-20", "categories": []}
{"id": "2209.08277", "title": "MiNL: Micro-images based Neural Representation for Light Fields", "abstract": "  Traditional representations for light fields can be separated into two types:\nexplicit representation and implicit representation. Unlike explicit\nrepresentation that represents light fields as Sub-Aperture Images (SAIs) based\narrays or Micro-Images (MIs) based lenslet images, implicit representation\ntreats light fields as neural networks, which is inherently a continuous\nrepresentation in contrast to discrete explicit representation. However, at\npresent almost all the implicit representations for light fields utilize SAIs\nto train an MLP to learn a pixel-wise mapping from 4D spatial-angular\ncoordinate to pixel colors, which is neither compact nor of low complexity.\nInstead, in this paper we propose MiNL, a novel MI-wise implicit neural\nrepresentation for light fields that train an MLP + CNN to learn a mapping from\n2D MI coordinates to MI colors. Given the micro-image's coordinate, MiNL\noutputs the corresponding micro-image's RGB values. Light field encoding in\nMiNL is just training a neural network to regress the micro-images and the\ndecoding process is a simple feedforward operation. Compared with common\npixel-wise implicit representation, MiNL is more compact and efficient that has\nfaster decoding speed (\\textbf{$\\times$80$\\sim$180} speed-up) as well as better\nvisual quality (\\textbf{1$\\sim$4dB} PSNR improvement on average).\n", "date": "2022-09-17", "categories": []}
{"id": "2209.08452", "title": "MetaDIP: Accelerating Deep Image Prior with Meta Learning", "abstract": "  Deep image prior (DIP) is a recently proposed technique for solving imaging\ninverse problems by fitting the reconstructed images to the output of an\nuntrained convolutional neural network. Unlike pretrained feedforward neural\nnetworks, the same DIP can generalize to arbitrary inverse problems, from\ndenoising to phase retrieval, while offering competitive performance at each\ntask. The central disadvantage of DIP is that, while feedforward neural\nnetworks can reconstruct an image in a single pass, DIP must gradually update\nits weights over hundreds to thousands of iterations, at a significant\ncomputational cost. In this work we use meta-learning to massively accelerate\nDIP-based reconstructions. By learning a proper initialization for the DIP\nweights, we demonstrate a 10x improvement in runtimes across a range of inverse\nimaging tasks. Moreover, we demonstrate that a network trained to quickly\nreconstruct faces also generalizes to reconstructing natural image patches.\n", "date": "2022-09-18", "categories": []}
{"id": "2209.08471", "title": "MIPI 2022 Challenge on RGBW Sensor Re-mosaic: Dataset and Report", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for in-depth exchange of views from\nindustry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). To bridge the gap, we introduce the first MIPI\nchallenge including five tracks focusing on novel image sensors and imaging\nalgorithms. In this paper, RGBW Joint Remosaic and Denoise, one of the five\ntracks, working on the interpolation of RGBW CFA to Bayer at full resolution,\nis introduced. The participants were provided with a new dataset including 70\n(training) and 15 (validation) scenes of high-quality RGBW and Bayer pairs. In\naddition, for each scene, RGBW of different noise levels was provided at 0dB,\n24dB, and 42dB. All the data were captured using an RGBW sensor in both outdoor\nand indoor conditions. The final results are evaluated using objective metrics\nincluding PSNR, SSIM, LPIPS, and KLD. A detailed description of all models\ndeveloped in this challenge is provided in this paper. More details of this\nchallenge and the link to the dataset can be found at\nhttps://github.com/mipi-challenge/MIPI2022.\n", "date": "2022-09-15", "categories": []}
{"id": "2209.08734", "title": "Magnetic Resonance Fingerprinting with compressed sensing and distance\n  metric learning", "abstract": "  Magnetic Resonance Fingerprinting (MRF) is a novel technique that\nsimultaneously estimates multiple tissue-related parameters, such as the\nlongitudinal relaxation time T1, the transverse relaxation time T2, off\nresonance frequency B0 and proton density, from a scanned object in just tens\nof seconds. However, the MRF method suffers from aliasing artifacts because it\nsignificantly undersamples the k-space data. In this work, we propose a\ncompressed sensing (CS) framework for simultaneously estimating multiple\ntissue-related parameters based on the MRF method. It is more robust to low\nsampling ratio and is therefore more efficient in estimating MR parameters for\nall voxels of an object. Furthermore, the MRF method requires identifying the\nnearest atoms of the query fingerprints from the MR-signal-evolution dictionary\nwith the L2 distance. However, we observed that the L2 distance is not always a\nproper metric to measure the similarities between MR Fingerprints. Adaptively\nlearning a distance metric from the undersampled training data can\nsignificantly improve the matching accuracy of the query fingerprints.\nNumerical results on extensive simulated cases show that our method\nsubstantially outperforms stateof-the-art methods in terms of accuracy of\nparameter estimation.\n", "date": "2022-09-19", "categories": []}
{"id": "2207.14741", "title": "End-to-end View Synthesis via NeRF Attention", "abstract": "  In this paper, we present a simple seq2seq formulation for view synthesis\nwhere we take a set of ray points as input and output colors corresponding to\nthe rays. Directly applying a standard transformer on this seq2seq formulation\nhas two limitations. First, the standard attention cannot successfully fit the\nvolumetric rendering procedure, and therefore high-frequency components are\nmissing in the synthesized views. Second, applying global attention to all rays\nand pixels is extremely inefficient. Inspired by the neural radiance field\n(NeRF), we propose the NeRF attention (NeRFA) to address the above problems. On\nthe one hand, NeRFA considers the volumetric rendering equation as a soft\nfeature modulation procedure. In this way, the feature modulation enhances the\ntransformers with the NeRF-like inductive bias. On the other hand, NeRFA\nperforms multi-stage attention to reduce the computational overhead.\nFurthermore, the NeRFA model adopts the ray and pixel transformers to learn the\ninteractions between rays and pixels. NeRFA demonstrates superior performance\nover NeRF and NerFormer on four datasets: DeepVoxels, Blender, LLFF, and CO3D.\nBesides, NeRFA establishes a new state-of-the-art under two settings: the\nsingle-scene view synthesis and the category-centric novel view synthesis.\n", "date": "2022-07-29", "categories": []}
{"id": "2209.02854", "title": "Video Restoration with a Deep Plug-and-Play Prior", "abstract": "  This paper presents a novel method for restoring digital videos via a Deep\nPlug-and-Play (PnP) approach. Under a Bayesian formalism, the method consists\nin using a deep convolutional denoising network in place of the proximal\noperator of the prior in an alternating optimization scheme. We distinguish\nourselves from prior PnP work by directly applying that method to restore a\ndigital video from a degraded video observation. This way, a network trained\nonce for denoising can be repurposed for other video restoration tasks. Our\nexperiments in video deblurring, super-resolution, and interpolation of random\nmissing pixels all show a clear benefit to using a network specifically\ndesigned for video denoising, as it yields better restoration performance and\nbetter temporal stability than a single image network with similar denoising\nperformance using the same PnP formulation. Moreover, our method compares\nfavorably to applying a different state-of-the-art PnP scheme separately on\neach frame of the sequence. This opens new perspectives in the field of video\nrestoration.\n", "date": "2022-09-06", "categories": []}
{"id": "2209.07057", "title": "MIPI 2022 Challenge on RGB+ToF Depth Completion: Dataset and Report", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems is prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for in-depth exchange of views from\nindustry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). To bridge the gap, we introduce the first MIPI\nchallenge including five tracks focusing on novel image sensors and imaging\nalgorithms. In this paper, RGB+ToF Depth Completion, one of the five tracks,\nworking on the fusion of RGB sensor and ToF sensor (with spot illumination) is\nintroduced. The participants were provided with a new dataset called\nTetrasRGBD, which contains 18k pairs of high-quality synthetic RGB+Depth\ntraining data and 2.3k pairs of testing data from mixed sources. All the data\nare collected in an indoor scenario. We require that the running time of all\nmethods should be real-time on desktop GPUs. The final results are evaluated\nusing objective metrics and Mean Opinion Score (MOS) subjectively. A detailed\ndescription of all models developed in this challenge is provided in this\npaper. More details of this challenge and the link to the dataset can be found\nat https://github.com/mipi-challenge/MIPI2022.\n", "date": "2022-09-15", "categories": []}
{"id": "2209.07060", "title": "MIPI 2022 Challenge on Quad-Bayer Re-mosaic: Dataset and Report", "abstract": "  Developing and integrating advanced image sensors with novel algorithms in\ncamera systems are prevalent with the increasing demand for computational\nphotography and imaging on mobile platforms. However, the lack of high-quality\ndata for research and the rare opportunity for in-depth exchange of views from\nindustry and academia constrain the development of mobile intelligent\nphotography and imaging (MIPI). To bridge the gap, we introduce the first MIPI\nchallenge, including five tracks focusing on novel image sensors and imaging\nalgorithms. In this paper, Quad Joint Remosaic and Denoise, one of the five\ntracks, working on the interpolation of Quad CFA to Bayer at full resolution,\nis introduced. The participants were provided a new dataset, including 70\n(training) and 15 (validation) scenes of high-quality Quad and Bayer pairs. In\naddition, for each scene, Quad of different noise levels was provided at 0dB,\n24dB, and 42dB. All the data were captured using a Quad sensor in both outdoor\nand indoor conditions. The final results are evaluated using objective metrics,\nincluding PSNR, SSIM, LPIPS, and KLD. A detailed description of all models\ndeveloped in this challenge is provided in this paper. More details of this\nchallenge and the link to the dataset can be found at\nhttps://github.com/mipi-challenge/MIPI2022.\n", "date": "2022-09-15", "categories": []}
{"id": "2209.06027", "title": "Two-Step Color-Polarization Demosaicking Network", "abstract": "  Polarization information of light in a scene is valuable for various image\nprocessing and computer vision tasks. A division-of-focal-plane polarimeter is\na promising approach to capture the polarization images of different\norientations in one shot, while it requires color-polarization demosaicking. In\nthis paper, we propose a two-step color-polarization demosaicking\nnetwork~(TCPDNet), which consists of two sub-tasks of color demosaicking and\npolarization demosaicking. We also introduce a reconstruction loss in the YCbCr\ncolor space to improve the performance of TCPDNet. Experimental comparisons\ndemonstrate that TCPDNet outperforms existing methods in terms of the image\nquality of polarization images and the accuracy of Stokes parameters.\n", "date": "2022-09-13", "categories": []}
{"id": "2209.04941", "title": "Synthetic Wavelength Imaging -- Utilizing Spectral Correlations for\n  High-Precision Time-of-Flight Sensing", "abstract": "  This book chapter describes how spectral correlations in scattered light\nfields can be utilized for high-precision time-of-flight sensing. The chapter\nshould serve as a gentle introduction and is intended for computational imaging\nscientists and students new to the fascinating topic of synthetic wavelength\nimaging. Technical details (such as detector or light source specifications)\nwill be largely omitted. Instead, the similarities between different methods\nwill be emphasized to \"draw the bigger picture.\"\n", "date": "2022-09-11", "categories": []}
{"id": "2209.01578", "title": "Spatial-Temporal Transformer for Video Snapshot Compressive Imaging", "abstract": "  Video snapshot compressive imaging (SCI) captures multiple sequential video\nframes by a single measurement using the idea of computational imaging. The\nunderlying principle is to modulate high-speed frames through different masks\nand these modulated frames are summed to a single measurement captured by a\nlow-speed 2D sensor (dubbed optical encoder); following this, algorithms are\nemployed to reconstruct the desired high-speed frames (dubbed software decoder)\nif needed. In this paper, we consider the reconstruction algorithm in video\nSCI, i.e., recovering a series of video frames from a compressed measurement.\nSpecifically, we propose a Spatial-Temporal transFormer (STFormer) to exploit\nthe correlation in both spatial and temporal domains. STFormer network is\ncomposed of a token generation block, a video reconstruction block, and these\ntwo blocks are connected by a series of STFormer blocks. Each STFormer block\nconsists of a spatial self-attention branch, a temporal self-attention branch\nand the outputs of these two branches are integrated by a fusion network.\nExtensive results on both simulated and real data demonstrate the\nstate-of-the-art performance of STFormer. The code and models are publicly\navailable at https://github.com/ucaswangls/STFormer.git\n", "date": "2022-09-04", "categories": []}
{"id": "2206.05049", "title": "Denoising Generalized Expectation-Consistent Approximation for MR Image\n  Recovery", "abstract": "  To solve inverse problems, plug-and-play (PnP) methods replace the proximal\nstep in a convex optimization algorithm with a call to an application-specific\ndenoiser, often implemented using a deep neural network (DNN). Although such\nmethods yield accurate solutions, they can be improved. For example, denoisers\nare usually designed/trained to remove white Gaussian noise, but the denoiser\ninput error in PnP algorithms is usually far from white or Gaussian.\nApproximate message passing (AMP) methods provide white and Gaussian denoiser\ninput error, but only when the forward operator is sufficiently random. In this\nwork, for Fourier-based forward operators, we propose a PnP algorithm based on\ngeneralized expectation-consistent (GEC) approximation -- a close cousin of AMP\n-- that offers predictable error statistics at each iteration, as well as a new\nDNN denoiser that leverages those statistics. We apply our approach to magnetic\nresonance (MR) image recovery and demonstrate its advantages over existing PnP\nand AMP methods.\n", "date": "2022-06-09", "categories": []}
{"id": "2209.01356", "title": "Masked Sinogram Model with Transformer for ill-Posed Computed Tomography\n  Reconstruction: a Preliminary Study", "abstract": "  Computed Tomography (CT) is an imaging technique where information about an\nobject are collected at different angles (called projections or scans). Then\nthe cross-sectional image showing the internal structure of the slice is\nproduced by solving an inverse problem. Limited by certain factors such as\nradiation dosage, projection angles, the produced images can be noisy or\ncontain artifacts. Inspired by the success of transformer for natural language\nprocessing, the core idea of this preliminary study is to consider a projection\nof tomography as a word token, and the whole scan of the cross-section (A.K.A.\nsinogram) as a sentence in the context of natural language processing. Then we\nexplore the idea of foundation model by training a masked sinogram model (MSM)\nand fine-tune MSM for various downstream applications including CT\nreconstruction under data collections restriction (e.g., photon-budget) and a\ndata-driven solution to approximate solutions of the inverse problem for CT\nreconstruction. Models and data used in this study are available at\nhttps://github.com/lzhengchun/TomoTx.\n", "date": "2022-09-03", "categories": []}
{"id": "2209.01725", "title": "Imaging with Equivariant Deep Learning", "abstract": "  From early image processing to modern computational imaging, successful\nmodels and algorithms have relied on a fundamental property of natural signals:\nsymmetry. Here symmetry refers to the invariance property of signal sets to\ntransformations such as translation, rotation or scaling. Symmetry can also be\nincorporated into deep neural networks in the form of equivariance, allowing\nfor more data-efficient learning. While there has been important advances in\nthe design of end-to-end equivariant networks for image classification in\nrecent years, computational imaging introduces unique challenges for\nequivariant network solutions since we typically only observe the image through\nsome noisy ill-conditioned forward operator that itself may not be equivariant.\nWe review the emerging field of equivariant imaging and show how it can provide\nimproved generalization and new imaging opportunities. Along the way we show\nthe interplay between the acquisition physics and group actions and links to\niterative reconstruction, blind compressed sensing and self-supervised\nlearning.\n", "date": "2022-09-05", "categories": []}
{"id": "2209.02417", "title": "Volume Rendering Digest (for NeRF)", "abstract": "  Neural Radiance Fields employ simple volume rendering as a way to overcome\nthe challenges of differentiating through ray-triangle intersections by\nleveraging a probabilistic notion of visibility. This is achieved by assuming\nthe scene is composed by a cloud of light-emitting particles whose density\nchanges in space. This technical report summarizes the derivations for\ndifferentiable volume rendering. It is a condensed version of previous reports,\nbut rewritten in the context of NeRF, and adopting its commonly used notation.\n", "date": "2022-08-29", "categories": []}
{"id": "2208.14022", "title": "Stabilize, Decompose, and Denoise: Self-Supervised Fluoroscopy Denoising", "abstract": "  Fluoroscopy is an imaging technique that uses X-ray to obtain a real-time 2D\nvideo of the interior of a 3D object, helping surgeons to observe pathological\nstructures and tissue functions especially during intervention. However, it\nsuffers from heavy noise that mainly arises from the clinical use of a low dose\nX-ray, thereby necessitating the technology of fluoroscopy denoising. Such\ndenoising is challenged by the relative motion between the object being imaged\nand the X-ray imaging system. We tackle this challenge by proposing a\nself-supervised, three-stage framework that exploits the domain knowledge of\nfluoroscopy imaging. (i) Stabilize: we first construct a dynamic panorama based\non optical flow calculation to stabilize the non-stationary background induced\nby the motion of the X-ray detector. (ii) Decompose: we then propose a novel\nmask-based Robust Principle Component Analysis (RPCA) decomposition method to\nseparate a video with detector motion into a low-rank background and a sparse\nforeground. Such a decomposition accommodates the reading habit of experts.\n(iii) Denoise: we finally denoise the background and foreground separately by a\nself-supervised learning strategy and fuse the denoised parts into the final\noutput via a bilateral, spatiotemporal filter. To assess the effectiveness of\nour work, we curate a dedicated fluoroscopy dataset of 27 videos (1,568 frames)\nand corresponding ground truth. Our experiments demonstrate that it achieves\nsignificant improvements in terms of denoising and enhancement effects when\ncompared with standard approaches. Finally, expert rating confirms this\nefficacy.\n", "date": "2022-08-30", "categories": []}
{"id": "2208.14369", "title": "SIGNet: Intrinsic Image Decomposition by a Semantic and Invariant\n  Gradient Driven Network for Indoor Scenes", "abstract": "  Intrinsic image decomposition (IID) is an under-constrained problem.\nTherefore, traditional approaches use hand crafted priors to constrain the\nproblem. However, these constraints are limited when coping with complex\nscenes. Deep learning-based approaches learn these constraints implicitly\nthrough the data, but they often suffer from dataset biases (due to not being\nable to include all possible imaging conditions).\n  In this paper, a combination of the two is proposed. Component specific\npriors like semantics and invariant features are exploited to obtain\nsemantically and physically plausible reflectance transitions. These\ntransitions are used to steer a progressive CNN with implicit homogeneity\nconstraints to decompose reflectance and shading maps.\n  An ablation study is conducted showing that the use of the proposed priors\nand progressive CNN increase the IID performance. State of the art performance\non both our proposed dataset and the standard real-world IIW dataset shows the\neffectiveness of the proposed method. Code is made available at\nhttps://github.com/Morpheus3000/SIGNet\n", "date": "2022-08-30", "categories": []}
{"id": "2208.13184", "title": "Towards Real-World Video Deblurring by Exploring Blur Formation Process", "abstract": "  This paper aims at exploring how to synthesize close-to-real blurs that\nexisting video deblurring models trained on them can generalize well to\nreal-world blurry videos. In recent years, deep learning-based approaches have\nachieved promising success on video deblurring task. However, the models\ntrained on existing synthetic datasets still suffer from generalization\nproblems over real-world blurry scenarios with undesired artifacts. The factors\naccounting for the failure remain unknown. Therefore, we revisit the classical\nblur synthesis pipeline and figure out the possible reasons, including shooting\nparameters, blur formation space, and image signal processor~(ISP). To analyze\nthe effects of these potential factors, we first collect an ultra-high\nframe-rate (940 FPS) RAW video dataset as the data basis to synthesize various\nkinds of blurs. Then we propose a novel realistic blur synthesis pipeline\ntermed as RAW-Blur by leveraging blur formation cues. Through numerous\nexperiments, we demonstrate that synthesizing blurs in the RAW space and\nadopting the same ISP as the real-world testing data can effectively eliminate\nthe negative effects of synthetic data. Furthermore, the shooting parameters of\nthe synthesized blurry video, e.g., exposure time and frame-rate play\nsignificant roles in improving the performance of deblurring models.\nImpressively, the models trained on the blurry data synthesized by the proposed\nRAW-Blur pipeline can obtain more than 5dB PSNR gain against those trained on\nthe existing synthetic blur datasets. We believe the novel realistic synthesis\npipeline and the corresponding RAW video dataset can help the community to\neasily construct customized blur datasets to improve real-world video\ndeblurring performance largely, instead of laboriously collecting real data\npairs.\n", "date": "2022-08-28", "categories": []}
