How to determine the similarity (in%) between two "black and white" images?

Question

I searched for all the possible related keywords that I can think of, but the results are actually not what I am looking for, since most of the algorithms found cause serious concern about COLOR.

The idea of ​​my application is to identify a pair of images with the highest similarity.

For example, my input is: a, the image pool contains b, c, d, e. The result will be approximately equal to b (90%), d (85%), e (80%), c (20%).

My question is, what approach can be used to calculate this “image similarity”? Or should I create my own code from scratch?

Answer 1

http://en.wikipedia.org/wiki/Shape_context , , , http://www.cs.berkeley.edu/~malik/papers/BMP-shape.pdf

Answer 2

, hough. . - .

Answer 3

SSIM . SSIM . , , . wikipedia

Answer 4

, OpenCV, .

, , .

approxPolyDP() (. ) ( ) . .

:

Improvements:

• you can count approximate segments
• You can calculate the angles between every two segments.
• you can find some geometric invariant measures (line length ratio, barycentric, area ratio ...), this is the best