Structure identification and tracking of multiple resolvable group targets with circular formation

This paper addresses the structure identification and tracking problem for multiple resolvable group targets, which includes multiple subgroups in a big group. Each subgroup moves in a circular formation in the normal case (the formation may change when the task changes suddenly), and we focus on target tracking, group division, structure detection, parameter identification as well as formation constrained estimation. To this end, the labeled multi-Bernoulli (LMB) filter is used to get the individual tracks of the group targets firstly. Then the Kullback-Leibler (KL) divergence is introduced to measure the similarity between the probability distribution functions (pdf) of two targets and the K-mediods clustering method is used to obtain the number and division of the subgroups. Next, the structure detection of each subgroup is performed to confirm whether the subgroup structure accords with the priori circular formation. Then the parameter identification and the formation constrained estimation based on the estimation projection (EP) method are only applied to those subgroups which conform to prior formation. Finally, a simulation study on the group targets tracking with circular formation is provided to demonstrate the proposed algorithm.