OpenCV 4 with Python Blueprints - Second Edition

Applying point matching with rich feature descriptors

A robust way of extracting important features from an image is by using the SIFT detector. In this chapter, we want to use it for two images, self.img1 and self.img2:

    def _extract_keypoints_sift(self):        # extract keypoints and descriptors from both images        detector = cv2.xfeatures2d.SIFT_create()        first_key_points, first_desc = detector.detectAndCompute(self.img1,                                                                 None)        second_key_points, second_desc = detector.detectAndCompute(self.img2,                                                                   None)

For feature matching, we will use a BruteForce matcher, so that other matchers (such as FLANN) can work as well:

        matcher = cv2.BFMatcher(cv2.NORM_L1, True)        matches = matcher.match(first_desc, second_desc)

For each of the matches, we need to recover ...