I have alot of .jpg
images with some duplicates (of the same size) that only differ slightly in their jpg comrpession level. I need to make a function that somehow returns the same md5 hash of all similar images based on their colors or color/shape/.. signature and ignore the compression level.
I thought about calculating the color average of each 4 pixels, and write it as 1 pixel into a grid, then use the result to get an md5 signature of the whole image?
I was wondering if there is a better way of doing this, maybe a library or an algorithme that does just this?