Remove Gaussian weighting of gradient orientation histogram in OpenCV SIFT (Python) -
In the SIFT algorithm, the entries of the gradient orientation histogram are Gaussian weighted, such as the orientation away from the main point center contributes less For the histogram, this ceremony can be seen in "ori_hist" I can get rid of this load as all shield orientations contribute equally to the histogram.
Basically, I will need to change the following line
w = exp (- (i * i + j * j) / exp_denom);
w = 1 Is there a Python connection that lets me do this? Or am I out of luck?
Download source code of OpenCV 2.4.4. Make the necessary changes to files (C ++), create and install it. I had a problem with the Cullman filter (in Python) once and for debugging, I had added some print statements in OpenCV source code and made it again, it works.
Since Python Binding is made up of CTIPs, if you change the C ++ code and make it, then it will work in Python.
So if you know C ++, and what you want to change, you can do it yourself.
Comments
Post a Comment