SIFT or SURF - Provide key points and retrieve descriptors

Question

I am trying to use OpenCV(2.4.6.0) to retrieve descriptors from key points that I have provided.

So far, I have not been successful...

l, des = surf.detectAndCompute(self.gray,None,useProvidedKeypoints = True)

where l is an array of feature points. I am not sure where to input the key points I already have...

Would anyone know how I can go about doing this with either SIFT or SURF?

Thank you for your help!

Answer 1

This looks like a problem with the Python bindings for detectAndCompute() , since the C++ equivalent does allow inputting of keypoints. Fortunately, there is a workaround. If you already have detected keypoints and stored them in l , then you can create a DescriptorExtractor object and compute descriptors for the provided keypoints.

An example generating FAST keypoints and then computing SURF descriptors follows:

im = cv2.imread(path_to_image)
fast = cv2.FeatureDetector_create('FAST')
l = fast.detect(im)
surf = cv2.DescriptorExtractor_create('SURF')
l, des = surf.compute(im, l)

This works equally well for SIFT features. Just pass 'SIFT' as an argument to cv2.DescriptorExtractor_create() instead.