Oriented FAST and Rotated BRIEF Algorithm

Oriented FAST and rotated BRIEF, known as ORB, is an algorithm for detecting and describing keypoints in an image, built as a fast, patent-free alternative to SIFT and SURF for tasks such as object recognition and real time tracking. It combines the FAST corner detector, extended with an added measure of each keypoint's orientation so that matching still works when the image is rotated, with a modified version of the BRIEF descriptor, a compact binary string built by comparing the brightness of pairs of nearby pixels, adjusted here to also account for that same orientation. Ethan Rublee, Vincent Rabaud, Kurt Konolige and Gary Bradski, working at the company then known as Willow Garage, presented ORB in 2011, showing it to be markedly faster to compute and match than SIFT or SURF while remaining free of their patent restrictions.