ORB-Based Fast Anti-Viewing Image Feature Matching Algorithm
LiangLiang Oi, Wei Liu, Ding Liu · 2018
Aiming at the problem that the original ORB algorithm is difficult to solve the effective matching problem under large angle of view transformation. This paper proposes an image feature matching algorithm based on ORB to resist the change of view according to the principle of ASIFT. The basic idea of the algorithm is as follows, Firstly, the longitude angle and latitude angle caused by the rotation of the camera are simulated, and the simulated image sequence set which is deformed by the change of the optical axis direction of the camera is accurately simulated. Then the feature points are extracted and mapped them into the original image by combining the FAST feature point detection method with the BRISK feature descriptors. Finally, the Brute-Force Matcher and the Random Sampling Consistency (RANSAC) method were used for feature matching and optimization. The experimental results show that the proposed algorithm can solve the problem that the image features cannot be matched under the large change of viewing angle (greater than 60°), and it has a significant improvement in real-time performance compared with ASIFT.