how to find percentage score from orb
I am using opencv 2.x with python. I need get similarity between images as percentage score using orb. I tried dividing number of good matches by total number of matches but the value seems really low. How do I proceed thanks in advance.
you must have misunderstood something.
feature matching is not intended to give you a matching score for image similarity
(but to find similar points in images, that can be used for pose estimation, stitching, etc)
it's simply the wrong tool for your job.
http://www.vlfeat.org/matlab/vl_ubcma... in addition to matches generates score with sift in matlab I guess it should be possible in opencv python too with orb. If it is not possible any suggestion other than simple histogram and phash dhash etc..