Improved Cardinalized Probability Hypothesis Density Filter for Multitarget Tracking Using Formation Priors

Effective observation of group targets (bird flocks, etc.) has significant importance in aviation safety and ecological research. Closely spaced group targets with swarm intelligence can form huge and complex linear formations. Tracking radar must feature a narrow beam and high bandwidth to discriminate different targets within the formation. However, there are two major problems in formation target tracking. First, the number of targets within the beam may fluctuate with the change of radar-target geometry; second, false measurements may be detected from the distributed Doppler spectra. This paper proposes a cardinalized probability hypothesis density (CPHD) filter for multitarget tracking using the formation priors form a scanning radar. The model of false measurements is optimized based on the experimental data of birds. The densities of new targets and the probability of target death are set according to the formation priors of group structure and radar-target geometry. Finally, the effectiveness of the proposed method is verified by experimental results.