Marked Multitarget Intensity Filters

Abstract

Probability Hypothesis Density and other intensity filters are based on modeling the multitarget state as a realization of a Poisson point process (PPP). Target identifiability is lost in these models; consequently, the filters require targets to have the same motion models and data likelihood functions to be the same for all targets. These are unrealistic limitations in some applications. The Marked Multitarget Intensity Filter (MMIF) presented here enables the use of heterogeneous target motion models and data likelihood functions. The MMIF uses a marked PPP target model together with a parameterized PPP intensity function. The parametric model is an affine, joint, linear-Gaussian sum on the joint measurement-target space. The “at most one measurement per target” rule is enforced in the mean.