{"id": "2208.13485", "title": "A Practical Calibration Method for RGB Micro-Grid Polarimetric Cameras", "abstract": "  Polarimetric imaging has been applied in a growing number of applications in\nrobotic vision (ex. underwater navigation, glare removal, de-hazing, object\nclassification, and depth estimation). One can find on the market RGB\nPolarization cameras that can capture both color and polarimetric state of the\nlight in a single snapshot. Due to the sensor's characteristic dispersion, and\nthe use of lenses, it is crucial to calibrate these types of cameras so as to\nobtain correct polarization measurements. The calibration methods that have\nbeen developed so far are either not adapted to this type of cameras, or they\nrequire complex equipment and time consuming experiments in strict setups. In\nthis paper, we propose a new method to overcome the need for complex optical\nsystems to efficiently calibrate these cameras. We show that the proposed\ncalibration method has several advantages such as that any user can easily\ncalibrate the camera using a uniform, linearly polarized light source without\nany a priori knowledge of its polarization state, and with a limited number of\nacquisitions. We will make our calibration code publicly available.\n", "date": "2022-08-29", "categories": []}
{"id": "2208.11836", "title": "Polarimetric Inverse Rendering for Transparent Shapes Reconstruction", "abstract": "  In this work, we propose a novel method for the detailed reconstruction of\ntransparent objects by exploiting polarimetric cues. Most of the existing\nmethods usually lack sufficient constraints and suffer from the over-smooth\nproblem. Hence, we introduce polarization information as a complementary cue.\nWe implicitly represent the object's geometry as a neural network, while the\npolarization render is capable of rendering the object's polarization images\nfrom the given shape and illumination configuration. Direct comparison of the\nrendered polarization images to the real-world captured images will have\nadditional errors due to the transmission in the transparent object. To address\nthis issue, the concept of reflection percentage which represents the\nproportion of the reflection component is introduced. The reflection percentage\nis calculated by a ray tracer and then used for weighting the polarization\nloss. We build a polarization dataset for multi-view transparent shapes\nreconstruction to verify our method. The experimental results show that our\nmethod is capable of recovering detailed shapes and improving the\nreconstruction quality of transparent objects. Our dataset and code will be\npublicly available at https://github.com/shaomq2187/TransPIR.\n", "date": "2022-08-25", "categories": []}
{"id": "2208.11622", "title": "Generative Adversarial Network (GAN) based Image-Deblurring", "abstract": "  This thesis analyzes the challenging problem of Image Deblurring based on\nclassical theorems and state-of-art methods proposed in recent years. By\nspectral analysis we mathematically show the effective of spectral\nregularization methods, and point out the linking between the spectral\nfiltering result and the solution of the regularization optimization objective.\nFor ill-posed problems like image deblurring, the optimization objective\ncontains a regularization term (also called the regularization functional) that\nencodes our prior knowledge into the solution. We demonstrate how to craft a\nregularization term by hand using the idea of maximum a posterior estimation.\nThen, we point out the limitations of such regularization-based methods, and\nstep into the neural-network based methods.\n  Based on the idea of Wasserstein generative adversarial models, we can train\na CNN to learn the regularization functional. Such data-driven approaches are\nable to capture the complexity, which may not be analytically modellable.\nBesides, in recent years with the improvement of architectures, the network has\nbeen able to output an image closely approximating the ground truth given the\nblurry observation. The Generative Adversarial Network (GAN) works on this\nImage-to-Image translation idea. We analyze the DeblurGAN-v2 method proposed by\nOrest Kupyn et al. [14] in 2019 based on numerical tests. And, based on the\nexperimental results and our knowledge, we put forward some suggestions for\nimprovement on this method.\n", "date": "2022-08-24", "categories": []}
{"id": "2208.08897", "title": "NeIF: Representing General Reflectance as Neural Intrinsics Fields for\n  Uncalibrated Photometric Stereo", "abstract": "  Uncalibrated photometric stereo (UPS) is challenging due to the inherent\nambiguity brought by unknown light. Existing solutions alleviate the ambiguity\nby either explicitly associating reflectance to light conditions or resolving\nlight conditions in a supervised manner. This paper establishes an implicit\nrelation between light clues and light estimation and solves UPS in an\nunsupervised manner. The key idea is to represent the reflectance as four\nneural intrinsics fields, i.e., position, light, specular, and shadow, based on\nwhich the neural light field is implicitly associated with light clues of\nspecular reflectance and cast shadow. The unsupervised, joint optimization of\nneural intrinsics fields can be free from training data bias as well as\naccumulating error, and fully exploits all observed pixel values for UPS. Our\nmethod achieves a superior performance advantage over state-of-the-art UPS\nmethods on public and self-collected datasets, under regular and challenging\nsetups. The code will be released soon.\n", "date": "2022-08-18", "categories": []}
{"id": "2208.09451", "title": "Guided-deconvolution for Correlative Light and Electron Microscopy", "abstract": "  Correlative light and electron microscopy is a powerful tool to study the\ninternal structure of cells. It combines the mutual benefit of correlating\nlight (LM) and electron (EM) microscopy information. However, the classical\napproach of overlaying LM onto EM images to assign functional to structural\ninformation is hampered by the large discrepancy in structural detail visible\nin the LM images. This paper aims at investigating an optimized approach which\nwe call EM-guided deconvolution. It attempts to automatically assign\nfluorescence-labelled structures to details visible in the EM image to bridge\nthe gaps in both resolution and specificity between the two imaging modes.\n", "date": "2022-08-19", "categories": []}
{"id": "2208.08604", "title": "Towards Practical Single-shot Phase Retrieval with Physics-Driven Deep\n  Neural Network", "abstract": "  Phase retrieval (PR), a long-established challenge for recovering a\ncomplex-valued signal from its Fourier intensity-only measurements, has\nattracted considerable attention due to its widespread applications in digital\nimaging. Recently, deep learning-based approaches were developed that achieved\nsome success in single-shot PR. These approaches require a single Fourier\nintensity measurement without the need to impose any additional constraints on\nthe measured data. Nevertheless, vanilla deep neural networks (DNN) do not give\ngood performance due to the substantial disparity between the input and output\ndomains of the PR problems. Physics-informed approaches try to incorporate the\nFourier intensity measurements into an iterative approach to increase the\nreconstruction accuracy. It, however, requires a lengthy computation process,\nand the accuracy still cannot be guaranteed. Besides, many of these approaches\nwork on simulation data that ignore some common problems such as saturation and\nquantization errors in practical optical PR systems. In this paper, a novel\nphysics-driven multi-scale DNN structure dubbed PPRNet is proposed. Similar to\nother deep learning-based PR methods, PPRNet requires only a single Fourier\nintensity measurement. It is physics-driven that the network is guided to\nfollow the Fourier intensity measurement at different scales to enhance the\nreconstruction accuracy. PPRNet has a feedforward structure and can be\nend-to-end trained. Thus, it is much faster and more accurate than the\ntraditional physics-driven PR approaches. Extensive simulations and experiments\non a practical optical platform were conducted. The results demonstrate the\nsuperiority and practicality of the proposed PPRNet over the traditional\nlearning-based PR methods.\n", "date": "2022-08-18", "categories": []}
{"id": "2207.00164", "title": "Deep Optical Coding Design in Computational Imaging", "abstract": "  Computational optical imaging (COI) systems leverage optical coding elements\n(CE) in their setups to encode a high-dimensional scene in a single or multiple\nsnapshots and decode it by using computational algorithms. The performance of\nCOI systems highly depends on the design of its main components: the CE pattern\nand the computational method used to perform a given task. Conventional\napproaches rely on random patterns or analytical designs to set the\ndistribution of the CE. However, the available data and algorithm capabilities\nof deep neural networks (DNNs) have opened a new horizon in CE data-driven\ndesigns that jointly consider the optical encoder and computational decoder.\nSpecifically, by modeling the COI measurements through a fully differentiable\nimage formation model that considers the physics-based propagation of light and\nits interaction with the CEs, the parameters that define the CE and the\ncomputational decoder can be optimized in an end-to-end (E2E) manner. Moreover,\nby optimizing just CEs in the same framework, inference tasks can be performed\nfrom pure optics. This work surveys the recent advances on CE data-driven\ndesign and provides guidelines on how to parametrize different optical elements\nto include them in the E2E framework. Since the E2E framework can handle\ndifferent inference applications by changing the loss function and the DNN, we\npresent low-level tasks such as spectral imaging reconstruction or high-level\ntasks such as pose estimation with privacy preserving enhanced by using optimal\ntask-based optical architectures. Finally, we illustrate classification and 3D\nobject recognition applications performed at the speed of the light using\nall-optics DNN.\n", "date": "2022-06-27", "categories": []}
{"id": "2208.04283", "title": "LWGNet: Learned Wirtinger Gradients for Fourier Ptychographic Phase\n  Retrieval", "abstract": "  Fourier Ptychographic Microscopy (FPM) is an imaging procedure that overcomes\nthe traditional limit on Space-Bandwidth Product (SBP) of conventional\nmicroscopes through computational means. It utilizes multiple images captured\nusing a low numerical aperture (NA) objective and enables high-resolution phase\nimaging through frequency domain stitching. Existing FPM reconstruction methods\ncan be broadly categorized into two approaches: iterative optimization based\nmethods, which are based on the physics of the forward imaging model, and\ndata-driven methods which commonly employ a feed-forward deep learning\nframework. We propose a hybrid model-driven residual network that combines the\nknowledge of the forward imaging system with a deep data-driven network. Our\nproposed architecture, LWGNet, unrolls traditional Wirtinger flow optimization\nalgorithm into a novel neural network design that enhances the gradient images\nthrough complex convolutional blocks. Unlike other conventional unrolling\ntechniques, LWGNet uses fewer stages while performing at par or even better\nthan existing traditional and deep learning techniques, particularly, for\nlow-cost and low dynamic range CMOS sensors. This improvement in performance\nfor low-bit depth and low-cost sensors has the potential to bring down the cost\nof FPM imaging setup significantly. Finally, we show consistently improved\nperformance on our collected real data.\n", "date": "2022-08-08", "categories": []}
{"id": "2208.05827", "title": "K-UNN: k-Space Interpolation With Untrained Neural Network", "abstract": "  Recently, untrained neural networks (UNNs) have shown satisfactory\nperformances for MR image reconstruction on random sampling trajectories\nwithout using additional full-sampled training data. However, the existing\nUNN-based approach does not fully use the MR image physical priors, resulting\nin poor performance in some common scenarios (e.g., partial Fourier, regular\nsampling, etc.) and the lack of theoretical guarantees for reconstruction\naccuracy. To bridge this gap, we propose a safeguarded k-space interpolation\nmethod for MRI using a specially designed UNN with a tripled architecture\ndriven by three physical priors of the MR images (or k-space data), including\nsparsity, coil sensitivity smoothness, and phase smoothness. We also prove that\nthe proposed method guarantees tight bounds for interpolated k-space data\naccuracy. Finally, ablation experiments show that the proposed method can more\naccurately characterize the physical priors of MR images than existing\ntraditional methods. Additionally, under a series of commonly used sampling\ntrajectories, experiments also show that the proposed method consistently\noutperforms traditional parallel imaging methods and existing UNNs, and even\noutperforms the state-of-the-art supervised-trained k-space deep learning\nmethods in some cases.\n", "date": "2022-08-11", "categories": []}
{"id": "2208.04679", "title": "Improved Multiple-Image-Based Reflection Removal Algorithm Using Deep\n  Neural Networks", "abstract": "  When imaging through a semi-reflective medium such as glass, the reflection\nof another scene can often be found in the captured images. It degrades the\nquality of the images and affects their subsequent analyses. In this paper, a\nnovel deep neural network approach for solving the reflection problem in\nimaging is presented. Traditional reflection removal methods not only require\nlong computation time for solving different optimization functions, their\nperformance is also not guaranteed. As array cameras are readily available in\nnowadays imaging devices, we first suggest in this paper a multiple-image based\ndepth estimation method using a convolutional neural network (CNN). The\nproposed network avoids the depth ambiguity problem due to the reflection in\nthe image, and directly estimates the depths along the image edges. They are\nthen used to classify the edges as belonging to the background or reflection.\nSince edges having similar depth values are error prone in the classification,\nthey are removed from the reflection removal process. We suggest a generative\nadversarial network (GAN) to regenerate the removed background edges. Finally,\nthe estimated background edge map is fed to another auto-encoder network to\nassist the extraction of the background from the original image. Experimental\nresults show that the proposed reflection removal algorithm achieves superior\nperformance both quantitatively and qualitatively as compared to the\nstate-of-the-art methods. The proposed algorithm also shows much faster speed\ncompared to the existing approaches using the traditional optimization methods.\n", "date": "2022-08-09", "categories": []}
{"id": "2208.05114", "title": "Ghost-free High Dynamic Range Imaging with Context-aware Transformer", "abstract": "  High dynamic range (HDR) deghosting algorithms aim to generate ghost-free HDR\nimages with realistic details. Restricted by the locality of the receptive\nfield, existing CNN-based methods are typically prone to producing ghosting\nartifacts and intensity distortions in the presence of large motion and severe\nsaturation. In this paper, we propose a novel Context-Aware Vision Transformer\n(CA-ViT) for ghost-free high dynamic range imaging. The CA-ViT is designed as a\ndual-branch architecture, which can jointly capture both global and local\ndependencies. Specifically, the global branch employs a window-based\nTransformer encoder to model long-range object movements and intensity\nvariations to solve ghosting. For the local branch, we design a local context\nextractor (LCE) to capture short-range image features and use the channel\nattention mechanism to select informative local details across the extracted\nfeatures to complement the global branch. By incorporating the CA-ViT as basic\ncomponents, we further build the HDR-Transformer, a hierarchical network to\nreconstruct high-quality ghost-free HDR images. Extensive experiments on three\nbenchmark datasets show that our approach outperforms state-of-the-art methods\nqualitatively and quantitatively with considerably reduced computational\nbudgets. Codes are available at\nhttps://github.com/megvii-research/HDR-Transformer\n", "date": "2022-08-10", "categories": []}
{"id": "2208.03161", "title": "Adversarial Robustness of MR Image Reconstruction under Realistic\n  Perturbations", "abstract": "  Deep Learning (DL) methods have shown promising results for solving ill-posed\ninverse problems such as MR image reconstruction from undersampled $k$-space\ndata. However, these approaches currently have no guarantees for reconstruction\nquality and the reliability of such algorithms is only poorly understood.\nAdversarial attacks offer a valuable tool to understand possible failure modes\nand worst case performance of DL-based reconstruction algorithms. In this paper\nwe describe adversarial attacks on multi-coil $k$-space measurements and\nevaluate them on the recently proposed E2E-VarNet and a simpler UNet-based\nmodel. In contrast to prior work, the attacks are targeted to specifically\nalter diagnostically relevant regions. Using two realistic attack models\n(adversarial $k$-space noise and adversarial rotations) we are able to show\nthat current state-of-the-art DL-based reconstruction algorithms are indeed\nsensitive to such perturbations to a degree where relevant diagnostic\ninformation may be lost. Surprisingly, in our experiments the UNet and the more\nsophisticated E2E-VarNet were similarly sensitive to such attacks. Our findings\nadd further to the evidence that caution must be exercised as DL-based methods\nmove closer to clinical practice.\n", "date": "2022-08-05", "categories": []}
{"id": "2207.00945", "title": "PS$^2$F: Polarized Spiral Point Spread Function for Single-Shot 3D\n  Sensing", "abstract": "  We propose a compact snapshot monocular depth estimation technique that\nrelies on an engineered point spread function (PSF). Traditional approaches\nused in microscopic super-resolution imaging such as the Double-Helix PSF\n(DHPSF) are ill-suited for scenes that are more complex than a sparse set of\npoint light sources. We show, using the Cram\\'er-Rao lower bound, that\nseparating the two lobes of the DHPSF and thereby capturing two separate images\nleads to a dramatic increase in depth accuracy. A special property of the phase\nmask used for generating the DHPSF is that a separation of the phase mask into\ntwo halves leads to a spatial separation of the two lobes. We leverage this\nproperty to build a compact polarization-based optical setup, where we place\ntwo orthogonal linear polarizers on each half of the DHPSF phase mask and then\ncapture the resulting image with a polarization-sensitive camera. Results from\nsimulations and a lab prototype demonstrate that our technique achieves up to\n$50\\%$ lower depth error compared to state-of-the-art designs including the\nDHPSF and the Tetrapod PSF, with little to no loss in spatial resolution.\n", "date": "2022-07-03", "categories": []}
{"id": "2208.02361", "title": "A Multibranch Convolutional Neural Network for Hyperspectral Unmixing", "abstract": "  Hyperspectral unmixing remains one of the most challenging tasks in the\nanalysis of such data. Deep learning has been blooming in the field and proved\nto outperform other classic unmixing techniques, and can be effectively\ndeployed onboard Earth observation satellites equipped with hyperspectral\nimagers. In this letter, we follow this research pathway and propose a\nmulti-branch convolutional neural network that benefits from fusing spectral,\nspatial, and spectral-spatial features in the unmixing process. The results of\nour experiments, backed up with the ablation study, revealed that our\ntechniques outperform others from the literature and lead to higher-quality\nfractional abundance estimation. Also, we investigated the influence of\nreducing the training sets on the capabilities of all algorithms and their\nrobustness against noise, as capturing large and representative ground-truth\nsets is time-consuming and costly in practice, especially in emerging Earth\nobservation scenarios.\n", "date": "2022-08-03", "categories": []}
{"id": "2208.01827", "title": "Fast Hierarchical Deep Unfolding Network for Image Compressed Sensing", "abstract": "  By integrating certain optimization solvers with deep neural network, deep\nunfolding network (DUN) has attracted much attention in recent years for image\ncompressed sensing (CS). However, there still exist several issues in existing\nDUNs: 1) For each iteration, a simple stacked convolutional network is usually\nadopted, which apparently limits the expressiveness of these models. 2) Once\nthe training is completed, most hyperparameters of existing DUNs are fixed for\nany input content, which significantly weakens their adaptability. In this\npaper, by unfolding the Fast Iterative Shrinkage-Thresholding Algorithm\n(FISTA), a novel fast hierarchical DUN, dubbed FHDUN, is proposed for image\ncompressed sensing, in which a well-designed hierarchical unfolding\narchitecture is developed to cooperatively explore richer contextual prior\ninformation in multi-scale spaces. To further enhance the adaptability, series\nof hyperparametric generation networks are developed in our framework to\ndynamically produce the corresponding optimal hyperparameters according to the\ninput content. Furthermore, due to the accelerated policy in FISTA, the newly\nembedded acceleration module makes the proposed FHDUN save more than 50% of the\niterative loops against recent DUNs. Extensive CS experiments manifest that the\nproposed FHDUN outperforms existing state-of-the-art CS methods, while\nmaintaining fewer iterations.\n", "date": "2022-08-03", "categories": []}
{"id": "2207.13460", "title": "Adaptive sampling for scanning pixel cameras", "abstract": "  A scanning pixel camera is a novel low-cost, low-power sensor that is not\ndiffraction limited. It produces data as a sequence of samples extracted from\nvarious parts of the scene during the course of a scan. It can provide very\ndetailed images at the expense of samplerates and slow image acquisition time.\nThis paper proposes a new algorithm which allows the sensor to adapt the\nsamplerate over the course of this sequence. This makes it possible to overcome\nsome of these limitations by minimising the bandwidth and time required to\nimage and transmit a scene, while maintaining image quality. We examine\napplications to image classification and semantic segmentation and are able to\nachieve similar results compared to a fully sampled input, while using 80%\nfewer samples\n", "date": "2022-07-27", "categories": []}
