Hi Mahila,
To apply PCA to a dataset containing multiple images, you need to convert the image data into a suitable matrix format for processing. Typically, each image is flattened into a vector, and these vectors are stacked to form a matrix where each row (or column) represents an image. Once you have this matrix, you can compute the covariance matrix and find the eigenvalues and eigenvectors. Following is an example code:
% Example: Assume images are grayscale and of size m x n
numImages = size(eye_dbcrop_m, 2);
imageSize = size(eye_dbcrop_m{1}); % Assuming all images are the same size
flattenedSize = prod(imageSize);
% Initialize a matrix to hold the flattened images
X = zeros(numImages, flattenedSize);
% Flatten each image and store it as a row in the matrix X
for i = 1:numImages
img = eye_dbcrop_m{i};
X(i, :) = img(:)'; % Flatten and store as a row
end
% Subtract the mean from each pixel (column)
X_mean = mean(X, 1);
X_centered = X - X_mean;
% Compute the covariance matrix
covMatrix = cov(X_centered);
% Compute eigenvalues and eigenvectors
[eigenvectors, eigenvaluesMatrix] = eig(covMatrix);
% Extract eigenvalues from the diagonal matrix
eigenvalues = diag(eigenvaluesMatrix);
% Sort eigenvalues and corresponding eigenvectors
[eigenvaluesSorted, idx] = sort(eigenvalues, 'descend');
eigenvectorsSorted = eigenvectors(:, idx);
For further information please refer to the following documentation:
