Interactive multiple-model probability hypothesis density filter for coexisting point and extended targets tracking

In this paper, an interactive multiple-model probability hypothesis density (IMM-PHD) filter for coexisting point and extended targets is proposed, the IMM-PHD filter extends the single-model PHD filter recursion to multiple motion models by exploiting the liner Gaussian jump Markov system (LGJMS). The model of the target directly affects the tracking accuracy in multi-target tracking, the single model is not enough to guarantee stable tracking performance for maneuvering targets. The motion model of the target is usually time-varying in most realistic tracking scenarios. Therefore, it is essential to estimate the model of the target adaptively online in practice. In addition, as the resolution of sensors increases, point targets and extended targets often coexist in the field of view. Therefore, it is necessary to construct a generalized measurement model for tracking coexisting points and extended targets. This paper models the generalized IMM-PHD intensity function for coexisting point and extended targets, which can accommodate both the point target model and extended target model. To obtain a closed-form solution, the generalized IMM-PHD intensity function is modeled as a mixture of the Gaussian distribution and the Gamma Gaussian Inverse Wishart (GGIW)distribution. Finally, according to the probability of the point target, the target is determined whether the point target or the extended target, and the model probability, kinematic state of each target, and the extended state of extended targets are estimated synchronously. Simulation results verify the effectiveness of the proposed method.