An ORB Corner Tracking Method Based on KLT

In order to solve the wrong corner tracking problem, a novel corner tracking method is proposed combining the ORB algorithm and the KLT algorithm. In this proposed method, the ORB detector is introduced to extract the corners, and the ORB descriptor is used to describe the tracked corners in KLT tracking process. The descriptors of pre-frame and cur-frame can be obtained which is used to calculate the similarity coefficient by brute force matching method. Then the wrong tracked corners can be removed by judging the similarity coefficient. Several simulations are made to examine the performance of the proposed method. The experimental results show that the proposed method can remove the wrong tracked corners and achieve robust corner tracking performance, which can be better used in the visual odometer.