Could probably get one of the many repos up and running pretty quickly [1].
Potentially what you could do is generate smaller versions of the images, test their hash matching under different conditions against multiple algorithms and then pick the parameters where you get fewest hash collisions.
Potentially what you could do is generate smaller versions of the images, test their hash matching under different conditions against multiple algorithms and then pick the parameters where you get fewest hash collisions.
[1] https://github.com/JohannesBuchner/imagehash