Track Coalescence Avoidance in Multi-target Tracking

Abstract

Multi-target tracking suffers from track coalescence due to closely spaced tracks and so, it avoids coupling between the tracks in clutter and uncertain measurements environment. When the motion of the multi-targets are in close vicinity, their track becomes cross-merged so that the multi-target turn out to be one target. We propose two ideas; one is to refine the target estimates by applying smoothing method, and the other is to ignore the influence of target measurement being tracked in a vicinity of a potential track. The proposed smoothing method uses the linear multi-target based on integrated track splitting filter (sLM-ITS) to avoid joint (common) multi-target measurements association while allowing them as pretended clutters. Hence, sLM-ITS updates a potential track without impact of the other tracks in its vicinity. The proposed method avoids track coales-cence significantly in a difficult multi-target situation. The sLM-ITS method provides improved smoothing as well as false-track discrimination (FTD) capabilities in comparison to the existing algorithms as illustrated in the simulation results.