Multi-target Track Extraction Method Based on Gaussian Mixture Probability Hypothesis Density Filter

Gaussian Mixture Probability Hypothesis Density (GM-PHD) filter can estimate the number and state of targets simultaneously, which is an effective multi-target tracking method. However, the identification of each target is needed to construct target track, and the filter cannot directly provide such information. In this paper, an improved multi-target tracking algorithm is proposed, which assigns a unique label value to each Gaussian component, and gives association between the states of each target over time. At the same time, to solve the problem of performance degradation under high clutter concentration, the update weight of Gaussian component is re-corrected in this paper, and finally the only observation corresponding to each target is extracted to ensure the continuity of track. Simulation results show the effectiveness of the proposed method.