I have some 2bit (black & white) images which contain a lot of noise and also images.
I need to detect roughly where the images are on the page.
What I'm thinking is if I separate the file into blocks, then count the number of black pixels in each block. I can then take an average for all the blocks. Then I know that blocks with significantly more black pixels in their block than average will contain an image.
What is the best and fastest way of coding this? I can do it in any language. It's being run from command line on Debian 6.
Thank you for your help!