matlab crashes when running code on ~600 images

Hi,

i want to remove duplicate images from my portable hardDisk (win 7, 64 bit)

i have ~ 650 images, and i don't care about efficiency / running time (+ don't want to deal with memory issue's) so i wrote the following naive code with double for loop, that loads and compare each image separately, and if the images are equal adds to one filename "_double".

the code works on small amount of images, but running over ~ 600 matlab crashes:

tic
dir_path = pwd;
list = dir([dir_path '\*.jpg']);

DupCounter = 0;
fprintf('list size is %d \n',length(list));
fprintf('processing file ');
for i=1:length(list)-1
    fprintf(' %d,', i);
    if strfind( list(i).name,'_double')
        continue;
    end
    currImage = imread(list(i).name);
    for j=i+1:length(list)
        if strfind( list(j).name,'_double')
            continue;
        end
        nextImage =  imread(list(j).name);
        if isequal(currImage, nextImage)

              movefile(list(j).name,strrep(list(j).name,'.jpg','_double.jpg')) % rename
              list(j).name = strrep(list(j).name,'.jpg','_double.jpg'); % change name also in the list
              DupCounter = DupCounter +1;
          end
      end
  end

msg = sprintf('Total # duplicate frames: %d', DupCounter);
disp(msg)
toc

i get this massage:

Problem signature:

Problem Event Name: APPCRASH

 Application Name:  MATLAB.exe
  Application Version:  8.1.0.0
  Application Timestamp:  50f7698f
  Fault Module Name:  rjpg8c.mexw64
  Fault Module Version:  0.0.0.0
  Fault Module Timestamp:  50f771f6
  Exception Code:  c0000005
  Exception Offset:  00000000000048fd
  OS Version:  6.1.7601.2.1.0.768.3
  Locale ID:  1037
  Additional Information 1:  2779
  Additional Information 2:  277989f6aaa1d3ad64e8aeb5379db6ad
  Additional Information 3:  afc2
  Additional Information 4:  afc2f6520a6347247fd65c23f35ecb40

any idea?

Answer 1

[Walter Roberson]

rjpg8c is the internal routine to read jpeg files. One possibility is that you have a corrupt jpg file. You could test that by looping and just reading them each once and seeing if that works. (You would of course output just before and after each imread() so you would be able to tell which file triggered the problem.)

Answer 2

[Image Analyst]

Inside the j loop, have it print out the image it's processing.

fprintf('Comparing image#%d: %s, with image #%d: %s\n', ...
i, list(i).name, j, list(j).name);

In case it's crashing in the i loop on hte currImage line, but before the j loop, write out the i image name there:

fprintf('About to compare all other images to image #%d,: %s', i, list(i).name);

See if it's crashing on the same one all the time. See if you can call imread() on that image from the command line: