Hello everyone. Can you suggest me some documentatios to learn about analyze picture in C++? I want to wirte a program comparing shape of something in the picture.
Take an example, the hands in http://imgur.com/a/X2xk1 are in the same shape
Sorry for my bad English
Thank you
Zernike moments will compare size invariant, rotational invariant, and if memory serves, mirror invariant and give you a fuzzy value to determine if they match.
We used to reduce the image first, cut it to 1/4 or 1/10 or whatever pixels make sense based off original sizes. You don't have to check 10+ million pixels in a high res image -- often a few hundred or a couple thousand will tell you enough and its much, much faster.
These 2 things solved this type of problem for us, but the z-moments are significant amount of coding.