App / src /services /cv2.py
amadmateen's picture
Setup: Unified repository structure with SatVision model and configs
fab54be
Raw History Blame Contribute Delete
5.23 kB
"""
OpenCV-based image processing services.
Handles water mask prediction, SAR processing, RGB preview generation, and overlay creation.
"""
import os
import numpy as np
import cv2
import tifffile as tiff
from src import config
def apply_hanning_window(tile_size):
"""Generate a 2D Hanning window for smooth tile blending."""
window_1d = np.hanning(tile_size)
return np.outer(window_1d, window_1d).astype(np.float32)
def apply_cloud_detection(prob_cloud, brightness):
"""Detect cloud pixels based on probability and brightness."""
return ((prob_cloud > config.CLOUD_DETECTION_THRESHOLD) &
(brightness > config.BRIGHTNESS_THRESHOLD)).astype(np.uint8)
def apply_water_detection(prob_water, ndwi, cloud_mask):
"""Detect water pixels based on probability, NDWI, and cloud exclusion."""
raw_water = (prob_water > config.NDWI_WATER_THRESHOLD) | (ndwi > config.NDWI_CLOUD_THRESHOLD)
return raw_water & (cloud_mask == 0) & (ndwi > config.NDWI_MIN_THRESHOLD)
def apply_morphological_cleanup(mask):
"""Apply closing and opening operations to clean mask."""
close_kernel = np.ones(config.CLOSE_KERNEL_SIZE, np.uint8)
mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, close_kernel)
open_kernel = np.ones(config.OPEN_KERNEL_SIZE, np.uint8)
mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, open_kernel)
return mask
def filter_small_contours(mask):
"""Remove contours smaller than minimum area threshold."""
contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
for cnt in contours:
area = cv2.contourArea(cnt)
if area < config.MIN_CONTOUR_AREA:
cv2.drawContours(mask, [cnt], -1, 0, -1)
return mask
def process_sar_image(tiff_path, preview_filename):
"""
Process SAR image for water detection and visualization.
"""
if not os.path.exists(tiff_path):
return None, None
try:
sar_db = tiff.imread(tiff_path).astype(np.float32)
water_min, water_max = config.SAR_WATER_DB_RANGE
water_mask = ((sar_db < water_min) & (sar_db > water_max)).astype(np.uint8) * 255
close_kernel = np.ones(config.CLOSE_KERNEL_SIZE, np.uint8)
water_mask = cv2.morphologyEx(water_mask, cv2.MORPH_CLOSE, close_kernel)
visual = np.clip((sar_db + config.SAR_DB_OFFSET) / config.SAR_DB_OFFSET, 0, 1) * 255
visual = visual.astype(np.uint8)
h, w = visual.shape
rgba = np.zeros((h, w, 4), dtype=np.uint8)
rgba[:, :, 0] = visual # B
rgba[:, :, 1] = visual # G
rgba[:, :, 2] = visual # R
rgba[:, :, 3] = np.where(sar_db < -90, 0, 255) # Transparency check for missing background data
save_path = os.path.join(config.DOWNLOADS_DIR, preview_filename)
cv2.imwrite(save_path, rgba)
return water_mask, preview_filename
except Exception as e:
print(f"❌ SAR Processing Error {tiff_path}: {e}", flush=True)
return None, None
def create_rgb_preview(tiff_path, filename):
"""
Create an enhanced RGB preview from optical or water mask data.
"""
if not os.path.exists(tiff_path):
return None
try:
image = tiff.imread(tiff_path).astype(np.float32)
if image.ndim == 2:
rgb = np.zeros((image.shape[0], image.shape[1], 3), dtype=np.uint8)
rgb[image == 255] = [255, 0, 0]
save_path = os.path.join(config.DOWNLOADS_DIR, filename)
cv2.imwrite(save_path, rgb)
return filename
if image.shape[0] == 6:
image = np.transpose(image, (1, 2, 0))
if np.max(image) > 1.0:
image = image / config.OPTICAL_NORMALIZATION_FACTOR
rgb = image[:, :, config.RGB_BAND_INDICES]
rgb = np.clip(rgb * config.RGB_SCALING_FACTOR, 0, 1) * 255
rgb = rgb.astype(np.uint8)
gaussian = cv2.GaussianBlur(rgb, (0, 0), config.GAUSSIAN_BLUR_SIGMA)
sharpened = cv2.addWeighted(
rgb,
config.GAUSSIAN_BLUR_SHARPEN_STRENGTH,
gaussian,
config.GAUSSIAN_BLUR_SUBTRACT_STRENGTH,
0
)
save_path = os.path.join(config.DOWNLOADS_DIR, filename)
cv2.imwrite(save_path, sharpened)
return filename
except Exception as e:
print(f"❌ RGB Preview Error {tiff_path}: {e}", flush=True)
return None
def create_overlay_png(mask, color, filename):
"""
Create a semi-transparent PNG overlay from a binary mask.
"""
if mask is None:
return None
try:
h, w = mask.shape
rgba = np.zeros((h, w, 4), dtype=np.uint8)
b, g, r = color
rgba[mask == 255, 0] = b
rgba[mask == 255, 1] = g
rgba[mask == 255, 2] = r
rgba[mask == 255, 3] = config.OVERLAY_ALPHA
save_path = os.path.join(config.DOWNLOADS_DIR, filename)
cv2.imwrite(save_path, rgba)
return filename
except Exception as e:
print(f"❌ Overlay Error {filename}: {e}", flush=True)
return None