An image registration algorithm based on FREAK-FAST for visual SLAM

In recent years, computer vision are popular in the simultaneous localization and mapping researchers, the widely used algorithm, SIFT and SURF consume more time in the image registration process, ORB algorithm which uses binary descriptors takes a breakthrough in real-time aspect, so an improved fast image registration algorithm is proposed which is based on the binary description of the algorithm FAST and FREAK. The proposed one combines FAST algorithm on the target image corner detection, FREAK algorithm on building descriptors, KNN algorithm on descriptors matching, RANSAC algorithm on eliminating mismatched pair. Evaluating this algorithm with SIFT, SURF and ORB algorithm by OpenCV in VS2010, the experimental results show that a better performance of the proposed algorithm.