{"id": "2208.00945", "title": "DoF-NeRF: Depth-of-Field Meets Neural Radiance Fields", "abstract": "  Neural Radiance Field (NeRF) and its variants have exhibited great success on\nrepresenting 3D scenes and synthesizing photo-realistic novel views. However,\nthey are generally based on the pinhole camera model and assume all-in-focus\ninputs. This limits their applicability as images captured from the real world\noften have finite depth-of-field (DoF). To mitigate this issue, we introduce\nDoF-NeRF, a novel neural rendering approach that can deal with shallow DoF\ninputs and can simulate DoF effect. In particular, it extends NeRF to simulate\nthe aperture of lens following the principles of geometric optics. Such a\nphysical guarantee allows DoF-NeRF to operate views with different focus\nconfigurations. Benefiting from explicit aperture modeling, DoF-NeRF also\nenables direct manipulation of DoF effect by adjusting virtual aperture and\nfocus parameters. It is plug-and-play and can be inserted into NeRF-based\nframeworks. Experiments on synthetic and real-world datasets show that,\nDoF-NeRF not only performs comparably with NeRF in the all-in-focus setting,\nbut also can synthesize all-in-focus novel views conditioned on shallow DoF\ninputs. An interesting application of DoF-NeRF to DoF rendering is also\ndemonstrated. The source code will be made available at\nhttps://github.com/zijinwuzijin/DoF-NeRF.\n", "date": "2022-08-01", "categories": []}
{"id": "2208.00950", "title": "Fast Two-step Blind Optical Aberration Correction", "abstract": "  The optics of any camera degrades the sharpness of photographs, which is a\nkey visual quality criterion. This degradation is characterized by the\npoint-spread function (PSF), which depends on the wavelengths of light and is\nvariable across the imaging field. In this paper, we propose a two-step scheme\nto correct optical aberrations in a single raw or JPEG image, i.e., without any\nprior information on the camera or lens. First, we estimate local Gaussian blur\nkernels for overlapping patches and sharpen them with a non-blind deblurring\ntechnique. Based on the measurements of the PSFs of dozens of lenses, these\nblur kernels are modeled as RGB Gaussians defined by seven parameters. Second,\nwe remove the remaining lateral chromatic aberrations (not contemplated in the\nfirst step) with a convolutional neural network, trained to minimize the\nred/green and blue/green residual images. Experiments on both synthetic and\nreal images show that the combination of these two stages yields a fast\nstate-of-the-art blind optical aberration compensation technique that competes\nwith commercial non-blind algorithms.\n", "date": "2022-08-01", "categories": []}
{"id": "2207.14392", "title": "A Deep Generative Approach to Oversampling in Ptychography", "abstract": "  Ptychography is a well-studied phase imaging method that makes non-invasive\nimaging possible at a nanometer scale. It has developed into a mainstream\ntechnique with various applications across a range of areas such as material\nscience or the defense industry. One major drawback of ptychography is the long\ndata acquisition time due to the high overlap requirement between adjacent\nillumination areas to achieve a reasonable reconstruction. Traditional\napproaches with reduced overlap between scanning areas result in\nreconstructions with artifacts. In this paper, we propose complementing\nsparsely acquired or undersampled data with data sampled from a deep generative\nnetwork to satisfy the oversampling requirement in ptychography. Because the\ndeep generative network is pre-trained and its output can be computed as we\ncollect data, the experimental data and the time to acquire the data can be\nreduced. We validate the method by presenting the reconstruction quality\ncompared to the previously proposed and traditional approaches and comment on\nthe strengths and drawbacks of the proposed approach.\n", "date": "2022-07-28", "categories": []}
{"id": "2207.14671", "title": "High Dynamic Range and Super-Resolution from Raw Image Bursts", "abstract": "  Photographs captured by smartphones and mid-range cameras have limited\nspatial resolution and dynamic range, with noisy response in underexposed\nregions and color artefacts in saturated areas. This paper introduces the first\napproach (to the best of our knowledge) to the reconstruction of\nhigh-resolution, high-dynamic range color images from raw photographic bursts\ncaptured by a handheld camera with exposure bracketing. This method uses a\nphysically-accurate model of image formation to combine an iterative\noptimization algorithm for solving the corresponding inverse problem with a\nlearned image representation for robust alignment and a learned natural image\nprior. The proposed algorithm is fast, with low memory requirements compared to\nstate-of-the-art learning-based approaches to image restoration, and features\nthat are learned end to end from synthetic yet realistic data. Extensive\nexperiments demonstrate its excellent performance with super-resolution factors\nof up to $\\times 4$ on real photographs taken in the wild with hand-held\ncameras, and high robustness to low-light conditions, noise, camera shake, and\nmoderate object motion.\n", "date": "2022-07-29", "categories": []}
{"id": "2207.14811", "title": "StyleLight: HDR Panorama Generation for Lighting Estimation and Editing", "abstract": "  We present a new lighting estimation and editing framework to generate\nhigh-dynamic-range (HDR) indoor panorama lighting from a single limited\nfield-of-view (LFOV) image captured by low-dynamic-range (LDR) cameras.\nExisting lighting estimation methods either directly regress lighting\nrepresentation parameters or decompose this problem into LFOV-to-panorama and\nLDR-to-HDR lighting generation sub-tasks. However, due to the partial\nobservation, the high-dynamic-range lighting, and the intrinsic ambiguity of a\nscene, lighting estimation remains a challenging task. To tackle this problem,\nwe propose a coupled dual-StyleGAN panorama synthesis network (StyleLight) that\nintegrates LDR and HDR panorama synthesis into a unified framework. The LDR and\nHDR panorama synthesis share a similar generator but have separate\ndiscriminators. During inference, given an LDR LFOV image, we propose a\nfocal-masked GAN inversion method to find its latent code by the LDR panorama\nsynthesis branch and then synthesize the HDR panorama by the HDR panorama\nsynthesis branch. StyleLight takes LFOV-to-panorama and LDR-to-HDR lighting\ngeneration into a unified framework and thus greatly improves lighting\nestimation. Extensive experiments demonstrate that our framework achieves\nsuperior performance over state-of-the-art methods on indoor lighting\nestimation. Notably, StyleLight also enables intuitive lighting editing on\nindoor HDR panoramas, which is suitable for real-world applications. Code is\navailable at https://style-light.github.io.\n", "date": "2022-07-29", "categories": []}
{"id": "2207.14230", "title": "Physics-informed neural networks for diffraction tomography", "abstract": "  We propose a physics-informed neural network as the forward model for\ntomographic reconstructions of biological samples. We demonstrate that by\ntraining this network with the Helmholtz equation as a physical loss, we can\npredict the scattered field accurately. It will be shown that a pretrained\nnetwork can be fine-tuned for different samples and used for solving the\nscattering problem much faster than other numerical solutions. We evaluate our\nmethodology with numerical and experimental results. Our physics-informed\nneural networks can be generalized for any forward and inverse scattering\nproblem.\n", "date": "2022-07-28", "categories": []}
{"id": "2207.13200", "title": "Deep Model-Based Architectures for Inverse Problems under Mismatched\n  Priors", "abstract": "  There is a growing interest in deep model-based architectures (DMBAs) for\nsolving imaging inverse problems by combining physical measurement models and\nlearned image priors specified using convolutional neural nets (CNNs). For\nexample, well-known frameworks for systematically designing DMBAs include\nplug-and-play priors (PnP), deep unfolding (DU), and deep equilibrium models\n(DEQ). While the empirical performance and theoretical properties of DMBAs have\nbeen widely investigated, the existing work in the area has primarily focused\non their performance when the desired image prior is known exactly. This work\naddresses the gap in the prior work by providing new theoretical and numerical\ninsights into DMBAs under mismatched CNN priors. Mismatched priors arise\nnaturally when there is a distribution shift between training and testing data,\nfor example, due to test images being from a different distribution than images\nused for training the CNN prior. They also arise when the CNN prior used for\ninference is an approximation of some desired statistical estimator (MAP or\nMMSE). Our theoretical analysis provides explicit error bounds on the solution\ndue to the mismatched CNN priors under a set of clearly specified assumptions.\nOur numerical results compare the empirical performance of DMBAs under\nrealistic distribution shifts and approximate statistical estimators.\n", "date": "2022-07-26", "categories": []}
{"id": "2207.13344", "title": "Inverse Airborne Optical Sectioning", "abstract": "  We present Inverse Airborne Optical Sectioning (IAOS) an optical analogy to\nInverse Synthetic Aperture Radar (ISAR). Moving targets, such as walking\npeople, that are heavily occluded by vegetation can be made visible and tracked\nwith a stationary optical sensor (e.g., a hovering camera drone above forest).\nWe introduce the principles of IAOS (i.e., inverse synthetic aperture imaging),\nexplain how the signal of occluders can be further suppressed by filtering the\nRadon transform of the image integral, and present how targets motion\nparameters can be estimated manually and automatically. Finally, we show that\nwhile tracking occluded targets in conventional aerial images is infeasible, it\nbecomes efficiently possible in integral images that result from IAOS.\n", "date": "2022-07-27", "categories": []}
{"id": "2207.13374", "title": "Efficient Video Deblurring Guided by Motion Magnitude", "abstract": "  Video deblurring is a highly under-constrained problem due to the spatially\nand temporally varying blur. An intuitive approach for video deblurring\nincludes two steps: a) detecting the blurry region in the current frame; b)\nutilizing the information from clear regions in adjacent frames for current\nframe deblurring. To realize this process, our idea is to detect the pixel-wise\nblur level of each frame and combine it with video deblurring. To this end, we\npropose a novel framework that utilizes the motion magnitude prior (MMP) as\nguidance for efficient deep video deblurring. Specifically, as the pixel\nmovement along its trajectory during the exposure time is positively correlated\nto the level of motion blur, we first use the average magnitude of optical flow\nfrom the high-frequency sharp frames to generate the synthetic blurry frames\nand their corresponding pixel-wise motion magnitude maps. We then build a\ndataset including the blurry frame and MMP pairs. The MMP is then learned by a\ncompact CNN by regression. The MMP consists of both spatial and temporal blur\nlevel information, which can be further integrated into an efficient recurrent\nneural network (RNN) for video deblurring. We conduct intensive experiments to\nvalidate the effectiveness of the proposed methods on the public datasets.\n", "date": "2022-07-27", "categories": []}
{"id": "2207.11617", "title": "Face Deblurring using Dual Camera Fusion on Mobile Phones", "abstract": "  Motion blur of fast-moving subjects is a longstanding problem in photography\nand very common on mobile phones due to limited light collection efficiency,\nparticularly in low-light conditions. While we have witnessed great progress in\nimage deblurring in recent years, most methods require significant\ncomputational power and have limitations in processing high-resolution photos\nwith severe local motions. To this end, we develop a novel face deblurring\nsystem based on the dual camera fusion technique for mobile phones. The system\ndetects subject motion to dynamically enable a reference camera, e.g.,\nultrawide angle camera commonly available on recent premium phones, and\ncaptures an auxiliary photo with faster shutter settings. While the main shot\nis low noise but blurry, the reference shot is sharp but noisy. We learn ML\nmodels to align and fuse these two shots and output a clear photo without\nmotion blur. Our algorithm runs efficiently on Google Pixel 6, which takes 463\nms overhead per shot. Our experiments demonstrate the advantage and robustness\nof our system against alternative single-image, multi-frame, face-specific, and\nvideo deblurring algorithms as well as commercial products. To the best of our\nknowledge, our work is the first mobile solution for face motion deblurring\nthat works reliably and robustly over thousands of images in diverse motion and\nlighting conditions.\n", "date": "2022-07-23", "categories": []}
{"id": "2207.11894", "title": "Sub-Aperture Feature Adaptation in Single Image Super-resolution Model\n  for Light Field Imaging", "abstract": "  With the availability of commercial Light Field (LF) cameras, LF imaging has\nemerged as an up and coming technology in computational photography. However,\nthe spatial resolution is significantly constrained in commercial microlens\nbased LF cameras because of the inherent multiplexing of spatial and angular\ninformation. Therefore, it becomes the main bottleneck for other applications\nof light field cameras. This paper proposes an adaptation module in a\npretrained Single Image Super Resolution (SISR) network to leverage the\npowerful SISR model instead of using highly engineered light field imaging\ndomain specific Super Resolution models. The adaption module consists of a Sub\naperture Shift block and a fusion block. It is an adaptation in the SISR\nnetwork to further exploit the spatial and angular information in LF images to\nimprove the super resolution performance. Experimental validation shows that\nthe proposed method outperforms existing light field super resolution\nalgorithms. It also achieves PSNR gains of more than 1 dB across all the\ndatasets as compared to the same pretrained SISR models for scale factor 2, and\nPSNR gains 0.6 to 1 dB for scale factor 4.\n", "date": "2022-07-25", "categories": []}
{"id": "2207.12570", "title": "Seeing Far in the Dark with Patterned Flash", "abstract": "  Flash illumination is widely used in imaging under low-light environments.\nHowever, illumination intensity falls off with propagation distance\nquadratically, which poses significant challenges for flash imaging at a long\ndistance. We propose a new flash technique, named ``patterned flash'', for\nflash imaging at a long distance. Patterned flash concentrates optical power\ninto a dot array. Compared with the conventional uniform flash where the signal\nis overwhelmed by the noise everywhere, patterned flash provides stronger\nsignals at sparsely distributed points across the field of view to ensure the\nsignals at those points stand out from the sensor noise. This enables\npost-processing to resolve important objects and details. Additionally, the\npatterned flash projects texture onto the scene, which can be treated as a\nstructured light system for depth perception. Given the novel system, we\ndevelop a joint image reconstruction and depth estimation algorithm with a\nconvolutional neural network. We build a hardware prototype and test the\nproposed flash technique on various scenes. The experimental results\ndemonstrate that our patterned flash has significantly better performance at\nlong distances in low-light environments.\n", "date": "2022-07-25", "categories": []}
{"id": "2207.12607", "title": "Physics Embedded Machine Learning for Electromagnetic Data Imaging", "abstract": "  Electromagnetic (EM) imaging is widely applied in sensing for security,\nbiomedicine, geophysics, and various industries. It is an ill-posed inverse\nproblem whose solution is usually computationally expensive. Machine learning\n(ML) techniques and especially deep learning (DL) show potential in fast and\naccurate imaging. However, the high performance of purely data-driven\napproaches relies on constructing a training set that is statistically\nconsistent with practical scenarios, which is often not possible in EM imaging\ntasks. Consequently, generalizability becomes a major concern. On the other\nhand, physical principles underlie EM phenomena and provide baselines for\ncurrent imaging techniques. To benefit from prior knowledge in big data and the\ntheoretical constraint of physical laws, physics embedded ML methods for EM\nimaging have become the focus of a large body of recent work.\n  This article surveys various schemes to incorporate physics in learning-based\nEM imaging. We first introduce background on EM imaging and basic formulations\nof the inverse problem. We then focus on three types of strategies combining\nphysics and ML for linear and nonlinear imaging and discuss their advantages\nand limitations. Finally, we conclude with open challenges and possible ways\nforward in this fast-developing field. Our aim is to facilitate the study of\nintelligent EM imaging methods that will be efficient, interpretable and\ncontrollable.\n", "date": "2022-07-26", "categories": []}
{"id": "2207.11725", "title": "Combining Internal and External Constraints for Unrolling Shutter in\n  Videos", "abstract": "  Videos obtained by rolling-shutter (RS) cameras result in spatially-distorted\nframes. These distortions become significant under fast camera/scene motions.\nUndoing effects of RS is sometimes addressed as a spatial problem, where\nobjects need to be rectified/displaced in order to generate their correct\nglobal shutter (GS) frame. However, the cause of the RS effect is inherently\ntemporal, not spatial. In this paper we propose a space-time solution to the RS\nproblem. We observe that despite the severe differences between their xy\nframes, a RS video and its corresponding GS video tend to share the exact same\nxt slices -- up to a known sub-frame temporal shift. Moreover, they share the\nsame distribution of small 2D xt-patches, despite the strong temporal aliasing\nwithin each video. This allows to constrain the GS output video using\nvideo-specific constraints imposed by the RS input video. Our algorithm is\ncomposed of 3 main components: (i) Dense temporal upsampling between\nconsecutive RS frames using an off-the-shelf method, (which was trained on\nregular video sequences), from which we extract GS \"proposals\". (ii) Learning\nto correctly merge an ensemble of such GS \"proposals\" using a dedicated\nMergeNet. (iii) A video-specific zero-shot optimization which imposes the\nsimilarity of xt-patches between the GS output video and the RS input video.\nOur method obtains state-of-the-art results on benchmark datasets, both\nnumerically and visually, despite being trained on a small synthetic RS/GS\ndataset. Moreover, it generalizes well to new complex RS videos with motion\ntypes outside the distribution of the training set (e.g., complex non-rigid\nmotions) -- videos which competing methods trained on much more data cannot\nhandle well. We attribute these generalization capabilities to the combination\nof external and internal constraints.\n", "date": "2022-07-24", "categories": []}
{"id": "2207.09629", "title": "Perspective Phase Angle Model for Polarimetric 3D Reconstruction", "abstract": "  Current polarimetric 3D reconstruction methods, including those in the\nwell-established shape from polarization literature, are all developed under\nthe orthographic projection assumption. In the case of a large field of view,\nhowever, this assumption does not hold and may result in significant\nreconstruction errors in methods that make this assumption. To address this\nproblem, we present the perspective phase angle (PPA) model that is applicable\nto perspective cameras. Compared with the orthographic model, the proposed PPA\nmodel accurately describes the relationship between polarization phase angle\nand surface normal under perspective projection. In addition, the PPA model\nmakes it possible to estimate surface normals from only one single-view phase\nangle map and does not suffer from the so-called $\\pi$-ambiguity problem.\nExperiments on real data show that the PPA model is more accurate for surface\nnormal estimation with a perspective camera than the orthographic model.\n", "date": "2022-07-20", "categories": []}
{"id": "2207.10123", "title": "Animation from Blur: Multi-modal Blur Decomposition with Motion Guidance", "abstract": "  We study the challenging problem of recovering detailed motion from a single\nmotion-blurred image. Existing solutions to this problem estimate a single\nimage sequence without considering the motion ambiguity for each region.\nTherefore, the results tend to converge to the mean of the multi-modal\npossibilities. In this paper, we explicitly account for such motion ambiguity,\nallowing us to generate multiple plausible solutions all in sharp detail. The\nkey idea is to introduce a motion guidance representation, which is a compact\nquantization of 2D optical flow with only four discrete motion directions.\nConditioned on the motion guidance, the blur decomposition is led to a\nspecific, unambiguous solution by using a novel two-stage decomposition\nnetwork. We propose a unified framework for blur decomposition, which supports\nvarious interfaces for generating our motion guidance, including human input,\nmotion information from adjacent video frames, and learning from a video\ndataset. Extensive experiments on synthesized datasets and real-world data show\nthat the proposed framework is qualitatively and quantitatively superior to\nprevious methods, and also offers the merit of producing physically plausible\nand diverse solutions. Code is available at\nhttps://github.com/zzh-tech/Animation-from-Blur.\n", "date": "2022-07-20", "categories": []}
