Multi-Target Tracking with the Progressive Gaussian Probability Hypothesis Density Filter

Particle flow filter implementations of random finite set filters have been proposed to tackle the issue of jointly estimating the number of targets and states. However, errors resulting from linearization are unavoidable. This paper presents a progressive Gaussian implementation of the probability hypothesis density filter, called the PG-PHD filter. The PG-PHD filter employed the progressive Gaussian filter to predict and update instead of the particle flow filter. The proposed algorithm addresses the drawback of Gaussian particle flow filter by using the progressive Gaussian method to migrate particles to the dense regions of the posterior while no need to linear the measurement function. The simulation results show that the performance of proposed PG-PHD improved significantly compared with the particle flow PHD filter.