Improved peak extraction algorithm in SMC implementation of PHD filter

Abstract

Based on the theory of random finite sets (RFS) and the generalized multi-target Bayes filtering, the probability hypothesis density (PHD) filter has emerged as a promising tool for the multi-target dynamic state estimation problem in recent years. However, except under some special circumstances, closed-form recursive update equations for the PHD filter do not exist and the sequential Monte Carlo (SMC) implementation has to be used. One problem caused by this SMC implementation is that the filter's output is a particle approximation of PHD, so some special algorithms are needed to extract the target's state estimation from those particles. Utilizing the information of both particles' weight and the spatial distribution, a new algorithm named C-Clean is proposed. Simulation results confirm its improved performance.