Occluded Target Tracking Method Based on Multi-feature JPDA

Abstract

Aiming at the problem of tracking instability or misalignment when the target is occluded or disturbed by similar targets, this paper proposes an improved occluded target tracking method based on JPDA (Joint Probability Data Association) based on the analysis of JPDA algorithm which is prone to misalignment when the target is occluded. The algorithm judges by adding multi-dimensional attributes (area, gray level, location, etc.). Accurate correlation of inter-frame targets is achieved. the accuracy of occluded target position prediction is improved to enhance the anti-occlusion ability. The measurement image sequence is used to simulate at finally. The simulation results show that the accuracy of occluded target location extraction is significantly improved by using the improved multi-feature based algorithm JPDA, which can meet the target tracking requirements in most occluded scenarios.