Using PCA for feature selection OCR Matlab

I have a matrix from feature extraction process (using DCT method). The size is 4096x601 (mxn) in double type.

I wanted to use PCA for dimensionality reduction without losing important information (appropriate features of the data). Expected ouput is kxn matrix.

This is my code for collecting feature. After get DCT result, I need PCA to make the dimension become smaller (efficient) to prevent 'out of memory' issue in the training stage.

srcFiles = dir('D:\...\datatrain\*.png');
    fetrain1 = [];
    for a = 1:length(srcFiles)
        file_name = strcat('D:\...\datatrain\',srcFiles(a).name);
        A = imread(file_name);
    %   [fiturtrain] = feature_extractor (A);
        [c] = CobaDCT (A);
        [fiturtrain] = cobazigzag(c);
        fetrain1 = [fetrain1 fiturtrain];

      %   vectorname = strcat(file_name,'_array.mat');
      end
       fetrain1 = pca (fetrain1);
       save ('fetrain1.mat','fetrain1');

Surely, pca function not give me the best solution, because after get a certain matrix and run training process, I get this error:

"Error using bsxfun

Non-singleton dimensions of the two input arrays must match each other."

. Any suggestions and help would be appreciated.