A Tracking Method for Dense Targets Within Resolvable Group Based on Collective Feature Correction

This article addresses the multi-target tracking problem for a dense but resolvable group, and designs a method that uses the velocity estimation of the group to correct the initialization of trajectories and optimize the trajectories identification for the sub-targets within the group. First, an extended target tracking algorithm based on the elliptical random hypersurface model (RHM) is adopted to obtain the overall motion state of the group. Second, the overall velocity estimation, i.e., the collective feature of the group, is used as a prior pseudo measurement information to assist in generating the newborn target state more accurately. Next, an adaptive Generalized Labeled MultiBernoulli (GLMB) algorithm is used to estimate the motion states of the dense targets within the group, and the sub-target motion states are modified by integrating the overall velocity estimation of the group. The simulation results verify that the velocity correction algorithm proposed in this paper can significantly improve the tracking performance of the dense group targets, and provide a theoretical guidance for the engineering applications in the group target tracking.