{"id": "2207.09594", "title": "ICRICS: Iterative Compensation Recovery for Image Compressive Sensing", "abstract": "  Closed-loop architecture is widely utilized in automatic control systems and\nattain distinguished performance. However, classical compressive sensing\nsystems employ open-loop architecture with separated sampling and\nreconstruction units. Therefore, a method of iterative compensation recovery\nfor image compressive sensing (ICRICS) is proposed by introducing closed-loop\nframework into traditional compresses sensing systems. The proposed method\ndepends on any existing approaches and upgrades their reconstruction\nperformance by adding negative feedback structure. Theory analysis on negative\nfeedback of compressive sensing systems is performed. An approximate\nmathematical proof of the effectiveness of the proposed method is also\nprovided. Simulation experiments on more than 3 image datasets show that the\nproposed method is superior to 10 competition approaches in reconstruction\nperformance. The maximum increment of average peak signal-to-noise ratio is\n4.36 dB and the maximum increment of average structural similarity is 0.034 on\none dataset. The proposed method based on negative feedback mechanism can\nefficiently correct the recovery error in the existing systems of image\ncompressive sensing.\n", "date": "2022-07-19", "categories": []}
{"id": "2207.01074", "title": "Variational Deep Image Restoration", "abstract": "  This paper presents a new variational inference framework for image\nrestoration and a convolutional neural network (CNN) structure that can solve\nthe restoration problems described by the proposed framework. Earlier CNN-based\nimage restoration methods primarily focused on network architecture design or\ntraining strategy with non-blind scenarios where the degradation models are\nknown or assumed. For a step closer to real-world applications, CNNs are also\nblindly trained with the whole dataset, including diverse degradations.\nHowever, the conditional distribution of a high-quality image given a diversely\ndegraded one is too complicated to be learned by a single CNN. Therefore, there\nhave also been some methods that provide additional prior information to train\na CNN. Unlike previous approaches, we focus more on the objective of\nrestoration based on the Bayesian perspective and how to reformulate the\nobjective. Specifically, our method relaxes the original posterior inference\nproblem to better manageable sub-problems and thus behaves like a\ndivide-and-conquer scheme. As a result, the proposed framework boosts the\nperformance of several restoration problems compared to the previous ones.\nSpecifically, our method delivers state-of-the-art performance on Gaussian\ndenoising, real-world noise reduction, blind image super-resolution, and JPEG\ncompression artifacts reduction.\n", "date": "2022-07-03", "categories": []}
{"id": "2207.07649", "title": "Localisation And Imaging Methods for Moving Target Ghost Imaging Radar\n  Based On Correlation Intensity Weighting", "abstract": "  Ghost imaging radar is a new system of gaze imaging radar with high detection\nsensitivity, super-resolution and better anti-interference performance, but the\nrelative motion between the radar system and the target will make the target\nimaging deteriorate. This paper proposes to perform absolute position\nlocalisation of a single target in the field of view by weighting the\ncorrelation strength of a single frame image of rough target, and to compensate\ntranslation of the reference arm speckle according to the localisation and\ntracking trajectory to accumulate the rough image into a high quality image.\nThe proposed correlation intensity weighted localization and tracking imaging\nmethod has been verified by simulation to be able to locate and image targets\nin the field of view well.\n", "date": "2022-06-22", "categories": []}
{"id": "2207.07301", "title": "Robust Deep Compressive Sensing with Recurrent-Residual Structural\n  Constraints", "abstract": "  Existing deep compressive sensing (CS) methods either ignore adaptive online\noptimization or depend on costly iterative optimizer during reconstruction.\nThis work explores a novel image CS framework with recurrent-residual\nstructural constraint, termed as R$^2$CS-NET. The R$^2$CS-NET first\nprogressively optimizes the acquired samplings through a novel recurrent neural\nnetwork. The cascaded residual convolutional network then fully reconstructs\nthe image from optimized latent representation. As the first deep CS framework\nefficiently bridging adaptive online optimization, the R$^2$CS-NET integrates\nthe robustness of online optimization with the efficiency and nonlinear\ncapacity of deep learning methods. Signal correlation has been addressed\nthrough the network architecture. The adaptive sensing nature further makes it\nan ideal candidate for color image CS via leveraging channel correlation.\nNumerical experiments verify the proposed recurrent latent optimization design\nnot only fulfills the adaptation motivation, but also outperforms classic long\nshort-term memory (LSTM) architecture in the same scenario. The overall\nframework demonstrates hardware implementation feasibility, with leading\nrobustness and generalization capability among existing deep CS benchmarks.\n", "date": "2022-07-15", "categories": []}
{"id": "2207.07317", "title": "Enhancement by Your Aesthetic: An Intelligible Unsupervised Personalized\n  Enhancer for Low-Light Images", "abstract": "  Low-light image enhancement is an inherently subjective process whose targets\nvary with the user's aesthetic. Motivated by this, several personalized\nenhancement methods have been investigated. However, the enhancement process\nbased on user preferences in these techniques is invisible, i.e., a \"black\nbox\". In this work, we propose an intelligible unsupervised personalized\nenhancer (iUPEnhancer) for low-light images, which establishes the correlations\nbetween the low-light and the unpaired reference images with regard to three\nuser-friendly attributions (brightness, chromaticity, and noise). The proposed\niUP-Enhancer is trained with the guidance of these correlations and the\ncorresponding unsupervised loss functions. Rather than a \"black box\" process,\nour iUP-Enhancer presents an intelligible enhancement process with the above\nattributions. Extensive experiments demonstrate that the proposed algorithm\nproduces competitive qualitative and quantitative results while maintaining\nexcellent flexibility and scalability. This can be validated by personalization\nwith single/multiple references, cross-attribution references, or merely\nadjusting parameters.\n", "date": "2022-07-15", "categories": []}
{"id": "2207.04175", "title": "Direct Handheld Burst Imaging to Simulated Defocus", "abstract": "  A shallow depth-of-field image keeps the subject in focus, and the foreground\nand background contexts blurred. This effect requires much larger lens\napertures than those of smartphone cameras. Conventional methods acquire RGB-D\nimages and blur image regions based on their depth. However, this approach is\nnot suitable for reflective or transparent surfaces, or finely detailed object\nsilhouettes, where the depth value is inaccurate or ambiguous.\n  We present a learning-based method to synthesize the defocus blur in shallow\ndepth-of-field images from handheld bursts acquired with a single small\naperture lens. Our deep learning model directly produces the shallow\ndepth-of-field image, avoiding explicit depth-based blurring. The simulated\naperture diameter equals the camera translation during burst acquisition. Our\nmethod does not suffer from artifacts due to inaccurate or ambiguous depth\nestimation, and it is well-suited to portrait photography.\n", "date": "2022-07-09", "categories": []}
{"id": "2207.06604", "title": "Rethinking Super-Resolution as Text-Guided Details Generation", "abstract": "  Deep neural networks have greatly promoted the performance of single image\nsuper-resolution (SISR). Conventional methods still resort to restoring the\nsingle high-resolution (HR) solution only based on the input of image modality.\nHowever, the image-level information is insufficient to predict adequate\ndetails and photo-realistic visual quality facing large upscaling factors (x8,\nx16). In this paper, we propose a new perspective that regards the SISR as a\nsemantic image detail enhancement problem to generate semantically reasonable\nHR image that are faithful to the ground truth. To enhance the semantic\naccuracy and the visual quality of the reconstructed image, we explore the\nmulti-modal fusion learning in SISR by proposing a Text-Guided Super-Resolution\n(TGSR) framework, which can effectively utilize the information from the text\nand image modalities. Different from existing methods, the proposed TGSR could\ngenerate HR image details that match the text descriptions through a\ncoarse-to-fine process. Extensive experiments and ablation studies demonstrate\nthe effect of the TGSR, which exploits the text reference to recover realistic\nimages.\n", "date": "2022-07-14", "categories": []}
{"id": "2207.06746", "title": "Single-Pixel Image Reconstruction Based on Block Compressive Sensing and\n  Deep Learning", "abstract": "  Single-pixel imaging (SPI) is a novel imaging technique whose working\nprinciple is based on the compressive sensing (CS) theory. In SPI, data is\nobtained through a series of compressive measurements and the corresponding\nimage is reconstructed. Typically, the reconstruction algorithm such as basis\npursuit relies on the sparsity assumption in images. However, recent advances\nin deep learning have found its uses in reconstructing CS images. Despite\nshowing a promising result in simulations, it is often unclear how such an\nalgorithm can be implemented in an actual SPI setup. In this paper, we\ndemonstrate the use of deep learning on the reconstruction of SPI images in\nconjunction with block compressive sensing (BCS). We also proposed a novel\nreconstruction model based on convolutional neural networks that outperforms\nother competitive CS reconstruction algorithms. Besides, by incorporating BCS\nin our deep learning model, we were able to reconstruct images of any size\nabove a certain smallest image size. In addition, we show that our model is\ncapable of reconstructing images obtained from an SPI setup while being priorly\ntrained on natural images, which can be vastly different from the SPI images.\nThis opens up opportunity for the feasibility of pretrained deep learning\nmodels for CS reconstructions of images from various domain areas.\n", "date": "2022-07-14", "categories": []}
{"id": "2207.06793", "title": "Neural apparent BRDF fields for multiview photometric stereo", "abstract": "  We propose to tackle the multiview photometric stereo problem using an\nextension of Neural Radiance Fields (NeRFs), conditioned on light source\ndirection. The geometric part of our neural representation predicts surface\nnormal direction, allowing us to reason about local surface reflectance. The\nappearance part of our neural representation is decomposed into a neural\nbidirectional reflectance function (BRDF), learnt as part of the fitting\nprocess, and a shadow prediction network (conditioned on light source\ndirection) allowing us to model the apparent BRDF. This balance of learnt\ncomponents with inductive biases based on physical image formation models\nallows us to extrapolate far from the light source and viewer directions\nobserved during training. We demonstrate our approach on a multiview\nphotometric stereo benchmark and show that competitive performance can be\nobtained with the neural density representation of a NeRF.\n", "date": "2022-07-14", "categories": []}
{"id": "2206.07050", "title": "Near-Exact Recovery for Tomographic Inverse Problems via Deep Learning", "abstract": "  This work is concerned with the following fundamental question in scientific\nmachine learning: Can deep-learning-based methods solve noise-free inverse\nproblems to near-perfect accuracy? Positive evidence is provided for the first\ntime, focusing on a prototypical computed tomography (CT) setup. We demonstrate\nthat an iterative end-to-end network scheme enables reconstructions close to\nnumerical precision, comparable to classical compressed sensing strategies. Our\nresults build on our winning submission to the recent AAPM DL-Sparse-View CT\nChallenge. Its goal was to identify the state-of-the-art in solving the\nsparse-view CT inverse problem with data-driven techniques. A specific\ndifficulty of the challenge setup was that the precise forward model remained\nunknown to the participants. Therefore, a key feature of our approach was to\ninitially estimate the unknown fanbeam geometry in a data-driven calibration\nstep. Apart from an in-depth analysis of our methodology, we also demonstrate\nits state-of-the-art performance on the open-access real-world dataset LoDoPaB\nCT.\n", "date": "2022-06-14", "categories": []}
{"id": "2207.05714", "title": "Bayesian Experimental Design for Computed Tomography with the Linearised\n  Deep Image Prior", "abstract": "  We investigate adaptive design based on a single sparse pilot scan for\ngenerating effective scanning strategies for computed tomography\nreconstruction. We propose a novel approach using the linearised deep image\nprior. It allows incorporating information from the pilot measurements into the\nangle selection criteria, while maintaining the tractability of a conjugate\nGaussian-linear model. On a synthetically generated dataset with preferential\ndirections, linearised DIP design allows reducing the number of scans by up to\n30% relative to an equidistant angle baseline.\n", "date": "2022-07-11", "categories": []}
{"id": "2207.04622", "title": "Edge-preserving Near-light Photometric Stereo with Neural Surfaces", "abstract": "  This paper presents a near-light photometric stereo method that faithfully\npreserves sharp depth edges in the 3D reconstruction. Unlike previous methods\nthat rely on finite differentiation for approximating depth partial derivatives\nand surface normals, we introduce an analytically differentiable neural surface\nin near-light photometric stereo for avoiding differentiation errors at sharp\ndepth edges, where the depth is represented as a neural function of the image\ncoordinates. By further formulating the Lambertian albedo as a dependent\nvariable resulting from the surface normal and depth, our method is\ninsusceptible to inaccurate depth initialization. Experiments on both synthetic\nand real-world scenes demonstrate the effectiveness of our method for detailed\nshape recovery with edge preservation.\n", "date": "2022-07-11", "categories": []}
{"id": "2207.03047", "title": "Single-image Defocus Deblurring by Integration of Defocus Map Prediction\n  Tracing the Inverse Problem Computation", "abstract": "  In this paper, we consider the problem in defocus image deblurring. Previous\nclassical methods follow two-steps approaches, i.e., first defocus map\nestimation and then the non-blind deblurring. In the era of deep learning, some\nresearchers have tried to address these two problems by CNN. However, the\nsimple concatenation of defocus map, which represents the blur level, leads to\nsuboptimal performance. Considering the spatial variant property of the defocus\nblur and the blur level indicated in the defocus map, we employ the defocus map\nas conditional guidance to adjust the features from the input blurring images\ninstead of simple concatenation. Then we propose a simple but effective network\nwith spatial modulation based on the defocus map. To achieve this, we design a\nnetwork consisting of three sub-networks, including the defocus map estimation\nnetwork, a condition network that encodes the defocus map into condition\nfeatures, and the defocus deblurring network that performs spatially dynamic\nmodulation based on the condition features. Moreover, the spatially dynamic\nmodulation is based on an affine transform function to adjust the features from\nthe input blurry images. Experimental results show that our method can achieve\nbetter quantitative and qualitative evaluation performance than the existing\nstate-of-the-art methods on the commonly used public test datasets.\n", "date": "2022-07-07", "categories": []}
{"id": "2207.03081", "title": "DRL-ISP: Multi-Objective Camera ISP with Deep Reinforcement Learning", "abstract": "  In this paper, we propose a multi-objective camera ISP framework that\nutilizes Deep Reinforcement Learning (DRL) and camera ISP toolbox that consist\nof network-based and conventional ISP tools. The proposed DRL-based camera ISP\nframework iteratively selects a proper tool from the toolbox and applies it to\nthe image to maximize a given vision task-specific reward function. For this\npurpose, we implement total 51 ISP tools that include exposure correction,\ncolor-and-tone correction, white balance, sharpening, denoising, and the\nothers. We also propose an efficient DRL network architecture that can extract\nthe various aspects of an image and make a rigid mapping relationship between\nimages and a large number of actions. Our proposed DRL-based ISP framework\neffectively improves the image quality according to each vision task such as\nRAW-to-RGB image restoration, 2D object detection, and monocular depth\nestimation.\n", "date": "2022-07-07", "categories": []}
{"id": "2207.02250", "title": "Array Camera Image Fusion using Physics-Aware Transformers", "abstract": "  We demonstrate a physics-aware transformer for feature-based data fusion from\ncameras with diverse resolution, color spaces, focal planes, focal lengths, and\nexposure. We also demonstrate a scalable solution for synthetic training data\ngeneration for the transformer using open-source computer graphics software. We\ndemonstrate image synthesis on arrays with diverse spectral responses,\ninstantaneous field of view and frame rate.\n", "date": "2022-07-05", "categories": []}
{"id": "2207.00943", "title": "Degradation-Guided Meta-Restoration Network for Blind Super-Resolution", "abstract": "  Blind super-resolution (SR) aims to recover high-quality visual textures from\na low-resolution (LR) image, which is usually degraded by down-sampling blur\nkernels and additive noises. This task is extremely difficult due to the\nchallenges of complicated image degradations in the real-world. Existing SR\napproaches either assume a predefined blur kernel or a fixed noise, which\nlimits these approaches in challenging cases. In this paper, we propose a\nDegradation-guided Meta-restoration network for blind Super-Resolution (DMSR)\nthat facilitates image restoration for real cases. DMSR consists of a\ndegradation extractor and meta-restoration modules. The extractor estimates the\ndegradations in LR inputs and guides the meta-restoration modules to predict\nrestoration parameters for different degradations on-the-fly. DMSR is jointly\noptimized by a novel degradation consistency loss and reconstruction losses.\nThrough such an optimization, DMSR outperforms SOTA by a large margin on three\nwidely-used benchmarks. A user study including 16 subjects further validates\nthe superiority of DMSR in real-world blind SR tasks.\n", "date": "2022-07-03", "categories": []}
{"id": "2207.01164", "title": "Aug-NeRF: Training Stronger Neural Radiance Fields with Triple-Level\n  Physically-Grounded Augmentations", "abstract": "  Neural Radiance Field (NeRF) regresses a neural parameterized scene by\ndifferentially rendering multi-view images with ground-truth supervision.\nHowever, when interpolating novel views, NeRF often yields inconsistent and\nvisually non-smooth geometric results, which we consider as a generalization\ngap between seen and unseen views. Recent advances in convolutional neural\nnetworks have demonstrated the promise of advanced robust data augmentations,\neither random or learned, in enhancing both in-distribution and\nout-of-distribution generalization. Inspired by that, we propose Augmented NeRF\n(Aug-NeRF), which for the first time brings the power of robust data\naugmentations into regularizing the NeRF training. Particularly, our proposal\nlearns to seamlessly blend worst-case perturbations into three distinct levels\nof the NeRF pipeline with physical grounds, including (1) the input\ncoordinates, to simulate imprecise camera parameters at image capture; (2)\nintermediate features, to smoothen the intrinsic feature manifold; and (3)\npre-rendering output, to account for the potential degradation factors in the\nmulti-view image supervision. Extensive results demonstrate that Aug-NeRF\neffectively boosts NeRF performance in both novel view synthesis (up to 1.5dB\nPSNR gain) and underlying geometry reconstruction. Furthermore, thanks to the\nimplicit smooth prior injected by the triple-level augmentations, Aug-NeRF can\neven recover scenes from heavily corrupted images, a highly challenging setting\nuntackled before. Our codes are available in\nhttps://github.com/VITA-Group/Aug-NeRF.\n", "date": "2022-07-04", "categories": []}
{"id": "2207.01229", "title": "Segmentation Guided Deep HDR Deghosting", "abstract": "  We present a motion segmentation guided convolutional neural network (CNN)\napproach for high dynamic range (HDR) image deghosting. First, we segment the\nmoving regions in the input sequence using a CNN. Then, we merge static and\nmoving regions separately with different fusion networks and combine fused\nfeatures to generate the final ghost-free HDR image. Our motion segmentation\nguided HDR fusion approach offers significant advantages over existing HDR\ndeghosting methods. First, by segmenting the input sequence into static and\nmoving regions, our proposed approach learns effective fusion rules for various\nchallenging saturation and motion types. Second, we introduce a novel memory\nnetwork that accumulates the necessary features required to generate plausible\ndetails in the saturated regions. The proposed method outperforms nine existing\nstate-of-the-art methods on two publicly available datasets and generates\nvisually pleasing ghost-free HDR results. We also present a large-scale motion\nsegmentation dataset of 3683 varying exposure images to benefit the research\ncommunity.\n", "date": "2022-07-04", "categories": []}
{"id": "2207.00004", "title": "Correction Algorithm of Sampling Effect and Its Application", "abstract": "  The sampling effect of the imaging acquisition device is long considered to\nbe a modulation process of the input signal, introducing additional error into\nthe signal acquisition process. This paper proposes a correction algorithm for\nthe modulation process that solves the sampling effect with high accuracy. We\nexamine the algorithm with perfect continuous Gaussian images and selected\ndigitized images, which indicate an accuracy increase of 106 for Gaussian\nimages, 102 at 15 times of Shannon interpolation for digitized images, and 105\nat 101 times of Shannon interpolation for digitized images. The accuracy limit\nof the Gaussian image comes from the truncation error, while the accuracy limit\nof the digitized images comes from their finite resolution, which can be\nimproved by increasing the time of Shannon interpolation.\n", "date": "2022-06-25", "categories": []}
