Index exceeds matrix dimensions using pca in matlab 2013a

here is the sample code:

temp_files = dir([temp_dir '\*.jpg']); im_waves = zeros(size(temp_files,1),2551);

        for j1 = 1:size(temp_files,1)

            filename = fullfile(temp_dir,temp_files(j1).name);
            temp_im = imresize(double(imread(filename)),[50 50]);
%             temp_im = imresize((imread(filename)),[50 50]);
            [im_waves(j1,:),~] =  feature_extraction_phase2(temp_im);

        end

        mean_waves = mean(im_waves);
        centered_waves = im_waves - repmat(mean_waves,[size(temp_files,1) 1]);
        [evectors,~, evalues] = princomp(centered_waves);
        num_eigenfaces = 50;
        evectors = evectors(:, 1:num_eigenfaces); // |the error is index exceeds matrix dimensions|
        reduced_output_waves = evectors' * centered_waves';
        reduced_input_waves = evectors' * (waves_ip - mean_waves)';
        temp_score = arrayfun(@(n) 1 / (1 + norm(reduced_output_waves(:,n) - reduced_input_waves)), 1:size(temp_files,1));

       score = [score,temp_score];

i checked with debugger too,

the centered

[Mahdi]

Can you show the error message as well please(so we know which variable the error is in)?

[Datti Nagadhara Harini]

Index exceeds matrix dimensions.

Error in wavelet_compare (line 28) evectors = evectors(:, 1:num_eigenfaces);

Answer 1

[Geoff Hayes]

Your code assumes that there are 50 columns in evectors. Why? With the debugger, prior to evaluating

 evectors = evectors(:, 1:num_eigenfaces);

in the Command Window type

 size(evectors)

and observe the results. It is most likely that the number of columns is less than num_eigenfaces. What you can do instead is

 num_eigenfaces = min(size(evectors,2),50);

to use the minimum of the number of columns in evectors and 50. This way you won't encounter the index exceeds matrix dimension error.