I use the SURF
cv2.xfeatures2d.SURF_create(400) and BRISK
cv2.BRISK_create() to detect and characterize features points in very large images in order to later register them. However, I get too many points detected when running the method
detectAndCompute. I know that for SURF I can increase the threshold for the Hessian in order to reduce the number of detected points, but I would have to test many different values to get precise the number of points I need.
Is there a way to sort all features points I get from the method
detectAndCompute and to classify them from more to less relevant?
When looking at the features I get for each point, I’m not sure what they represent. I could I know exactly what they are?