DataProcessTools/Generalized_descriptors/PC_calculator_v4.m

%% PC_calculator
%{
Version: v4
Description: This code calculate PCs from image data of different heat
             loads.
Last modified: 2025/10/07
Author: Lige Zhang
Reference: International Journal of Heat and Mass Transfer 255 (2026): 127894.
%}
%% ------------------- Setup -------------------
clear; clc; close all;
format long;
%% ------------------- Configuration -------------------
% Define the main data directory where high-speed images are stored.
MainData_DIR = "\path\to\your\data"; % <-- IMPORTANT: Set your actual data path here

% Define subfolders for each heat load
heatLoads = ["15W", "30W"];

% Create a main directory for saving results
outputDir = fullfile(MainData_DIR, "PC_Results");
if ~exist(outputDir, 'dir')
    mkdir(outputDir);
end

% --- Parameters to define for PCA calculation ---
totalImages = 3000;           % Total number of images to process per heat load (N)
imageFileFormat = '%05d.jpg'; % The file name format that the high speed images are saved. 
numSubsets = 10;              % Number of subsets to divide the data into
numPCs = 10;                  % Number of principal components to calculate and save
savePCdata = true;            % Set to true to save results, false to just run

fprintf('PCA Calculation Started...\n');

%% ------------------- Main Processing Loop -------------------
% Loop over each heat load condition
for i = 1:length(heatLoads)
    currentHeatLoad = heatLoads(i);
    imageDataDir = fullfile(MainData_DIR, currentHeatLoad);
    fprintf('\nProcessing Heat Load: %s\n', currentHeatLoad);
    
    % Create a specific output folder for this heat load's results
    heatLoadOutputDir = fullfile(outputDir, currentHeatLoad);
    if ~exist(heatLoadOutputDir, 'dir')
        mkdir(heatLoadOutputDir);
    end

    imagesPerSubset = totalImages / numSubsets; % (n)

    % Loop over each of the 10 subsets
    for s = 1:numSubsets
        fprintf('  Calculating for Subset %d/%d...\n', s, numSubsets);
        
        % --- 1. Load, Resize, and Vectorize Images for the Subset ---
        
        % Determine the first and last image index for the current subset
        firstImageIdx = (s - 1) * imagesPerSubset;
        lastImageIdx = s * imagesPerSubset - 1;

        % Pre-allocate matrix to hold image data for efficiency
        % First, read one image to get its dimensions after resizing
        sampleImPath = fullfile(imageDataDir, sprintf(imageFileFormat, firstImageIdx));
        I_sample = imread(sampleImPath);
        
        [m, n] = size(I_sample); % Get dimensions of the image
        
        % Now pre-allocate the data matrix: p-by-n (pixels-by-images)
        imageData = zeros(m * n, imagesPerSubset);
        
        % Load all images for the current subset
        for k = 1:imagesPerSubset
            imgIndex = firstImageIdx + (k - 1);
            imFile = sprintf(imageFileFormat, imgIndex);
            
            % Read, convert to double for math, and resize
            I = double(imread(fullfile(imageDataDir, imFile)));
            % I_resized = imresize(I, 0.1);
            
            % Normalize (0-255 -> 0-1) and flatten into a column vector
            imageData(:, k) = reshape(I / 255.0, [m * n, 1]);
        end

        % --- 2. Perform PCA using SVD ---
        
        % Calculate the mean image (vector) of the subset
        meanImageVector = mean(imageData, 2);
        
        % Subtract the mean from every image (mean-centering)
        meanCenteredData = imageData - meanImageVector;
        
        % Perform SVD on the transpose of the mean-centered data
        % 'econ' is crucial for efficiency, as it calculates the "economy size" SVD
        [U, S, V] = svd(meanCenteredData', 'econ');

        

        % --- 3. Calculate and Save Principal Component Scores ---

        

        % The PC scores are simply U*S.

        

        all_PC_scores = U * S;

        

        % Extract the first 10 PCs

        PC_scores_to_save = all_PC_scores(:, 1:numPCs);

        

        % Save the results to a CSV file if requested

        if savePCdata

            % Define a descriptive filename

            fileName = sprintf('Subset_%02d_PCs.csv', s);
            csvFilePath = fullfile(heatLoadOutputDir, fileName);
            
            % Write the matrix of PC scores to the file
            writematrix(PC_scores_to_save, csvFilePath);
        end
    end
end

fprintf('\nPCA Calculation Complete. Results saved in: %s\n', outputDir);