Single scan multi-target tracking using joint state particle filters

In this paper, we compare some of the existing joint state particle filtering algorithms for closely spaced target tracking problem. Both maximum a posteriori (MAP) and minimum mean square error (MMSE) estimation outputs of four different algorithms are compared. We also include comparison of a non-joint state particle filter and Kalman filter for a baseline. Simulation results show that claimed performance of MAP based output is misleading and non-joint state particle filtering seems more appealing in terms of estimation performance than joint state counterparts.