An improved GM-PHD tracker with track management for multiple target tracking
Huanqing Zhang, Jinlong Yang, Hongwei Ge, Le Yang · 2015
The probability hypothesis density (PHD) filter is a promising tool for tracking the time-varying number of targets in real time. The Gaussian mixture PHD filter is an analytic solution to the PHD filter for linear Gaussian multi-target models. By using Gaussian component labels in GM-PHD filter, the identities of individual target can be obtained. However, the labeling GM-PHD filter cannot correctly discriminate tracks of individual targets when targets are moving closely to each other. To solve this problem, an improved GM-PHD tracker is proposed, which is able to identify and manage the track of individual target effectively. First, a detection-guided dynamic reweight scheme is employed in the GM-PHD filter to alleviate the weight error of closely spaced targets. Then, a novel track management scheme is introduced to form and maintain the tracks of individual targets. The simulation results demonstrate the better performance of the proposed algorithm in comparison with the labeling GM-PHD filter.