Gamma Gaussian inverse-Wishart Poisson multi-Bernoulli filter for extended target tracking

Paper i proceeding, 2016

This paper presents a gamma-Gaussian-inverse Wishart (GGIW) implementation of a Poisson multi-Bernoulli mixture (PMBM) filter for multiple extended target tracking. The GGIW density is the single extended target conjugate prior assuming a Poisson distributed number of Gaussian distributed measurements, and the PMBM density is the multi-object conju- gate prior assuming Poisson target measurements, Poisson clutter, and Poisson target birth. Specifically, the Poisson part of the GGIW-PMBM multi-object density represents the distribution of targets that have not yet been detected, and the multi-Bernoulli mixture part of the GGIW-PMBM multi-object density represents the distribution of targets that have been detected at least once. The update and the prediction of the GGIW-PMBM density parameters are given, and the filter is evaluated in a simulation study. The results show that the GGIW-PMBM filter outperforms PHD and CPHD filters for extended target tracking.