Model Parameter Adaptive Approach of Extended Object Tracking Using Random Matrix and Identification

Abstract

In many scenarios, the motion state and shape changes of the target need to be taken into account when tracking the target, as this allows for a better description of the target state. Many tracking methods based on random matrix (RM) theory tend to share a common drawback of inaccurate estimation of the extended state of the target when it undergoes high maneuvers. In this paper, an improved adaptive tracking method based on RM theory is proposed mainly for elliptical extended targets or group targets. The method uses convex packet algorithm to introduce the identification information, which successfully overcomes the drawback that the original method cannot achieve tracking due to random matrix divergence. The simulation results show that the improved adaptive method can effectively improve the tracking accuracy for elliptical extended targets.