The variable structure multiple model GM-PHD filter based on likely-model set algorithm
Peng Dong, Zhongliang Jing, Minzhe Li, Han Chang Pan · International Conference on Information Fusion · 2016
The multiple model (MM) version of Gaussian mixture probability hypothesis density (GM-PHD) filter is an effective method for multiple maneuvering target tracking. However, the model set used in the MM version of GM-PHD (MM-GM-PHD) filter is the same for each target at each time step. In this paper, we present a variable structure MM-GM-PHD (VSMM-GM-PHD) filter. Different model sets at different time are used for each target, and the GM-PHD filter for variable structure MM (VSMM) is also developed. Then the likely-model set (LMS) algorithm is employed to determine the model sets used for the different targets at different time steps. In this paper, the VSMM-GM-PHD filter based on LMS is proposed. The simulation results show that the proposed algorithm can work more efficiently with better accuracy compared with the effective MM-GM-PHD filter.