{"id": "2206.12930", "title": "SVBR-NET: A Non-Blind Spatially Varying Defocus Blur Removal Network", "abstract": "  Defocus blur is a physical consequence of the optical sensors used in most\ncameras. Although it can be used as a photographic style, it is commonly viewed\nas an image degradation modeled as the convolution of a sharp image with a\nspatially-varying blur kernel. Motivated by the advance of blur estimation\nmethods in the past years, we propose a non-blind approach for image deblurring\nthat can deal with spatially-varying kernels. We introduce two encoder-decoder\nsub-networks that are fed with the blurry image and the estimated blur map,\nrespectively, and produce as output the deblurred (deconvolved) image. Each\nsub-network presents several skip connections that allow data propagation from\nlayers spread apart, and also inter-subnetwork skip connections that ease the\ncommunication between the modules. The network is trained with synthetically\nblur kernels that are augmented to emulate blur maps produced by existing blur\nestimation methods, and our experimental results show that our method works\nwell when combined with a variety of blur estimation methods.\n", "date": "2022-06-26", "categories": []}
{"id": "2206.13419", "title": "DeStripe: A Self2Self Spatio-Spectral Graph Neural Network with Unfolded\n  Hessian for Stripe Artifact Removal in Light-sheet Microscopy", "abstract": "  Light-sheet fluorescence microscopy (LSFM) is a cutting-edge volumetric\nimaging technique that allows for three-dimensional imaging of mesoscopic\nsamples with decoupled illumination and detection paths. Although the selective\nexcitation scheme of such a microscope provides intrinsic optical sectioning\nthat minimizes out-of-focus fluorescence background and sample photodamage, it\nis prone to light absorption and scattering effects, which results in uneven\nillumination and striping artifacts in the images adversely. To tackle this\nissue, in this paper, we propose a blind stripe artifact removal algorithm in\nLSFM, called DeStripe, which combines a self-supervised spatio-spectral graph\nneural network with unfolded Hessian prior. Specifically, inspired by the\ndesirable properties of Fourier transform in condensing striping information\ninto isolated values in the frequency domain, DeStripe firstly localizes the\npotentially corrupted Fourier coefficients by exploiting the structural\ndifference between unidirectional stripe artifacts and more isotropic\nforeground images. Affected Fourier coefficients can then be fed into a graph\nneural network for recovery, with a Hessian regularization unrolled to further\nensure structures in the standard image space are well preserved. Since in\nrealistic, stripe-free LSFM barely exists with a standard image acquisition\nprotocol, DeStripe is equipped with a Self2Self denoising loss term, enabling\nartifact elimination without access to stripe-free ground truth images.\nCompetitive experimental results demonstrate the efficacy of DeStripe in\nrecovering corrupted biomarkers in LSFM with both synthetic and real stripe\nartifacts.\n", "date": "2022-06-27", "categories": []}
{"id": "2206.11629", "title": "Global Sensing and Measurements Reuse for Image Compressed Sensing", "abstract": "  Recently, deep network-based image compressed sensing methods achieved high\nreconstruction quality and reduced computational overhead compared with\ntraditional methods. However, existing methods obtain measurements only from\npartial features in the network and use them only once for image\nreconstruction. They ignore there are low, mid, and high-level features in the\nnetwork\\cite{zeiler2014visualizing} and all of them are essential for\nhigh-quality reconstruction. Moreover, using measurements only once may not be\nenough for extracting richer information from measurements. To address these\nissues, we propose a novel Measurements Reuse Convolutional Compressed Sensing\nNetwork (MR-CCSNet) which employs Global Sensing Module (GSM) to collect all\nlevel features for achieving an efficient sensing and Measurements Reuse Block\n(MRB) to reuse measurements multiple times on multi-scale. Finally,\nexperimental results on three benchmark datasets show that our model can\nsignificantly outperform state-of-the-art methods.\n", "date": "2022-06-23", "categories": []}
{"id": "2206.02510", "title": "Single pixel imaging at high pixel resolutions", "abstract": "  The usually reported pixel resolution of single pixel imaging (SPI) varies\nbetween $32 \\times 32$ and $256 \\times 256$ pixels falling far below imaging\nstandards with classical methods. Low resolution results from the trade-off\nbetween the acceptable compression ratio, the limited DMD modulation frequency,\nand reasonable reconstruction time, and has not improved significantly during\nthe decade of intensive research on SPI. In this paper we show that image\nmeasurement at the full resolution of the DMD, which lasts only a fraction of a\nsecond, is possible for sparse images or in a situation when the field of view\nis limited but is a priori unknown. We propose the sampling and reconstruction\nstrategies that enable us to reconstruct sparse images at the resolution of\n$1024 \\times 768$ within the time of $0.3~$s. Non-sparse images are\nreconstructed with less details. The compression ratio is on the order of $0.4\n\\%$ which corresponds to an acquisition frequency of $7~$Hz. Sampling is\ndifferential, binary, and non-adaptive, and includes information on multiple\npartitioning of the image which later allows us to determine the actual field\nof view. Reconstruction is based on the differential Fourier domain regularized\ninversion (D-FDRI). The proposed SPI framework is an alternative to both\nadaptive SPI, which is challenging to implement in real time, and to classical\ncompressive sensing image recovery methods, which are very slow at high\nresolutions.\n", "date": "2022-06-06", "categories": []}
{"id": "2206.09243", "title": "Structured Light with Redundancy Codes", "abstract": "  Structured light (SL) systems acquire high-fidelity 3D geometry with active\nillumination projection. Conventional systems exhibit challenges when working\nin environments with strong ambient illumination, global illumination and\ncross-device interference. This paper proposes a general-purposed technique to\nimprove the robustness of SL by projecting redundant optical signals in\naddition to the native SL patterns. In this way, projected signals become more\ndistinguishable from errors. Thus the geometry information can be more easily\nrecovered using simple signal processing and the ``coding gain\" in performance\nis obtained. We propose three applications using our redundancy codes: (1) Self\nerror-correction for SL imaging under strong ambient light, (2) Error detection\nfor adaptive reconstruction under global illumination, and (3) Interference\nfiltering with device-specific projection sequence encoding, especially for\nevent camera-based SL and light curtain devices. We systematically analyze the\ndesign rules and signal processing algorithms in these applications.\nCorresponding hardware prototypes are built for evaluations on real-world\ncomplex scenes. Experimental results on the synthetic and real data demonstrate\nthe significant performance improvements in SL systems with our redundancy\ncodes.\n", "date": "2022-06-18", "categories": []}
{"id": "2206.05283", "title": "Poissonian Blurred Image Deconvolution by Framelet based Local Minimal\n  Prior", "abstract": "  Image production tools do not always create a clear image, noisy and blurry\nimages are sometimes created. Among these cases, Poissonian noise is one of the\nmost famous noises that appear in medical images and images taken in astronomy.\nBlurred image with Poissonian noise obscures important details that are of\ngreat importance in medicine or astronomy. Therefore, studying and increasing\nthe quality of images that are affected by this type of noise is always\nconsidered by researchers. In this paper, in the first step, based on framelet\ntransform, a local minimal prior is introduced, and in the next step, this tool\ntogether with fractional calculation is used for Poissonian blurred image\ndeconvolution. In the following, the model is generalized to the blind case. To\nevaluate the performance of the presented model, several images such as real\nimages have been investigated.\n", "date": "2022-06-10", "categories": []}
{"id": "2206.05047", "title": "A GPU-Accelerated Light-field Super-resolution Framework Based on Mixed\n  Noise Model and Weighted Regularization", "abstract": "  This paper presents a GPU-accelerated computational framework for\nreconstructing high resolution (HR) LF images under a mixed Gaussian-Impulse\nnoise condition. The main focus is on developing a high-performance approach\nconsidering processing speed and reconstruction quality. From a statistical\nperspective, we derive a joint $\\ell^1$-$\\ell^2$ data fidelity term for\npenalizing the HR reconstruction error taking into account the mixed noise\nsituation. For regularization, we employ the weighted non-local total variation\napproach, which allows us to effectively realize LF image prior through a\nproper weighting scheme. We show that the alternating direction method of\nmultipliers algorithm (ADMM) can be used to simplify the computation complexity\nand results in a high-performance parallel computation on the GPU Platform. An\nextensive experiment is conducted on both synthetic 4D LF dataset and natural\nimage dataset to validate the proposed SR model's robustness and evaluate the\naccelerated optimizer's performance. The experimental results show that our\napproach achieves better reconstruction quality under severe mixed-noise\nconditions as compared to the state-of-the-art approaches. In addition, the\nproposed approach overcomes the limitation of the previous work in handling\nlarge-scale SR tasks. While fitting within a single off-the-shelf GPU, the\nproposed accelerator provides an average speedup of 2.46$\\times$ and\n1.57$\\times$ for $\\times 2$ and $\\times 3$ SR tasks, respectively. In addition,\na speedup of $77\\times$ is achieved as compared to CPU execution.\n", "date": "2022-06-09", "categories": []}
{"id": "2206.05236", "title": "Optical Diffraction Tomography based on 3D Physics-Inspired Neural\n  Network (PINN)", "abstract": "  Optical diffraction tomography (ODT) is an emerging 3D imaging technique that\nis used for the 3D reconstruction of the refractive index (RI) for\nsemi-transparent samples. Various inverse models have been proposed to\nreconstruct the 3D RI based on the holographic detection of different samples\nsuch as the Born and the Rytov approximations. However, such approximations\nusually suffer from the so-called missing-cone problem that results in an\nelongation of the final reconstruction along the optical axis. Different\niterative schemes have been proposed to solve the missing cone problem relying\non physical forward models and an error function that aims at filling in the\nk-space and thus eliminating the missing-cone problem and reaching better\nreconstruction accuracy. In this paper, we propose a different approach where a\n3D neural network (NN) is employed. The NN is trained with a cost function\nderived from a physical model based on the physics of optical wave propagation.\nThe 3D NN starts with an initial guess for the 3D RI reconstruction (i.e. Born,\nor Rytov) and aims at reconstructing better 3D reconstruction based on an error\nfunction. With this technique, the NN can be trained without any examples of\nthe relation between the ill-posed reconstruction (Born or Rytov) and the\nground truth (true shape).\n", "date": "2022-06-10", "categories": []}
{"id": "2206.04124", "title": "DRHDR: A Dual branch Residual Network for Multi-Bracket High Dynamic\n  Range Imaging", "abstract": "  We introduce DRHDR, a Dual branch Residual Convolutional Neural Network for\nMulti-Bracket HDR Imaging. To address the challenges of fusing multiple\nbrackets from dynamic scenes, we propose an efficient dual branch network that\noperates on two different resolutions. The full resolution branch uses a\nDeformable Convolutional Block to align features and retain high-frequency\ndetails. A low resolution branch with a Spatial Attention Block aims to attend\nwanted areas from the non-reference brackets, and suppress displaced features\nthat could incur on ghosting artifacts. By using a dual branch approach we are\nable to achieve high quality results while constraining the computational\nresources required to estimate the HDR results.\n", "date": "2022-06-08", "categories": []}
{"id": "2206.02838", "title": "Invertible Sharpening Network for MRI Reconstruction Enhancement", "abstract": "  High-quality MRI reconstruction plays a critical role in clinical\napplications. Deep learning-based methods have achieved promising results on\nMRI reconstruction. However, most state-of-the-art methods were designed to\noptimize the evaluation metrics commonly used for natural images, such as PSNR\nand SSIM, whereas the visual quality is not primarily pursued. Compared to the\nfully-sampled images, the reconstructed images are often blurry, where\nhigh-frequency features might not be sharp enough for confident clinical\ndiagnosis. To this end, we propose an invertible sharpening network\n(InvSharpNet) to improve the visual quality of MRI reconstructions. During\ntraining, unlike the traditional methods that learn to map the input data to\nthe ground truth, InvSharpNet adapts a backward training strategy that learns a\nblurring transform from the ground truth (fully-sampled image) to the input\ndata (blurry reconstruction). During inference, the learned blurring transform\ncan be inverted to a sharpening transform leveraging the network's\ninvertibility. The experiments on various MRI datasets demonstrate that\nInvSharpNet can improve reconstruction sharpness with few artifacts. The\nresults were also evaluated by radiologists, indicating better visual quality\nand diagnostic confidence of our proposed method.\n", "date": "2022-06-06", "categories": []}
{"id": "2206.03069", "title": "Patch-based image Super Resolution using generalized Gaussian mixture\n  model", "abstract": "  Single Image Super Resolution (SISR) methods aim to recover the clean images\nin high resolution from low resolution observations.A family of patch-based\napproaches have received considerable attention and development. The minimum\nmean square error (MMSE) methodis a powerful image restoration method that uses\na probability model on the patches of images. This paper proposes an algorithm\nto learn a jointgeneralized Gaussian mixture model (GGMM) from a pair of the\nlow resolution patches and the corresponding high resolution patches fromthe\nreference data. We then reconstruct the high resolution image based on the MMSE\nmethod. Our numerical evaluations indicate that theMMSE-GGMM method competes\nwith other state of the art methods.\n", "date": "2022-06-07", "categories": []}
{"id": "2206.01740", "title": "Denoising Fast X-Ray Fluorescence Raster Scans of Paintings", "abstract": "  Macro x-ray fluorescence (XRF) imaging of cultural heritage objects, while a\npopular non-invasive technique for providing elemental distribution maps, is a\nslow acquisition process in acquiring high signal-to-noise ratio XRF volumes.\nTypically on the order of tenths of a second per pixel, a raster scanning probe\ncounts the number of photons at different energies emitted by the object under\nx-ray illumination. In an effort to reduce the scan times without sacrificing\nelemental map and XRF volume quality, we propose using dictionary learning with\na Poisson noise model as well as a color image-based prior to restore noisy,\nrapidly acquired XRF data.\n", "date": "2022-06-03", "categories": []}
{"id": "2206.02452", "title": "Universal Photometric Stereo Network using Global Lighting Contexts", "abstract": "  This paper tackles a new photometric stereo task, named universal photometric\nstereo. Unlike existing tasks that assumed specific physical lighting models;\nhence, drastically limited their usability, a solution algorithm of this task\nis supposed to work for objects with diverse shapes and materials under\narbitrary lighting variations without assuming any specific models. To solve\nthis extremely challenging task, we present a purely data-driven method, which\neliminates the prior assumption of lighting by replacing the recovery of\nphysical lighting parameters with the extraction of the generic lighting\nrepresentation, named global lighting contexts. We use them like lighting\nparameters in a calibrated photometric stereo network to recover surface normal\nvectors pixelwisely. To adapt our network to a wide variety of shapes,\nmaterials and lightings, it is trained on a new synthetic dataset which\nsimulates the appearance of objects in the wild. Our method is compared with\nother state-of-the-art uncalibrated photometric stereo methods on our test data\nto demonstrate the significance of our method.\n", "date": "2022-06-06", "categories": []}
{"id": "2206.02715", "title": "Day-to-Night Image Synthesis for Training Nighttime Neural ISPs", "abstract": "  Many flagship smartphone cameras now use a dedicated neural image signal\nprocessor (ISP) to render noisy raw sensor images to the final processed\noutput. Training nightmode ISP networks relies on large-scale datasets of image\npairs with: (1) a noisy raw image captured with a short exposure and a high ISO\ngain; and (2) a ground truth low-noise raw image captured with a long exposure\nand low ISO that has been rendered through the ISP. Capturing such image pairs\nis tedious and time-consuming, requiring careful setup to ensure alignment\nbetween the image pairs. In addition, ground truth images are often prone to\nmotion blur due to the long exposure. To address this problem, we propose a\nmethod that synthesizes nighttime images from daytime images. Daytime images\nare easy to capture, exhibit low-noise (even on smartphone cameras) and rarely\nsuffer from motion blur. We outline a processing framework to convert daytime\nraw images to have the appearance of realistic nighttime raw images with\ndifferent levels of noise. Our procedure allows us to easily produce aligned\nnoisy and clean nighttime image pairs. We show the effectiveness of our\nsynthesis framework by training neural ISPs for nightmode rendering.\nFurthermore, we demonstrate that using our synthetic nighttime images together\nwith small amounts of real data (e.g., 5% to 10%) yields performance almost on\npar with training exclusively on real nighttime images. Our dataset and code\nare available at https://github.com/SamsungLabs/day-to-night.\n", "date": "2022-06-06", "categories": []}
{"id": "2206.02748", "title": "Compound Multi-branch Feature Fusion for Real Image Restoration", "abstract": "  Image restoration is a challenging and ill-posed problem which also has been\na long-standing issue. However, most of learning based restoration methods are\nproposed to target one degradation type which means they are lack of\ngeneralization. In this paper, we proposed a multi-branch restoration model\ninspired from the Human Visual System (i.e., Retinal Ganglion Cells) which can\nachieve multiple restoration tasks in a general framework. The experiments show\nthat the proposed multi-branch architecture, called CMFNet, has competitive\nperformance results on four datasets, including image dehazing, deraindrop, and\ndeblurring, which are very common applications for autonomous cars. The source\ncode and pretrained models of three restoration tasks are available at\nhttps://github.com/FanChiMao/CMFNet.\n", "date": "2022-06-02", "categories": []}
{"id": "2206.01429", "title": "Learning rich optical embeddings for privacy-preserving lensless image\n  classification", "abstract": "  By replacing the lens with a thin optical element, lensless imaging enables\nnew applications and solutions beyond those supported by traditional camera\ndesign and post-processing, e.g. compact and lightweight form factors and\nvisual privacy. The latter arises from the highly multiplexed measurements of\nlensless cameras, which require knowledge of the imaging system to recover a\nrecognizable image. In this work, we exploit this unique multiplexing property:\ncasting the optics as an encoder that produces learned embeddings directly at\nthe camera sensor. We do so in the context of image classification, where we\njointly optimize the encoder's parameters and those of an image classifier in\nan end-to-end fashion. Our experiments show that jointly learning the lensless\noptical encoder and the digital processing allows for lower resolution\nembeddings at the sensor, and hence better privacy as it is much harder to\nrecover meaningful images from these measurements. Additional experiments show\nthat such an optimization allows for lensless measurements that are more robust\nto typical real-world image transformations. While this work focuses on\nclassification, the proposed programmable lensless camera and end-to-end\noptimization can be applied to other computational imaging tasks.\n", "date": "2022-06-03", "categories": []}
{"id": "2206.01430", "title": "LenslessPiCam: A Hardware and Software Platform for Lensless\n  Computational Imaging with a Raspberry Pi", "abstract": "  Lensless imaging seeks to replace/remove the lens in a conventional imaging\nsystem. The earliest cameras were in fact lensless, relying on long exposure\ntimes to form images on the other end of a small aperture in a darkened\nroom/container (camera obscura). The introduction of a lens allowed for more\nlight throughput and therefore shorter exposure times, while retaining sharp\nfocus. The incorporation of digital sensors readily enabled the use of\ncomputational imaging techniques to post-process and enhance raw images (e.g.\nvia deblurring, inpainting, denoising, sharpening). Recently, imaging\nscientists have started leveraging computational imaging as an integral part of\nlensless imaging systems, allowing them to form viewable images from the highly\nmultiplexed raw measurements of lensless cameras (see [5] and references\ntherein for a comprehensive treatment of lensless imaging). This represents a\nreal paradigm shift in camera system design as there is more flexibility to\ncater the hardware to the application at hand (e.g. lightweight or flat\ndesigns). This increased flexibility comes however at the price of a more\ndemanding post-processing of the raw digital recordings and a tighter\nintegration of sensing and computation, often difficult to achieve in practice\ndue to inefficient interactions between the various communities of scientists\ninvolved. With LenslessPiCam, we provide an easily accessible hardware and\nsoftware framework to enable researchers, hobbyists, and students to implement\nand explore practical and computational aspects of lensless imaging. We also\nprovide detailed guides and exercises so that LenslessPiCam can be used as an\neducational resource, and point to results from our graduate-level signal\nprocessing course.\n", "date": "2022-06-03", "categories": []}
