0

I have applied SIFT on one image and got descriptors, then I have used Euclidean distance to find similar descriptors, now I want to use k-d tree to find which descriptors are more similar and reprocess them in a data structure. Please help me how I can do it. Thanks

user615864
  • 1
  • 2

1 Answers1

0

OpenSIFT - which is a popular SIFT implementation, provides a convenient match command-line tool to compare a pair of images while using a kd-tree to pre-compute keypoints correspondences (a.k.a ratio test).

You should definitely refer to match.c#L55 for more details on how to proceed.

deltheil
  • 15,496
  • 2
  • 44
  • 64
  • thanks for your answer.I have only one image not pair of images. In one image i want to find which areas are similar to each other in case of any copy-move attack. i want to Detects SIFT features in one image and finds matches. I have to do it in MATLAB. – user615864 Mar 26 '13 at 14:01
  • The main principles stay the same (i.e you index a set of descriptors, and you loop over each point to perform a query - except that in your use case you will have to skip itself from returned NN-s). That being said I can't help specifically on MATLAB. If it is not the case you should have a look at [VLFeat MATLAB API-s](http://www.vlfeat.org/mdoc/mdoc.html). – deltheil Mar 26 '13 at 21:56