Implementation of Efficient Cropping-Resistant Robust Image Hashing paper?

I’m not sure if this paper has come across your desk at all, as it were: DOI: 10.1109/ARES.2014.85 https://ieeexplore.ieee.org/document/6980335

It seems an interesting concept, though much slower than other image hashing options. It basically segments an image, and then hashes each segment. But it sees a drastic improvement in matching cropped images (When the image is cropped 50%, it can match 174/175 examples, as compared to 10/175 for simple ). It does also output data a bit differently though, giving a hash for each region it detects during image segmentation, and the check for matches is then a little more complicated than a simple ==.

Has this been considered for addition to the library? It does seem very powerful, even if it is a fair bit slower, has a different data format and check, and has a lot of tunable parameters.

[JohannesBuchner]

I would be happy to merge it if you make a contribution. #95 may also be related.

[JohannesBuchner]

@lucaguarro Please try to latest release and let us know of any bugs.