{"id": "2206.00812", "title": "Modeling sRGB Camera Noise with Normalizing Flows", "abstract": "  Noise modeling and reduction are fundamental tasks in low-level computer\nvision. They are particularly important for smartphone cameras relying on small\nsensors that exhibit visually noticeable noise. There has recently been renewed\ninterest in using data-driven approaches to improve camera noise models via\nneural networks. These data-driven approaches target noise present in the\nraw-sensor image before it has been processed by the camera's image signal\nprocessor (ISP). Modeling noise in the RAW-rgb domain is useful for improving\nand testing the in-camera denoising algorithm; however, there are situations\nwhere the camera's ISP does not apply denoising or additional denoising is\ndesired when the RAW-rgb domain image is no longer available. In such cases,\nthe sensor noise propagates through the ISP to the final rendered image encoded\nin standard RGB (sRGB). The nonlinear steps on the ISP culminate in a\nsignificantly more complex noise distribution in the sRGB domain and existing\nraw-domain noise models are unable to capture the sRGB noise distribution. We\npropose a new sRGB-domain noise model based on normalizing flows that is\ncapable of learning the complex noise distribution found in sRGB images under\nvarious ISO levels. Our normalizing flows-based approach outperforms other\nmodels by a large margin in noise modeling and synthesis tasks. We also show\nthat image denoisers trained on noisy images synthesized with our noise model\noutperforms those trained with noise from baselines models.\n", "date": "2022-06-02", "categories": []}
{"id": "2206.00831", "title": "Dynamic Cardiac MRI Reconstruction Using Combined Tensor Nuclear Norm\n  and Casorati Matrix Nuclear Norm Regularizations", "abstract": "  Low-rank tensor models have been applied in accelerating dynamic magnetic\nresonance imaging (dMRI). Recently, a new tensor nuclear norm based on t-SVD\nhas been proposed and applied to tensor completion. Inspired by the different\nproperties of the tensor nuclear norm (TNN) and the Casorati matrix nuclear\nnorm (MNN), we introduce a combined TNN and Casorati MNN regularizations\nframework to reconstruct dMRI, which we term as TMNN. The proposed method\nsimultaneously exploits the spatial structure and the temporal correlation of\nthe dynamic MR data. The optimization problem can be efficiently solved by the\nalternating direction method of multipliers (ADMM). In order to further improve\nthe computational efficiency, we develop a fast algorithm under the Cartesian\nsampling scenario. Numerical experiments based on cardiac cine MRI and\nperfusion MRI data demonstrate the performance improvement over the traditional\nCasorati nuclear norm regularization method.\n", "date": "2022-06-02", "categories": []}
{"id": "2206.00878", "title": "EfficientNeRF: Efficient Neural Radiance Fields", "abstract": "  Neural Radiance Fields (NeRF) has been wildly applied to various tasks for\nits high-quality representation of 3D scenes. It takes long per-scene training\ntime and per-image testing time. In this paper, we present EfficientNeRF as an\nefficient NeRF-based method to represent 3D scene and synthesize novel-view\nimages. Although several ways exist to accelerate the training or testing\nprocess, it is still difficult to much reduce time for both phases\nsimultaneously. We analyze the density and weight distribution of the sampled\npoints then propose valid and pivotal sampling at the coarse and fine stage,\nrespectively, to significantly improve sampling efficiency. In addition, we\ndesign a novel data structure to cache the whole scene during testing to\naccelerate the rendering speed. Overall, our method can reduce over 88\\% of\ntraining time, reach rendering speed of over 200 FPS, while still achieving\ncompetitive accuracy. Experiments prove that our method promotes the\npracticality of NeRF in the real world and enables many applications.\n", "date": "2022-06-02", "categories": []}
{"id": "2606.21027", "title": "Self-Supervised Dual-Frequency Phase Decomposition for Single-Shot Composite Fringe Projection Profilometry", "abstract": "Single-shot fringe projection profilometry (FPP) has been actively studied for real-time measurement, dynamic object reconstruction, and motion-sensitive environments. Composite fringe patterns are advantageous in single-shot FPP because multiple frequency components can be encoded in a single pattern, enabling phase ambiguity resolution. Existing approaches mainly rely on Fourier transform-based methods or supervised deep learning methods. However, Fourier transform-based methods often suffer from limited accuracy and degraded performance in complex regions, while supervised methods require dense phase or depth labels, which are costly to obtain. In this work, we propose a self-supervised phase refinement framework for single-shot composite fringe patterns without requiring phase or depth labels. The proposed method exploits the scale and direction relationships between low- and high-frequency phase gradients, improving the reliability of phase separation. We also introduce a soft edge consistency loss to preserve object boundaries and fine geometric structures. Experimental results show that the proposed method achieves MAE_z and RMSE_z of 0.367 mm and 1.804 mm, respectively, outperforming the best-performing transform-based baseline, which obtains 0.402 mm and 2.785 mm. The proposed method also improves the valid-pixel ratio from 84.75 % to 95.07 %. These results demonstrate the effectiveness of self-supervised dual-frequency phase refinement for reliable single-shot 3D reconstruction without ground-truth label supervision.", "date": "2026-06-19", "categories": []}
{"id": "2606.21099", "title": "ShuffleFlow: Scalable Posterior Inference for Bayesian Inverse Imaging", "abstract": "Variational inference (VI) is a powerful method for principled posterior inference for scientific inverse imaging. VI learns the posterior distribution, often with a flow-based network, which can cheaply generate posterior samples upon optimization, and can flexibly incorporate score-based or classic priors. However, its application to large-scale image reconstruction is severely hindered by the poor scalability of the flow-based networks. In this work, we introduce ShuffleFlow, a scalable VI framework to address this challenge. Our method breaks down the problem into three parts: a pixel-unshuffling-based image coordinate sampler, a neural field as feature encoder, and a conditional normalizing flow (CNF) as posterior estimator. Specifically, our framework partitions an image into a stack of sub-images with pixel-unshuffling and uses a shared CNF to model the joint distribution of the sub-image stack. We condition the CNF on the output of a neural field, which embeds feature vectors corresponding to pixel-unshuffling sample locations to capture spatial structures, and share the flow's latent variable across the channels to model their correlations. We demonstrate our method's effectiveness and efficiency on both linear and nonlinear imaging inverse problems, and show its ability to more rapidly generate a high-sample-count posterior than diffusion samplers.", "date": "2026-06-19", "categories": []}
{"id": "2606.21270", "title": "Non-line-of-sight imaging with arbitrary relay surface geometries via 3D Gaussian Transient Rendering", "abstract": "Imaging objects hidden outside the direct line of sight expands the effective field of view and is critical for applications such as autonomous driving and robotic perception. Despite impressive progress in time-of-flight (ToF)-based non-line-of-sight (NLOS) imaging, real-world deployment remains challenging because practical measurements are often collected over spatially limited, arbitrarily shaped relay regions-conditions that violate the planar-wall and dense-sampling assumptions made by most existing methods. To address these limitations, we propose a LOS-guided NLOS imaging pipeline that imposes no geometric assumptions on the relay surface and naturally supports both confocal and non-confocal configurations. Our method represents the hidden scene using 3D Gaussian primitives and couples them with an efficient, differentiable transient rendering model, enabling end-to-end optimization directly from measured transients. We validate our approach on real-world measurements from both a public dataset and a custom-built capture system. Across settings, our method achieves state-of-the-art reconstruction fidelity under spatially limited, sparsely sampled conditions, and significantly outperforms existing methods on complex, arbitrary relay surface geometries.", "date": "2026-06-19", "categories": []}
{"id": "2606.21381", "title": "OSOG: A Differentiable, Physics-Informed Synthetic Data Engine for Micro-Optical Environments", "abstract": "Deep learning in computational microscopy is severely constrained by the scarcity of densely annotated datasets. While synthetic data generation has bridged this gap in macroscopic computer vision, traditional graphics engines rely on geometric ray-tracing, failing to capture the micro-optical phenomena required for microscopy. Conversely, while wave-optics formulations exist, rendering them computationally tractable at the scale required for deep learning remains a massive systems challenge. To address this, we introduce the Optical Synthetic Object Generator (OSOG), a high-performance, fully differentiable forward-modeling engine. Drawing on established physical models of diffraction and phase retardation, OSOG maps continuous Optical Path Difference (OPD) calculations into a highly optimized, PyTorch-native Structure-of-Arrays (SoA) architecture. We validate this computational framework across three axes: First, object detection models (YOLOv11-OBB) trained purely on OSOG-generated data achieve robust zero-shot transfer to real-world highly occluded Lysozyme micrographs. Second, we introduce DiffOSOG, demonstrating that the engine's end-to-end differentiability allows for the exact recovery of continuous optical parameters via curriculum-guided inverse rendering. Finally, OSOG bypasses the $\\mathcal{O}(N)$ bottlenecks of sequential ray-tracing, demonstrating sub-linear scaling by synthesizing 40,000 complex wave-optic particles in under 50 milliseconds (\\>20 FPS). By providing a fast, scalable, and physically grounded tensor pipeline, OSOG enables true real-time, on-the-fly dataset generation.", "date": "2026-06-19", "categories": []}
{"id": "2606.21674", "title": "Enlight: Fast Low-Light Image Enhancement via Multi-Objective Optimization and Shadow-Aware Refinement", "abstract": "We present ENLIGHT, a fast and training free framework for low-light image enhancement based on direct optimization of a perceptual objective. Unlike deep learning approaches that require large scale training data and supervision, ENLIGHT operates in a zero-shot manner by optimizing image quality at inference time. The method employs a two stage global to local optimization strategy. In the first stage, ENLIGHT performs global illumination adjustment to improve visibility while maintaining structural consistency and avoiding excessive noise enhancement. In the second stage, a shadow aware refinement selectively improves low-intensity regions through masked local optimization, enhancing visibility without overexposure. To balance quality and efficiency, we introduce two modes: Fast, which uses a multi-objective formulation combining entropy, gradient preservation, and noise regularization, and Ultrafast, which reduces computational cost via a lightweight approximation of the same objective. The framework is optimizer agnostic and supports both evolutionary and lightweight local search methods. Experiments on BAID, Backlit300, LIME, MEF, NPE, and DICM demonstrate that ENLIGHT achieves competitive perceptual quality (MUSIQ, NIQE, BRISQUE) with significantly lower inference time. Qualitative results further show improved contrast, preserved structural details, and controlled noise amplification, making ENLIGHT a practical and interpretable alternative to learning based methods.", "date": "2026-06-19", "categories": []}
{"id": "2606.21713", "title": "Adaptive Beam Selection for Efficient Scanning Probe Tomography", "abstract": "In X-ray tomography, reconstruction quality generally improves with larger numbers of projections. However, more projections increase experiment costs, acquisition time and the radiation dose imparted to the sample. One mitigation to these trade-offs is to adopt a sequential design of experiments, in which each subsequent measurement is determined as a function of previously acquired data in order to maximize information gain. In practice, a widely used heuristic to maximize information is to align beams with the edges of the sample. A key challenge, however, is that the true sample is unknown, so identifying edge-aligned beams typically requires reconstructing the sample based on available measurements. This work proposes a novel sequential design method that identifies edge-aligned measurements directly from the sinogram, bypassing any reconstruction, thereby improving computational efficiency and reducing the experimental design's susceptibility to reconstruction errors. Our method dynamically selects the next set of measurement beams by maximizing an acquisition function that balances exploration and exploitation over the domain of all possible measurements, improving reconstruction quality while reducing measurement redundancy.", "date": "2026-06-19", "categories": []}
{"id": "2606.22477", "title": "Physically-guided Image Generation for Multi-Projection Mapping", "abstract": "Projection Mapping (PM) enables seamless superimposition of digital content onto real-world 3D objects, serving as a fundamental technique for immersive visualization, digital twins, and interactive art. Although text-to-image diffusion models have greatly facilitated customized content creation, directly integrating them into practical PM pipelines remains challenging due to the mismatch between idealized 2D generation and physical constraints. To bridge this gap, this paper formalizes two application-level generative paradigms: the cooperative paradigm (harmonizing generated semantics with physical attributes) and the adversarial paradigm (eliminating surface interference via radiometric compensation). Based on this, we propose ConPhyG, a unified controllable physically-guided generative multi-projection mapping framework that enables creators to interactively adjust physical constraints and flexibly switch generative paradigms. In cooperative mode, multi-dimensional physical priors (per-pixel gamut, depth, and edges) are injected into the diffusion process. In adversarial mode, the framework releases the generative potential and applies bounded numerical optimization for multi-projector radiometric compensation. It allows users to dynamically switch constraints to balance artistic freedom with physical feasibility. Furthermore, we extend ConPhyG to 360-degree multi-view consistent PM using a sequential generation strategy. Quantitative and qualitative evaluations on a real-world four-projector setup demonstrate that ConPhyG significantly outperforms state-of-the-art methods in geometric alignment, gamut utilization, and semantic fidelity.", "date": "2026-06-21", "categories": []}
{"id": "2606.23098", "title": "Poisson2Gaussian: Noise Gaussianization to Enhance Image Denoising", "abstract": "The quantum nature of light determines the inherent Poisson stochasticity of photon detection, which is ubiquitous in photography, microscopy, and astronomy. However, our controlled numerical studies reveal that the signal-dependency, heteroscedasticity, and statistical asymmetry of Poisson-mixed noise make it challenging for existing denoisers to learn. In contrast, i.i.d. Gaussian noise, with its statistical independence and symmetric distribution, is easier to model for networks. To address this gap, we propose Poisson2Gaussian (P2G), a noise Gaussianization method that explicitly converts complex real-world noise to i.i.d. Gaussian noise via probability density matching beyond low-order moments. We also design an unbiased denoising framework that synergizes P2G with downstream denoisers, ensuring convergence to the underlying signal without requiring paired clean data or explicit noise parameters. Extensive experiments demonstrate that P2G consistently achieves state-of-the-art performance across diverse datasets. In challenging scenarios where noise strongly deviates from Gaussian statistics, our method improves the PSNR by up to 0.75 dB. Notably, P2G is architecture-agnostic and can provide universal improvements for various denoisers. The source code will be publicly available.", "date": "2026-06-22", "categories": []}
{"id": "2606.23944", "title": "Stochastic Expectation Maximization for Robust State-Space Radio Interferometric Imaging", "abstract": "State--space models provide a flexible framework for analyzing dynamical systems, yet they often rely on Gaussian assumptions that fail to capture heavy-tailed or outlier-prone measurement noise. We propose a robust estimation scheme for linear state--space models subject to compound-Gaussian noise, as encountered for instance in radio interferometry affected by radio-frequency interference (RFI). The method relies on a Stochastic Approximation Expectation--Maximization (SAEM) algorithm in which the standard E-step is replaced by Monte Carlo sampling of the latent states and noise texture through closed-form Gibbs updates, enabling tractable inference despite the heavy-tailed likelihood. Numerical experiments show that the proposed method significantly improves reconstruction fidelity and robustness to RFI, outperforming a Gaussian EM algorithm and even an oracle RTS smoother. These results highlight the benefits of heavy-tailed state--space modeling and SAEM-based inference in interference-dominated imaging scenarios.", "date": "2026-06-22", "categories": []}
{"id": "2606.24072", "title": "Fabric Image Demoir\\'eing Benchmark from Synthesis to Restoration", "abstract": "Fabric moir\\'e is a sampling-induced aliasing artifact caused by the interaction between fine textile patterns and camera sensor grids, producing structured interference that severely degrades image quality. Unlike screen-induced moir\\'e, which stems from strictly periodic display lattices, fabric moir\\'e is intrinsically more challenging due to the broadband and semi-periodic nature of textile weaves. The heavy spectral overlap between intrinsic texture and aliasing components renders fabric demoir\\'eing substantially more ill-posed. Consequently, existing models trained on screen moir\\'e datasets generalize poorly to these complex textile patterns. Despite its practical importance, fabric image demoir\\'eing remains underexplored and lacks standardized benchmarks. We present the first comprehensive benchmark for fabric image demoir\\'eing. To address the difficulty of acquiring pixel-aligned real-world pairs, we develop a physically motivated synthesis framework and construct a large-scale dataset comprising 16,050 paired multi-resolution fabric images with controllable aliasing severity. Furthermore, we customize a baseline model, which establishes promising performance on the proposed benchmark dataset with strong generalization ability. Our benchmark provides a standardized platform for advancing research in fabric image demoir\\'eing.", "date": "2026-06-23", "categories": []}
{"id": "2606.24153", "title": "Differential Unfolding: Efficient Unfolding Reconstruction for Video Snapshot Compressive Imaging", "abstract": "While Deep Unfolding Networks (DUNs) dominate video Snapshot Compressive Imaging (SCI), they remain constrained by a uniform design philosophy. Existing methods repeatedly stack high-complexity priors with identical structures, ignoring the fact that optimization trajectories converge toward static states. This results in representation stagnation, where high-cost computations are wasted on minimal feature updates. To address this inefficiency, we present Differential Unfolding (DU), a heterogeneous framework that replaces uniform repetition with dynamic evolution. Central to DU is the Differential Evolutionary Framework (DEF), which partitions the unfolding process into two complementary roles: structural anchoring and differential evolution. In this scheme, high-parameter general stages are sparsely deployed to generate high-fidelity feature foundations. Complementing these, lightweight differential stages employ a Differential Representation Prior (DRP) to propagate and refine these foundational features through a differential mechanism. By integrating Differential Representation Attention (DRA) for evolving attention maps and a Differential Modulated FFN (DM-FFN) for feature rectification, DRP effectively models cross-stage variations with minimal overhead. By focusing computational resources on dynamic evolution rather than static redundancy, DU achieves a superior trade-off between accuracy and efficiency. Extensive experiments verify that our method establishes new state-of-the-art results while significantly slashing computational overhead. https://github.com/Muyuan-Zhang/DU", "date": "2026-06-23", "categories": []}
{"id": "2606.24516", "title": "What Do Flow-Based Inverse Solvers Approximate? A Posterior-Transport View", "abstract": "A growing family of training-free solvers -- FlowDPS, FLOWER, PnP-Flow and their diffusion ancestors (DPS, DAPS) -- repurpose a pretrained flow-matching prior to solve imaging inverse problems by adding a measurement-guidance term to the deterministic probability-flow ODE. Despite strong empirical results, what these per-step corrections actually approximate -- and how far the resulting samples are from the true posterior $p(x\\mid y)$ -- has not been characterized. We give a posterior-transport account of flow-based inverse problem solving. Our starting point is a simple but consequential fact: for a \\emph{deterministic} flow prior, Bayesian conditioning is realized entirely by a \\emph{reweighting of the source distribution}, not by a drift correction; pushing the reweighted source through the \\emph{unmodified} velocity field yields exact posterior samples. From this we show that trajectory-guidance solvers can be read as the minimum-kinetic-energy \\emph{correction} field needed to morph the unconditional source into the posterior, and that FlowDPS / FLOWER / PnP-Flow correspond to distinct zeroth-order / Gaussian / proximal approximations of this single object; we bound the resulting posterior bias in Wasserstein distance. A controlled $2$D study with a closed-form posterior confirms the theory decisively: source reweighting matches the true posterior to the Monte-Carlo floor on every metric, whereas trajectory guidance incurs $200$--$800\\times$ larger error and collapses posterior modes, \\emph{regardless of guidance strength}. Guided by the analysis we propose a cheap, principled velocity-correction solver that is competitive across two in-domain priors (AFHQ, CelebA) and two out-of-distribution settings while, unlike point-estimate source-space optimizers, producing diverse posterior samples with uncertainty that correlates with reconstruction error.", "date": "2026-06-23", "categories": []}
