7

I see HOG is often used with SVM for target detection, can it be used in matching keypoints in two images?

and btw, where could I find OpenCV sample of using HOGDescriptor?

solvingPuzzles
  • 8,541
  • 16
  • 69
  • 112
Nauhc
  • 185
  • 2
  • 9

2 Answers2

4

HOG can be used without SVM for feature matching. just choose some points ( edge, for example ) and calculate the feature of HOG inside ROI with those points centered. HOGDescriptor seems only for GPU programming. I created Descriptor for HOG as a Mat in openCV and it also works for OpenCV matching functions.

Nauhc
  • 185
  • 2
  • 9
  • 1
    can you share some code ? Thanks! Especially the part where you do the matching between 2 images with HOG – dynamic Jul 28 '12 at 11:22
2

If you are working with images you can use SIFT/SURF with SVM. There is nothing that stops you from using HOG for keypoint matching, but bear in mind that the effectiveness depends on discrimination power and robustness of the descriptor.

Edit: My bad in understanding when I originally mentioned HOG being for video only. Somehow I was thinking about histogram of optical flow vectors which is very effective for video activity description.

Edit 2 [Oct '12]: I now suggest people to try ORB or BRISK for those looking for license friendly descriptors that are fast and quite effective for keypoint matching.

KMS
  • 766
  • 1
  • 7
  • 10
  • SIFT/SURF are both patented however. I think(?) both make allowances for non-commercial use, but certainly not for commercial use. –  Jun 06 '11 at 18:49
  • Yes, it is true that SIFT/SURF are patented. There are a few non-patented descriptors that are more or less equally effective. Your options are to either implement one of these on your own or to license SIFT/SURF. – KMS Jun 07 '11 at 23:46
  • HOG is not patented? It is later than SIFT as I know. – Nauhc Jun 28 '11 at 01:11