Research on Object Tracking Algorithm Based on KCF

To solve the tracking failure problem of the Kernelized Correlation Filter (KCF) algorithm due to occlusion, the KCF-AO algorithm which uses the confidence of the response graph to make judgments on the tracking results of each frame is proposed in this paper. When the judgment is occlusion, the template update is stopped; When it is judged that the target is lost, the A-KAZE feature point matching algorithm and the normalized correlation coefficient matching algorithm are used to complete the re-detection of the target, and the position information is returned to the KCF to continue tracking. The partial sequence of the OTB100 data set is selected for testing. The experimental results show that the algorithm in this paper can find the target and track it accurately in the case of occlusion and target loss.