{"id": "2606.24561", "title": "Quantum CT via Dynamic Interval Encoding and Prior-Balanced QUBO Reconstruction", "abstract": "Quadratic unconstrained binary optimization (QUBO)-based quantum computed tomography (CT) casts reconstruction as a binary quadratic problem for quantum annealing and hybrid quantum--classical solvers. For grayscale CT, however, image encoding is constrained by the binary-variable budget: fixed global bit-plane encodings increase QUBO size and coupling complexity as gray-level precision improves, whereas low-bit encodings introduce quantization error. We propose a QUBO-based grayscale CT reconstruction framework that combines dynamic interval encoding with prior-balanced optimization. Each refinement round encodes active pixels only within local gray-level intervals around the current estimate, and a boundary-hit-guided update rule adaptively switches between search expansion and local refinement. To improve optimization stability, the method balances projection-domain data consistency and an edge-preserving quadratic prior before forming the final QUBO. Sparse-view and limited-angle fan-beam CT experiments show that the proposed method recovers structures and gray-level distributions more faithfully than the evaluated analytic, iterative, variational, and representation-based baselines. Expressivity analysis and ablation studies further indicate that the improvement mainly arises from effective gray-level representation through dynamic local encoding and more stable data-fidelity--prior coupling. Experiments on the D-Wave hybrid binary quadratic model (BQM) solver further demonstrate that the formulation is executable on a hardware-backed hybrid quantum--classical backend.", "date": "2026-06-23", "categories": []}
{"id": "2606.24567", "title": "Multilevel Stochastic Plug-and-Play for Sparse-View CT Reconstruction", "abstract": "Sparse-view computed tomography (SVCT) reduces radiation exposure and acquisition time, but the limited number of projection views makes the reconstruction problem severely ill-posed and leads to streak artifacts when analytical methods are used. Plug-and-Play (PnP) methods provide an effective way to combine data fidelity with learned image priors, while stochastic PnP methods further improve robustness by matching the denoiser input distribution through re-noising. However, these methods often require many iterations to converge, which limits their practical efficiency. In this work, we propose a multilevel (ML) stochastic PnP method for SVCT that accelerates stochastic PnP reconstruction. We highlight that, in the stochastic setting, directly enforcing prior coherence across levels would require accurately estimating fine-level prior gradients through multiple denoiser function evaluations, which substantially increases the computational cost. Motivated by this observation, we perform the multilevel steps in multiresolution analysis (MRA) approximation spaces. This choice is supported by the structure of the wavelet decomposition, which causes the prior-coherence correction to vanish in expectation, thereby avoiding costly estimation of fine-level stochastic prior gradients for the coarse-level corrections. Experiments on SVCT reconstruction show that our method, called Multilevel Stochastic Plug-and-Play (ML-SPnP), achieves reconstruction quality comparable to state-of-the-art methods while substantially reducing runtime.", "date": "2026-06-23", "categories": []}
{"id": "2606.25545", "title": "TensorLDM: A Component-Wise Latent Diffusion Model for Volumetric DTI Reconstruction from Sparse DWIs", "abstract": "Reconstructing diffusion tensors from sparse DWIs is critical for accelerating Diffusion Tensor Imaging (DTI) in clinical settings, yet current deep learning approaches frequently yield anatomically inconsistent or physically implausible tensors. We introduce TensorLDM, a component-wise latent diffusion model that processes the six tensor components through two group-specific encoders (for diagonal and off-diagonal elements) while maintaining anatomical consistency via shared DWI conditioning. TensorLDM uses an Anatomy-Conditioned Autoencoder that encourages the latent to focus on tensor properties rather than re-encoding structural information. A shared Cross-Component Attention (CCA) mechanism, applied in both autoencoder refinement and diffusion fine-tuning, models inter-component dependencies, while a Mixture-of-Experts (MoE) DWI conditioner provides component-adaptive conditioning. On the Human Connectome Project (HCP) dataset under a single-shell, four-volume sparse acquisition, TensorLDM produces the most accurate downstream tractography and tensors with near-ground-truth physical validity (SPD-violation rate 1.54% vs. 1.40%), with the best or comparable voxel-wise reconstruction accuracy. Geodesic tensor error measured by the Log-Euclidean Metric (LEM) corroborates these gains.", "date": "2026-06-24", "categories": []}
{"id": "2606.25962", "title": "A Benchmark for Heterogeneous Stereo Deblurring with Physically- and Epipolar-constrained Cross Attention", "abstract": "Modern stereo-capable smartphones enable immersive XR content capture. However, hardware heterogeneity across camera modules often causes severe asymmetric blur artifacts. Existing methods and benchmarks largely assume homogeneous stereo setups and therefore do not explicitly address such asymmetric degradation. To bridge this gap, we present a dedicated framework for heterogeneous stereo deblurring. First, we introduce the heterogeneous stereo deblurring (HSD) dataset, constructed from real smartphone stereo captures via multi-frame integration. Second, we propose physically- and epipolar-constrained cross attention (PECA), a lightweight module that restricts cross-view matching to an epipolar search window bounded by a optics-derived disparity upper bound. By enforcing physically valid disparity constraints, PECA enables efficient and reliable cross-view feature fusion. Moreover, our confidence-weighted attention with residual fusion emphasizes cross-guided deblurring when correspondences are reliable, while naturally falling back to self-deblurring in occluded or unreliable regions. PECA is architecture-agnostic and consistently improves CNN-, Transformer-, and NAFNet-based baselines. Extensive experiments on HSD show that PECA-enhanced models achieve improved restoration performance with favorable efficiency.", "date": "2026-06-24", "categories": []}
{"id": "2606.26609", "title": "LogicIR: Logic Gate Networks for Image Restoration", "abstract": "Image restoration aims to reconstruct high-quality images from degraded low-quality inputs. As the computational demands of image restoration models continue to rise, there is growing interest in lightweight architectures optimized for fast and efficient inference. Logic gate networks (LGNs), which operate using fundamental logic operations such as NAND and XOR, have recently emerged as a promising direction for achieving highly efficient computation. However, their potential remains largely untapped in the domain of image restoration. In this work, we introduce LogicIR, the first LGN specifically designed for image restoration tasks. LogicIR incorporates a UNet-inspired architecture composed entirely of logic gates. In addition, we propose a differentiable bit decoding layer and an index shuffling mechanism that improves information propagation across logic gates. Experimental results across multiple image restoration benchmarks demonstrate that LogicIR achieves strong performance with significantly reduced computational cost, establishing LogicIR as a viable and efficient alternative for image restoration. The source code is available at https://github.com/jimmy9704/LogicIR", "date": "2026-06-25", "categories": []}
{"id": "2606.26716", "title": "Dual-Prior Guided Null-Space Learning with Mixture-of-Splines for Arbitrary Medical Slice Super-Resolution", "abstract": "Arbitrary slice super-resolution reconstructs isotropic volumes from anisotropic clinical acquisitions by synthesizing intermediate slices at arbitrary scales. However, treating this ill-posed inverse problem as unconstrained residual-based regression risks hallucinating anatomically implausible structures or altering the originally observed data. To address both concerns, this paper presents the Dual-Prior Null-space Learning (DP-NSL) framework, which reformulates the task as a constrained recovery process guided by two complementary priors. A Measurement-Consistent Projection (MCP) enforces a Deterministic Observation Prior: the reconstruction undergoes an exact orthogonal projection that reproduces every acquired slice with zero error, confining all learned details to the unobservable null space. Within this null space, a Mixture-of-Splines (MoS) module imposes a Geometric Continuity Prior by dynamically mixing B-spline experts of different analytic orders, allowing each anatomical region to be modeled with a content-aware level of continuity. To promote spatial coherence, a Local Spatial Consistency Decoder (LSCD) further injects local inductive bias. Experiments on three CT and one MRI benchmark show that DP-NSL outperforms existing approaches while strictly preserving measurement consistency. Code is available at https://github.com/DeepMed-Lab-ECNU/Medical-Image-Reconstruction.", "date": "2026-06-25", "categories": []}
{"id": "2606.27576", "title": "DeLux: Cross-Modal Local Artifact Restoration in Video Using Neuromorphic Data", "abstract": "Conventional RGB cameras suffer from lighting artifacts such as flare, glare, flicker, and overexposure, leading to irrecoverable information loss that necessitates computational restoration. However, existing approaches treat these problems in isolation, failing to recover structural details completely obscured by complex spatially discrete image degradations. In this paper, we propose a novel cross-modal restoration paradigm and present DeLux, a modular proof-of-concept pipeline that leverages neuromorphic event streams as a structural prior to guide the targeted detection and inpainting of lighting artifacts in RGB video. Validation on synthetic benchmarks and real-world automotive footage demonstrates that DeLux effectively suppresses local artifacts and restores affected regions. The proposed approach outperforms existing RGB-only baselines and event-guided HDR models, achieving an average MS-SSIM of over 0.99 across all artifact types and demonstrating up to an 88% reduction in artifact severity in real-world automotive footage. The synthetic artifact generation tools and curated real-world evaluation datasets are made publicly available to foster future research on cross-modal restoration.", "date": "2026-06-25", "categories": []}
{"id": "2606.27635", "title": "Denoising ICF Images with Multiplicative Uniform Noise: A Self-Supervised Study Based on the Log-Domain Noisier2Inverse Framework", "abstract": "This paper documents the implementation and evaluation of a self-supervised denoising framework on Inertial Confinement Fusion (ICF) images corrupted by Multiplicative Uniform noise: the \\emph{Log-Domain Noisier2Inverse} framework. This framework is developed and analysed in this work; the key theoretical result -- that minimising the log-domain self-supervised loss is equivalent to supervised learning in the transformed domain -- is presented with full proof. We document significant implementation challenges arising from the unique characteristics of ICF imagery, describe the fixes applied at each stage, and report final quantitative results. The log-domain approach with per-image JSON Uniform noise loading (Variant~B) achieves the best result: a mean PSNR of $21.41\\db$ and SSIM of $0.8358$, a $+19.46\\db$ improvement over the noisy input baseline of $1.95\\db$, substantially outperforming BM3D log-domain ($4.47\\db$, SSIM $0.5181$) and Noise2Self ($4.75\\db$, SSIM $0.0177$). Variant~A, using fixed Gaussian noise loading, achieves $21.39\\db$ PSNR and SSIM $0.8436$. Of the three evaluated methods, Log-Domain Noisier2Inverse and Noise2Self are entirely self-supervised during training, requiring no clean ground truth data; BM3D is a classical filter-based method requiring no training at all. The clean reference images are used solely for quantitative evaluation of all three methods.", "date": "2026-06-26", "categories": []}
{"id": "2606.27646", "title": "VLM-Aware Meta-Optic Front-End Design for Frozen Vision-Language Models", "abstract": "Conventional machine-vision pipelines typically rely on high-quality optics that produce clean, human-interpretable images, and optical design has therefore been driven by image-level criteria such as resolution, aberration correction, and pixel fidelity. However, such optics are often impractical for size-, cost-, or form-factor-constrained applications, where compact meta-optics offer an attractive alternative but operate under strict physical efficiency limits. We propose CODA, a co-design framework that optimizes a continuous-density meta-optic front-end for frozen-model recognition using differentiable image formation and adjoint-gradient updates of Maxwell-based simulations. CODA directly optimizes the cross-entropy loss of a fixed zero-shot CLIP classifier without learned reconstruction, image signal processing, or image-fidelity auxiliary objectives. In a two-dimensional simulated imaging benchmark on ImageNet-100, CODA improves CLIP ViT-L/14 zero-shot accuracy from 53.75 $\\pm$ 3.57$\\%$ with a focal-concentration baseline to 65.41 $\\pm$ 3.99$\\%$. The optimized optics further transfer without re-optimization across CLIP, SigLIP, and DINOv2 on ImageNet-100, CIFAR-100, and Food-101. These results demonstrate that, under constrained meta-optic imaging, downstream recognition can be improved by aligning optical design with frozen vision-model objectives rather than conventional image-formation criteria.", "date": "2026-06-26", "categories": []}
{"id": "2606.27905", "title": "There and Back Again: A Flexible-Frame Transformer for Multi-Exposure Fusion", "abstract": "Multi-exposure fusion (MEF) brings the dynamic range of conventional cameras closer to that of human vision, producing images with rich scene content. Given the large variability in scene luminance, exposure strategies often require different numbers of frames to capture the full radiance range faithfully. However, conventional MEF techniques are typically designed for a fixed number of inputs, forcing deployment systems to maintain separate models for different frame-count requirements, which undermines deployment efficiency. To address this limitation, we propose FreeMEF, the first flexible-frame transformer for MEF that seamlessly accommodates varying numbers of input exposures without retraining or architectural changes. The proposed approach consists of two key modules. First, we introduce a recurrent state space module (RSSM) that sequentially fuses features from arbitrary sequences via adaptive alignment and state-space recurrent modeling, thereby providing global information guidance for the subsequent restoration. Second, we devise a global feature guided block (GFGB) incorporating an extremity-aware hybrid attention (EAHA) and an affine-injection feed-forward network (AFFN), which effectively resolves the similarity paradox while simultaneously optimizing contrast and brightness regulation. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our method, which performs favorably against state-of-the-art methods both quantitatively and qualitatively.", "date": "2026-06-26", "categories": []}
{"id": "2606.28112", "title": "BiDeMem: Bidirectional Degradation Memory for Explainable Image Restoration", "abstract": "Degradation-aware prompts, conditions, and latent priors are increasingly used in image restoration, yet they are usually judged by a single endpoint: whether the restored image obtains higher PSNR. This is a weak test of semantics. A condition can help by adding capacity, acting as a global correction bias, or exploiting dataset shortcuts, without becoming an interpretable degradation prior. We propose BiDeMem, a bidirectional degradation memory for explainable image restoration. A query built from restoration features and input statistics retrieves a compact top-k subset of memory slots. The same selected slot identity supports the restoration path at inference time and a training-only forward-degradation explanation path. The study centers on verifiability in a controlled multi-degradation NAFNet setting. New controls separate the gain from a correction head alone, a dense query prior, and a static global prior: these variants are 0.2588, 0.2586, and 0.2839 dB below BiRank, respectively. Strong residual supervision and a wider degradation head also remain below the full bidirectional memory model. Intervention probes show that BiRank preserves restoration quality while increasing wrong-prior and native-prior sensitivity, framing degradation memory as both a restoration module and a falsifiable explanation mechanism.", "date": "2026-06-26", "categories": []}
{"id": "2606.28158", "title": "Recovering Sharp Conductivity Features in the Finite-Data Calder\\'on Problem with Physics-Informed Neural Networks", "abstract": "Physics-informed neural networks (PINNs) have recently emerged as a promising framework for addressing the Calder\\'on inverse problem from limited boundary data. In this work, we revisit neural Calder\\'on inversion by introducing multiscale boundary excitations based on randomized wavelet functions and investigating the role of Fourier-feature encoding (FFE) for representing sharp conductivity variations. We propose a physics-informed reconstruction framework that represents the unknown conductivity and the associated family of electric potentials with separate neural networks conditioned on the applied boundary excitations. The governing elliptic PDE is enforced through physics-informed residuals, while finite Dirichlet-to-Neumann (DtN) data are incorporated through boundary losses. Using synthetic data from a finite-difference forward solver, we evaluate the method on conductivity fields with inclusions, sharp interfaces, smooth profiles, and heterogeneous media. Results show that the framework recovers dominant conductivity structures from finite boundary measurements with relative errors between $3\\%-12\\%$ approximately. We show that FFE improves the reconstruction of localized sharp features, particularly for inclusions and interfaces, but are not universally optimal, with raw-coordinate networks performing competitively for smoother fields. These results highlight coordinate representations and boundary excitation design as key factors in neural Calder\\'on inversion.", "date": "2026-06-26", "categories": []}
{"id": "2606.28431", "title": "A Zero-Shot Deep Image Prior Framework for Denoising and Deconvolution in Fluorescence Microscopy", "abstract": "Fluorescence microscopy images are degraded by noise and diffraction-induced blur, which compromise structural fidelity and limit quantitative analysis. Supervised deep learning methods achieve impressive restoration performance but require large-scale paired datasets that are difficult to obtain in practice. To address this issue, we propose SDIP, a zero-shot deep image prior (DIP) framework that sequentially performs denoising and deconvolution without external training data. An aSeqDIP-based module first suppresses noise while preserving fine structures through sequential autoencoding regularization. In the deconvolution stage, a wavelet-based background correction step is incorporated before the proposed RLG-DIP module performs artifact-reduced deconvolution. RLG-DIP uses the Richardson-Lucy deconvolution result as a physically consistent guidance prior, integrating the imaging model with the implicit prior of DIP to stabilize the ill-posed deconvolution process. Experiments on the BioSR dataset across multiple cellular structures demonstrate that SDIP improves both signal-to-noise ratio and resolution, achieving superior visual quality and improved quantitative performance on most evaluated structures. The proposed framework may also provide useful insights for designing physically guided DIP methods for other inverse problems.", "date": "2026-06-26", "categories": []}
{"id": "2606.28448", "title": "Measured-Subspace Consistency: A Plug-and-Play Operator for Diffusion Posterior Sampling in Accelerated MRI Reconstruction", "abstract": "Diffusion posterior samplers for accelerated MRI can reconstruct accurately yet still disagree on the acquired k-space across samples, placing posterior variability on coefficients the scanner has already measured. We identify this measured-subspace leakage as a physical-admissibility failure. Under a hard-constraint model it violates the measurement constraint and inflates the reported uncertainty with disagreement about coefficients the scanner has already determined. To quantify this leakage, we introduce complementary measured- and unmeasured-subspace k-space dispersion metrics (MSD/USD). We then present Measured-Subspace Consistency (MSC), a training-free terminal correction that wraps any compatible image-space posterior sampler with a standard multi-coil consistency lock. The ideal lock follows classical range/null-space data consistency. Our contribution is to repurpose it as a black-box posterior audit and correction rather than a new reconstructor or learned sampler. Theoretically, we prove that the ideal transform confines pairwise sample differences to the MRI null space and bound the residual cross-subspace coupling left by practical sensitivity-weighted implementations. Across six base samplers and two MRI anatomies, including out-of-distribution transfer where a knee prior reconstructs brain, MSC substantially reduces measured-subspace dispersion for Soft samplers (a median 16.5x reduction for DPS across five brain contrasts, up to ~29x), while preserving unmeasured-subspace diversity and acting as a near-identity map for Consistent ones. Furthermore, MSC maintains or modestly improves PSNR/SSIM, with no retraining, retuning, or significant computational overhead.", "date": "2026-06-26", "categories": []}
