Descriptor based on Haar wavelet response

For the extraction of the descriptor, the first step consists of constructing a square region centered around the interest point and oriented along the orientation selected in the previous section. This is not required for U-SURF. The size of the window is 20. The following steps are taken to find the descriptors:

  1. Split the interest region into 4 x 4 square subregions with 5 x 5 regularly spaced sample points inside.
  2. Calculate Haar wavelet responses dx and dy (dx = Haar wavelet response in x direction; dy = Haar wavelet response in y direction. The filter size used is 2 s).
  3. Weight the response with a Gaussian kernel centered at the interest point.


  1. Sum the response over each subregion for dx

Get Computer Vision with Python 3 now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.