By using SIFT descriptor, it returns some key point with related descriptor (for instance 100). How can I select 10 strongest key point among them? Should I use hessian matrix? Is there any MATLAB code to do it?
you can change thresholds when you find extremums in DOG images or find difference between extermum and mean value of window( window which consists of 28 neighbor's pixel value(see sift algorithm details)) and save them to find stronger points. The greater difference value, the stronger key point.
hi, I store SIFT Feauters in some variable name such as siftpoints but selectStrongest(scenePoints, 100) function select strongest with metric properties but sift features don't have metric properties