The trajectory CPHD filter for spawning targets

Abstract

This paper proposes a new trajectory cardinality probability hypothesis density (TCPHD) filter to track multiple spawning targets, referred to as the S-TCPHD filter. Firstly, we build the spawning events of single trajectory as a zero-inflated Poisson process and present corresponding trajectory set dynamic model. Then, the prediction equations of the S-TCPHD filter is derived based on the dynamic model, while the update step is the same as the TCPHD. In addition, we also develop the linear Gaussian mixture (LGM) implementations of the prediction step. The resulting S-TCPHD filter can estimate the trajectory information of targets, thereby being versatile and suitable for spawning target applications. The trajectory probability hypothesis density (TPHD) filter for spawning targets, namely S-TPHD, can be directly obtained from the S-TCPHD filter. Finally, the simulation results verify the tracking performance of the proposed S-TCPHD/S-TPHD filters in a multiple spawning targets tracking scenario.