A novel CIF-based SMC-PHD approach for tracking multiple nonlinear targets

In multi-target tracking, the conventional sequential Monte Carlo probability hypothesis density (SMC-PHD) approaches use the transition density density as the importance sampling (IS) function, leading to great tracking error in nonlinear case. In this paper, we present a novel IS function approximation approach to enhance the tracking accuracy of the conventional SMC-PHD approaches in nonlinear scenarios. As for our approach, we incorporate the cubature information filter (CIF) with a gating method into the IS function approximation. Benefiting from high estimating accuracy of CIF in nonlinear target tracking, our IS function approximation approach is capable of estimating the time varying states and number in nonlinear scenario. Simulation results demonstrate the effectiveness of our approach in nonlinear multi-target tracking.