PHD Filter Based Traffic Target Tracking Framework with FMCW Radar

Aiming at tracking traffic target based on frequency modulated continuous wave (FMCW) radar in traffic scenes, this paper proposes a robust Gaussian mixture probability hypothesis density (GM-PHD) filter combined with measurement processing. Due to the different recognition requirements of moving and static targets in this application, the raw measurements are separated into moving and static parts. To deal with the problem of extended target, a density-based spatial clustering of applications with noise (DBSCAN) algorithm is used to cluster the moving measurements and then extract the feature measurements. After the above two steps, the number of measurements input to PHD filter is reduced, which ensures the real-time performance and accuracy of the algorithm. In order to solve the problems of the PHD filter with track information in traffic scenes, this paper proposes some improvements, including target birth intensity generation, track merging and extraction, and target weight correction. The experimental data is used to compare the proposed algorithm with the conventional algorithm, which proves the advantages of the proposed algorithm.