A novel augmented particle filter for extended target tracking with star-convex random hypersurface model

The bootstrap particle filter has been used for extended target tracking problems with random hypersurface models. However, its importance function relies solely on the state equation , which leads to particle degeneracy phenomenon. To mitigate this phenomenon, we propose a novel augmented particle filter for the star-convex random hypersurface model. This filter combines the particles from the state equation and measurement equation. To sample from the measurement equation, we develop a new method by combining Gaussian-approximated likelihood. Based on this sampling method, the novel augmented particle filtering algorithm is presented. Simulation results are presented to verify the effectiveness of our proposed method. • Introduced a novel particle filter for extended target tracking . • Developed a sampling technique combining Gaussian-approximated likelihood. • The novel method can be applied to star-convex random hypersurface models. • The proposed filter can effectively limit particle degeneracy.