An extended target particle probability hypothesis density filter based on the Star-Convex shape estimation

Abstract

Abstract. The extended target is characterized by common centroid kinematic state and extended information, extended forms not only can be treated as a state to be estimated separately, including the size, shape and direction information will effectively enhance the performance of filter with proper use. For this reason, a new algorithm of the extended target particle probability hypothesis density filter modeling for Star-Convex is proposed, the algorithm take local clustering trend analysis into account and propose a method of extended target track initiation based on Star-Convex gate, then, according to the different characteristics of measurement sets, we propose an adaptive measurement partition algorithm based on extended information of Star-Convex. Simulation results show that the false initiation and computational cost both reduce significantly. In the intersection or the neighbor target tracking scenario, the proposed partition algorithm can maintain a better performance and improve the stability of the filter.