How to identify similar images even though some of them are cropped?

Question

I have 100 images, all visually similar, and I need to look for duplicates. I have an algorithm that can match one color / pattern / editing, etc., but it does not support cropping. If two similar images, one of which is cropped, the result will be different.

I need an algorithm that can match two similar images, despite cropping, somehow like tineye works. I got some links, but nothing worked.

Here's a sample article for reference .

Answer 1

I think you're on the right track, using hashing to identify almost duplicates. I believe that location-sensitive hashing can give you the extra mileage you need. It takes into account the terrain of the image functions from which it computes the hash key, and thus provides better performance for the task of detecting almost duplicates.