I got a template image of a drawing.

I also have several images that may contain attempts to replicate it with variations (size, position, rotation).

I want to give a score of accuracy for each attempt compared to the template. I tried some opencv techniques like Hu moments, don't really get good results. Can you suggest a more effective approach or algorithm to achieve this?

I'm a debutant in image processing, so please explain in simple terms. I'm currently working with openCV in Python3 but the solution must works in Java too.