{"id": "2606.28513", "title": "HDDPM: Heteroscedastic Denoising Diffusion Probabilistic Model for Quantitative Low-Count Brain PET Recovery", "abstract": "Positron emission tomography (PET) seeks to balance diagnostic quality with ra-diation dose. Low-count PET noise is non-Gaussian, non-stationary, and spatial-ly dependent. It scales directly with local activity and is shaped by iterative recon-struction and physical corrections. Standard denoising diffusion probabilistic models (DDPMs) ignore these PET properties. Their forward process adds iso-tropic, homoscedastic Gaussian noise to the target. Such an approach fails to cap-ture the realistic physical degradation generated by the imaging system. To ad-dress the above limitations, this study introduces a heteroscedastic residual diffu-sion model (HDDPM) for low-count brain PET recovery in which the forward corruption is itself intensity-aware. We designed a fixed, Poisson-based variance module to generate voxel-wise noise maps. These maps naturally place stronger noise perturbation on low-activity regions than high-activity ones, meanwhile the network predicts the low-to-standard-count residual under explicit dose-fraction conditioning. We evaluated our proposed model (HDDPM) alongside generative frameworks across three different scanners, using both internal and external da-tasets at various simulated dose levels (1% to 50%). HDDPM and isotropic DDPM showed comparable overall image quality, but HDDPM stood out in the lowest-dose (1%) external scans. It is highly reliable and significantly reduces measurement errors in both high- and low-activity regions, compared to the standard model. These results support that heteroscedastic noising with the pro-posed HDDPM is feasible, and it provides a physically motivated inductive bias for quantitative low-count PET recovery by reflecting the activity-dependent noise structure of PET.", "date": "2026-06-26", "categories": []}
{"id": "2606.28635", "title": "AEGIR: Modeling Area Emitters for Indoor Inverse Rendering using Gaussian Splatting", "abstract": "Inverse rendering requires separating illumination from surface materials, which is highly ambiguous due to their tight coupling in observed images. While Gaussian Splatting is efficient for novel view synthesis, existing relightable methods approximate scene lighting using discrete point lights, global environment maps, or implicit representations. By ignoring the physical spatial extent of real-world emitters, these approaches produce incorrect light attenuation and unrealistic shadows. We present AEGIR (Area Emitters for Gaussian Inverse Rendering), a framework that explicitly models local area emitters within a relightable Gaussian Splatting representation. Joint optimization of emitters, materials, and geometry is challenging due to flexible emitter parameterization, which increases both the number of parameters and the ambiguity between illumination and materials. We address this by introducing a differentiable deferred rendering pipeline that integrates multiple importance sampling with targeted regularization. As a result, AEGIR accurately simulates local light transport and achieves more consistent decomposition. Experiments show that explicit area emitters improve illumination reconstruction and enhance downstream tasks, including novel view synthesis, controlled relighting, and virtual object insertion, particularly in scenes with complex local lighting.", "date": "2026-06-26", "categories": []}
{"id": "2606.28785", "title": "Stochastic Optimal Control Sampling for Diffusion Inverse Problems", "abstract": "Benefiting from the strong ability to capture data distributions, diffusion models have become powerful tools for solving image inverse problems. The key is to controllably steer the sampling trajectory toward the measurements while respecting the diffusion prior. In this work, we introduce Stochastic Optimal Control Sampling (SOCS), which models the denoising process as a dynamical system and injects control signals via SOC. Previous SOC-based approach addresses inverse problems by optimizing over the entire trajectory, which is computationally expensive. In contrast, we derive a closed-form control update and apply it at each sampling step, pulling the measurement-consistent clean prediction back onto the denoising flow. In SOCS, we can readily modulate the control strength to align with the diffusion model's native capabilities and thereby enhance perceptual quality. Our method is compatible with a variety of linear stochastic differential equation backbones. Extensive experiments across a broad spectrum of image inverse tasks demonstrate that SOCS achieves accurate measurement-aligned reconstructions with improved visual fidelity and stronger quantitative performance.", "date": "2026-06-27", "categories": []}
{"id": "2606.28787", "title": "BREIT: A Framework for Brain Stroke Reconstruction using Multi-Frequency 3D EIT", "abstract": "Multi-Frequency Electrical Impedance Tomography (MF-EIT) is a non-invasive, low-cost modality that reconstructs electrical property distributions from boundary voltages. For stroke imaging, progress in 3D deep-learning reconstruction is limited by the lack of large-scale datasets with paired ground-truth (GT) volumes and by non-standardized pipelines for data generation, simulation, and evaluation. We introduce BREIT, a modular framework for 3D MF-EIT stroke reconstruction providing: (i) a neuroimaging-to-EIT pipeline that converts CT/MRI into frequency-dependent GT admittivity volumes; (ii) a self-contained Python 3D Complete Electrode Model (CEM) forward solver for simulating MF-EIT voltages; and (iii) a 3D D-bar implementation supporting non-uniform electrode layouts. Building on BREIT, we propose dFNO-bar, which integrates Fourier Neural Operators into D-bar by learning a mapping from scattering data $t(\\xi)$ to conductivity $\\sigma(x){=}\\Re\\{\\gamma\\}$. We evaluate dFNO-bar against D-bar, Deep D-bar, and Gauss--Newton reconstructions on UCLH-matched synthetic data, and observe higher brain SSIM with comparable CC across noise settings.", "date": "2026-06-27", "categories": []}
{"id": "2606.28971", "title": "Self-Evolving Agentic Image Restoration via Deliberate Planning and Intuitive Execution", "abstract": "Real-world image restoration (IR) remains challenging due to complex and coupled degradations. While recent agentic IR frameworks leverage Large Language Models for flexible tool planning, they face two critical limitations. First, from a search scheme perspective, excessive reliance on greedy strategies fails to balance exploration and exploitation. Second, existing agentic systems underutilize information, exhibiting episodic amnesia. To address these challenges, we propose \\textbf{Self-Evolving Agentic Image Restoration (SEAR)}, which formulates restoration as a sequential decision-making problem. Inspired by the dual-process theory, SEAR comprises an Intuitive Executor and a Deliberate Planner, respectively following the fast-thinking \\textit{System 1} and slow-thinking \\textit{System 2} principles. The Deliberate Planner employs Pruning-Aware Monte Carlo Tree Search for long-horizon reasoning, utilizing a hybrid no-reference reward and a Multimodal Large Language Model (MLLM)-based tournament to prevent metric exploitation. Complementarily, the Intuitive Executor leverages a self-evolving episodic memory indexed by degradation-aware state fingerprints. This mechanism distills expensive search trajectories into adaptive expertise, overcoming episodic amnesia while progressively amortizing cold-start exploration costs through memory reuse. Extensive experiments on synthetic and real-world benchmarks demonstrate its strong perceptual and quantitative performance.", "date": "2026-06-27", "categories": []}
{"id": "2606.29093", "title": "From Fog Chamber to Aircraft Window: Pixel-Registered Imaging and Synthetic Fine-Tuning Enable Cross-Domain Defogging", "abstract": "A deep defogging pipeline pretrained on controlled laboratory fog and fine-tuned with domain-randomized synthetic fog applied to clear outdoor scenes generalizes across a graded sequence of out-of-distribution settings with no target-domain training, from chamber-free free-flowing fog to iPhone video recorded through an aircraft cabin window in flight, an entirely unseen sensor, scene, and optical path. This directly addresses an open transfer limitation reported for real-world binocular defogging. Two design choices support the transfer. First, a single-camera fog imager photographs a flat-panel display through an artificial-fog enclosure with a fixed 114~mm scattering path, producing 5{,}495 pixel-aligned foggy/clear pairs. Exact registration permits a paired Laplacian ratio that predicts per-image restoration quality far better than single-image proxies (Spearman $\\rho = 0.632$ versus $0.399$) and supports pixel-exact $L_1$ reconstruction training that avoids adversarial hallucination. Second, the fog-chamber checkpoint is fine-tuned on Mapillary Vistas crops overlaid with on-the-fly randomized synthetic fog spanning a broad range of strengths, spatial variations, airlights, and noise conditions. On a 552-image held-out split, a uniform comparison of 30 restoration backbones places NAFNet at the top (24.33~dB~/~0.7912~SSIM), with a compact alternative within 1.29~dB at 3\\% of the parameter count, and a ResNet-50 classifier confirms that the restoration preserves semantic content rather than only pixel-level structure. On unpaired aircraft-window video, NIQE decreases from a mean of 6.22 to 4.97 after fine-tuning, with temporally stable output across full-motion sequences. The same backbone, under paired supervision, also reaches 20.71~dB~/~0.683~SSIM on a non-overlapping O-HAZE/NH-HAZE split (a transferability check rather than a competitive ranking).", "date": "2026-06-27", "categories": []}
{"id": "2606.29430", "title": "EvLIR: Learning Illumination Residuals from Ordered Events for Low-Light Image Enhancement", "abstract": "Low-light image enhancement is severely ill-posed when the input frame contains missing structure, saturated noise, and weak local contrast. Event cameras provide asynchronous brightness-change observations with high temporal resolution, but prior works often treat voxel channels as an unordered or static feature stack before fusion, rather than explicitly modeling their within-window temporal evolution, weakening the temporal evidence that makes events useful. We propose EvLIR, a temporal-residual enhancement framework that learns illumination residuals from ordered events for low-light image enhancement. Given a low-light frame and its aligned event voxel, EvLIR preserves the ordered temporal bins of the event stream and introduces a Temporal Event Residual Module (TERM) to encode short-window event dynamics with a lightweight ConvGRU. The resulting temporal state is converted into a bounded illumination correction, which provides spatially adaptive photometric guidance for Retinex-style illumination estimation and subsequent reliability-aware image-event restoration. On SDE and SDSD indoor/outdoor benchmarks, EvLIR achieves the best result on eleven of twelve dataset-metric pairs, with average scores of 25.63~dB PSNR, 28.30~dB PSNR*, and 0.827 SSIM across the four benchmarks.", "date": "2026-06-28", "categories": []}
{"id": "2606.29845", "title": "Bricker to BRACE: A Bracket Exposure RAW Dataset and Restoration Model for Flicker-Banding", "abstract": "Flicker-banding (FB), arises from temporal aliasing between a camera's rolling shutter and a display's brightness modulation, degrading screen-captured image readability with color shifts and jagged patterns. Existing single-frame methods with simplified parametric stripe models cannot reliably distinguish these artifacts from genuine texture. To address this, we conduct a systematic analysis of complex FB morphologies and reveal their significant variation across exposure settings, motivating a multi-frame bracketed RAW restoration paradigm. We construct Bricker, a synthetic-real bracketed RAW dataset built via ray-tracing-based physical simulation and automated multi-exposure capture tool. We further propose BRACE: Bracketed RAW Flicker-Banding Removal, a multi-frame restoration model that utilizes frequency-aware banding prior and a multi-scale spatial cross-attention modulator (MSCAM) for cross-exposure spatial fusion. We also introduce the Stripe Frequency Consistency (SFC) metric to evaluate banding removal. Experiments demonstrate state-of-the-art performance on both synthetic and real benchmarks. Our dataset and code are available at: https://github.com/ZZH-qwq/BRACE.", "date": "2026-06-29", "categories": []}
{"id": "2606.30030", "title": "CogSENet: Blind Image Deblurring with Blur-Conditioned Semantic Routing and Explicit Frequency Fusion", "abstract": "Blind image deblurring demands the recovery of high-fidelity details and coherent structures from complex, unknown degradations. Current blind image deblurring methods struggle with real-world, spatially varying degradations, and lack the semantic awareness necessary to reliably differentiate valid textures from artifacts. To bridge this gap, we propose CogSENet, a dynamic, semantic-aligned reconstruction framework inspired by the eagle's visual system. By mimicking the eagle's active saccadic scanning, we devise a Semantic-Driven State Space Module (SDSSM) with semantic-aware token regrouping via differentiable routing, enabling prompt-conditioned long-range dependency modeling. To ensure physically interpretable recovery of textures and structures, a BiFreqFusionBlock (BFFB) mirrors functional differentiation of the eagle's retina by decomposing features into high and low frequencies using wavelet transforms. Finally, we estimate a continuous Blur Field (CBF) from blur image and fuse it with CLIP semantic priors to modulate the deepest latent features, emulating focal adaptation and enabling adaptive restoration under spatially non-uniform blur. Extensive experiments demonstrate that CogSENetoutperforms state-of-the-art deblurring methods in both visual quality and structural fidelity with fewer parameters, while also performing favorably on dehazing, deraining, and denoising tasks.", "date": "2026-06-29", "categories": []}
{"id": "2606.30159", "title": "A Dual-domain Refinement Network with FBP-based Jacobian Learning for Sparse-view Dual-Energy CT Material Decomposition", "abstract": "Dual-energy CT (DECT) exploits attenuation differences across different X-ray spectra to provide richer material information and has been widely used in medical imaging. While sparse-view acquisition can lower radiation exposure, it makes DECT material decomposition even more challenging, as the problem is nonlinear and ill-posed. Existing deep unrolling approaches generally do not explicitly incorporate the Jacobian operator induced by the nonlinear forward model, and their sparsity priors are still mainly built on conventional convolutions, which are insufficient for modeling global structural information. This study addresses the challenge of DECT multi-material decomposition in sparse-view settings by representing it as a sparse-regularized nonlinear least-squares problem. To solve it, we propose an iterative dual-domain refinement network (DECT-DRNet). In each iteration, the filtered back-projection (FBP)-based Jacobian approximation module is used first to generate an intermediate material decomposition result. Here, we characterize the forward process of material decomposition using a nonlinear operator, and then construct a theoretically grounded learnable approximation of the adjoint Jacobian operator by integrating the FBP algorithm with a U-Net into the backward process. In addition, to address the limitation of existing deep learning-based decomposition methods in globally suppressing noise and artifacts, we introduce a learnable sparse dual domain regularization term that incorporates Fourier convolutional residual blocks. This refinement block combines geometric feature extraction in the image domain with noise suppression in the frequency domain, allowing the model to capture both global and local features while maintaining structural details. DECT-DRNet demonstrates its ability to achieve more accurate material decomposition under sparse-view conditions.", "date": "2026-06-29", "categories": []}
{"id": "2606.31318", "title": "Wavelet-Optimized Pseudo-3D Accelerated Diffusion Model for Truncated Computed Laminography", "abstract": "Computed Laminography (CL) is a key technology for the nondestructive testing of large plate-shaped objects. However, field-of-view (FOV) limitations inevitably lead to truncation of projected data, an ill-posed inverse problem that causes severe reconstruction artifacts. Existing deep learning methods typically rely on 2D architectures that lack rigorous data consistency constraints. Furthermore, they conventionally confine artifact removal strictly to the FOV, discarding potentially recoverable information outside it. To overcome these limitations, we first introduce a comprehensive CL FOV analysis, categorizing the space into data-complete, data-incomplete, and data-free regions. By extending our reconstruction target to encompass the data-incomplete region, we significantly expand the effective imaging range and enhance scanning efficiency. To achieve this, we propose a novel wavelet-optimized pseudo-3D accelerated diffusion model for CL truncation reconstruction (CL-DM). Our method utilizes a standard 2D diffusion model for slice aggregation, combined with a 3D model-based iterative reconstruction (MBIR) method to ensure strict data consistency. To mitigate inter-slice discontinuities, we introduce wavelet regularization along the z-direction, paired with a translation-invariant (TI) mechanism and a low-frequency preservation strategy. Finally, we introduce a 3D fast sampling architecture, significantly accelerating inference speed. Extensive simulations and real-world experiments demonstrate that CL-DM is superior in effectively eliminating truncation artifacts and restoring high-fidelity, continuous 3D structures.", "date": "2026-06-30", "categories": []}
{"id": "2606.31353", "title": "RCL-Mamba: A Dual-domain State Space Model for Measurement-oriented Image Restoration in Rotational Sparse-View Scanning Computed Laminography", "abstract": "Rotational Scanning Computed Laminography (RCL) is widely utilized for the Non-Destructive Testing(NDT) of large planar components. However, to facilitate rapid inspection, continuous sparse-view scanning is often employed, where the angular integration effect during exposure induces rotational blur in the projection domain. Furthermore, the data incompleteness inherent in sparse sampling manifests as sparse artifacts in the reconstructed image domain. To address these cross-domain degradations, this paper proposes RCL-Mamba, a measurement-oriented dual-domain State Space Model (SSM)-based image restoration network. The framework adopts a cascaded joint processing strategy: it first corrects the rotational blur in the projection domain and subsequently suppresses the sparse artifacts in the image domain. Additionally, we design a Mamba-CNN dual-branch module to adaptively balance large-scale blur correction with local detail recovery. Evaluations on both simulated datasets and real-world Printed Circuit Board (PCB) scans demonstrate that RCL-Mamba outperforms existing baselines in blur removal, artifact suppression, and structural preservation. Line-profile-based structural measurement further verifies that the proposed method better preserves via/pad boundaries and slender trace profiles. Crucially, by reducing the required scanning views from 512 to 64, our method enhances inspection efficiency by approximately 8-fold without compromising reconstruction quality, offering a robust measurement-oriented restoration solution for high-throughput RCL inspection with improved structural measurement fidelity.", "date": "2026-06-30", "categories": []}
{"id": "2606.31959", "title": "AnyBokeh: Physics-Guided Any-to-Any Bokeh Editing with Optical Fingerprint Transfer", "abstract": "Depth-of-field control is a fundamental tool in photography, yet post-capture bokeh editing from a single image remains challenging. A practical editor should handle images captured under arbitrary focus and aperture settings. Existing methods typically assume an all-in-focus input, or first recover an all-in-focus image before rendering new bokeh. Such pipelines can discard useful blur cues from the source image and propagate reconstruction artifacts into the final edit. We introduce AnyBokeh, a physics-guided framework for any-to-any bokeh editing. Instead of treating source blur merely as a degradation to be removed, AnyBokeh estimates the source blur state with a signed circle-of-confusion map and a disparity map. By modeling the linear relation between signed circle of confusion and disparity difference, AnyBokeh estimates a source-specific optical fingerprint and transfers the source optical characteristics to the desired focus and aperture setting. A generative editor conditioned on both source and target circle-of-confusion maps then performs relative blur synthesis, enabling spatially adaptive deblurring, preservation, and defocus rendering. To support physically supervised learning, we further construct a high-fidelity synthetic dataset with accurate depth, focus distance, and full EXIF metadata. Experiments on real-world benchmarks show that AnyBokeh achieves faithful and controllable editing across any-to-any bokeh editing, all-in-focus-to-bokeh rendering, and defocus deblurring, while avoiding all-in-focus reconstruction and test-time bokeh-level calibration commonly required by existing approaches. The code and dataset will be available at https://github.com/itsmag11/AnyBokeh.", "date": "2026-06-30", "categories": []}
{"id": "2607.00251", "title": "Leveraging Phase Information to Boost Unrolled Network Learning for Image Deblurring", "abstract": "While most image deblurring techniques directly restore the spatial image variable, we propose an amplitude and phase decomposition recognizing the importance of accurate phase estimation in recovering sharp image details. To that end, we first develop novel linear minimum mean squared (LMMSE) estimators of the amplitude and phase of the blurred, noisy image observation. An iterative optimization algorithm follows that recovers the sharp image using the aforementioned LMMSE estimators. Finally, matrix parameters that are statistically determined and fixed in the iterative algorithm are now learned using a training dataset of clean and degraded observations. Our deblurring engine is dubbed UPADNet (Unrolled Phase and Amplitude Decomposition Network), such that each iteration of the underlying phase and amplitude recovery algorithm is parameterized and trained end-to-end. Experiments over benchmark evaluation datasets such as GoPro, RealBlur and COCO datasets confirm that UPADNet outperforms state of the art deep networks including those based on algorithm unrolling in the image domain. The benefits of UPADNet are even more pronounced in high noise and limited training data regimes.", "date": "2026-06-30", "categories": []}