I've been working on how to sort images by how similar they are (not looking for duplicates), I've tried pHash and other hash algorithms, but I feel like what I need is a way to detect similar edges (for example mountains have similar edges). Can anyone suggest some methods (maybe grayscaling, contrasting, resizing?), I just can't put together how you would analyse the pixels to determine "similar" patterns
-
1use can use correlation coefficient on some image property like histogram, DFT, DCT, SIFT/SURF ... or go the other way around train a neural network with similar images so it can classify what is similar and what not (but that is for limited number of classes only)– SpektreCommented Nov 22, 2015 at 9:24
Add a comment
|