Multitarget tracking based on PHD smoother with unknown clutter spatial density

Abstract

Conventional multitarget tracking techniques assume that clutter density is known a priori and use it directly in the recursive processing. However, in practical surveillance systems, the spatial distribution density of measurements generated by clutter is unknown and time-variant. Therefore, in order to achieve better tracking performance as well as the ability to evaluate the surveillance environment, we propose a fully forward-backward probability hypothesis density (PHD) smoother integrated with clutter spatial density estimator in this paper. Details on the sequential Monte Carlo (SMC) implementation method are presented as well. Simulation results of tracking performance evaluation verify the effectiveness of the proposed PHD smoother.