Sounds like you want to compare the files and not the pictures?

sub comp {
   my ($qfn1, $qfn2) = @_;

   open(my $fh1, '<:raw:stdio', $qfn1)
      or die("Unable to open file \"$qfn1\": $!\n");
   open(my $fh2, '<:raw:stdio', $qfn2)
      or die("Unable to open file \"$qfn2\": $!\n");

   my $size1 = -f $fh1 && -s _;
   my $size2 = -f $fh2 && -s _;
   return 0 if $size1 && $size2 && $size1 != $size2;

   local $/ = \(16*1024);
   for (;;) {
      my $blk1 = <$fh1>;
      my $blk2 = <$fh2>;
      return 0 if defined($blk1) xor defined($blk2);
      return 1 if !defined($blk1);
      return 0 if if $blk1 ne $blk2;
   }
}
[download]

Taking the a digest (such as MD5) of the file which speed things up if you need to compare a file against multiple other files.

Update: Added size check up front.

Why write it manually? Someone else already did it.

use File::Compare;
compare("foo.jpg", "bar.jpg");
[download]

Depending on what the OP is asking for, checking the first 16K of the file won't provide the correct answer if a) the OP is trying to determine are the images the same (rather than just are the files identical) and b) the files are in different formats (say, JPG and BMP) but are actually the same image.

Update Oops -- ikegami is quite right, the loop checks the entire file, not just the first 16K.

Alex / talexb / Toronto

"Groklaw is the open-source mentality applied to legal research" ~ Linus Torvalds

The code checks the entire file (not just the first 16K). "for (;;)" can be read as "for ever". And the post already says the code compares files not images.

looking for the fastest way to compare two picture-Files

Not sure if you are trying to do a "this picture file is a duplicate file of this picture file" or if you're tyring to do some sort of image recognition so I don't know if this is what you are asking for.

If you are just looking at two files to see if they are duplicates one thought I have is to do an MD5 sum on the two files and see if the hash matches. If they match chances are they are the same image. If not they are definitely not hte same image.

Why do people always suggest finding the digest to compare two files? Graciously assuming calculating the digest is instantaneous, using the digest method requires reading the entire file whereas one can usually exit after one pair of reads using a byte for byte comparison.

Using the digest method is great when comparing a file against multiple others over a long time period. It's a poor method when comparing two files.

If they match chances are they are the same image.

If you're going to follow up by checking the files byte for byte, it might make more sense to find the digest of the first X bytes of the file instead of the digest of the entire file.

It's out of similar considerations that some time ago I wrote dupseek. Even for multiple files, just using parts of file contents instead of digests works quite well (and rules out even the very small risk of collisions).


The stupider the astronaut, the easier it is to win the trip to Vega - A. Tucket