PROJECT TITLE :

PHD and CPHD Filtering With Unknown Detection Probability - 2018

ABSTRACT:

A priori knowledge of target detection likelihood is of vital importance within the Gaussian mixture probability hypothesis density (PHD) and cardinalized PHD (CPHD) filters. Additionally, these two filters need that the method noise and measurement noise of the state propagated within the recursion be Gaussian. These limitations may prohibit the 2 filters application in real issues. To accommodate unknown target detection probability and nonnegative non-Gaussian parameters, this Project proposes a brand new implementation based mostly on inverse gamma Gaussian mixtures, introducing a location freelance feature whose posterior likelihood density and likelihood function are nonnegative non-Gaussian inverse gamma and gamma functions to see detection chance incorporated into the recursions. The derivation of the merging inverse gamma components is additionally presented to stop the unbounded increase of mixture parts by minimizing the Kullback-Leibler divergence. Initial, a real significant-litter state of affairs is employed to validate the effectiveness of the proposed filters in track initiation and target tracking while not known detection chance. Then, simulations are presented to demonstrate that the proposed CPHD and PHD filters will achieve multitarget tracking performance almost like the quality counterparts with known target detection probability, which they outperform the quality counterparts in scenarios with unknown and dynamically changing detection likelihood. The robustness of the proposed filters is tested in both real and simulation situations. It is also shown that the analytical and empirical computational complexities of the proposed filters are just like those of their customary counterparts.