Ask Your Question

Good feature descriptor/matcher for matching photo of different size

asked 2012-11-23 01:22:34 -0500

Stephen gravatar image

updated 2012-11-23 01:25:33 -0500

Hi All,

First of all, sorry if answer to this question is too obvious/easy. I have been playing around opencv for couple weeks, but without any background in computer vision, i am hitting a road block now and am hoping your expertise can point me to the right direction :)

i am trying to match two identical images but with different size (dimension), or one might be a cropped version of the other. There is no rotation variation.

Here are the steps that my program follows

  1. extract feature (key point)
  2. compute descriptor from key point
  3. matching descriptors of the two images

I have tried using SURF for step 1 and 2 and a brute force L2 matcher for step3. It works quite well, but since it is patented, I am trying to find a alternative. I have tried a few different combination of feature detector and descriptor extractor, but the result is not good. Is there a good combination of "feature detector/descriptor extractor/matcher"choice that might worth a try?

Thanks so much

edit retag flag offensive close merge delete

2 answers

Sort by » oldest newest most voted

answered 2012-11-26 01:49:44 -0500

Geppertm gravatar image

Hi there,

I´m using a combination of StarFeatureDetector, FREAK and BFMatcher(NORM_HAMMING).knnMatch which gives very good results. i´m also Trying a combination of BRISK/FREAK/BFMATCHER which gives less Keypoints but runs much faster.

Hope this will help.

edit flag offensive delete link more

answered 2013-09-30 05:21:51 -0500

Rui Marques gravatar image

I find ORB detector and descriptor to have a nice balance between accuracy and performance.

edit flag offensive delete link more
Login/Signup to Answer

Question Tools



Asked: 2012-11-23 01:22:34 -0500

Seen: 908 times

Last updated: Sep 30 '13