篇名 | A Multi-target Tracking Algorithm for Pseudo Missed-detection Based on the PHD Filter |
---|---|
卷期 | 31:3 |
作者 | Li Gao 、 Huan-Qing Zhang |
頁次 | 072-085 |
關鍵字 | multi-target tracking 、 probability hypothesis density filter 、 pseudo missed-detection 、 random finite sets 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 202006 |
DOI | 10.3966/199115992020063103006 |
As the promising and efficient approximation of the Bayesian filter for multi-target tracking, the probability hypothesis density (PHD) filter iteratively propagates the first-order statistical moment of the multi-target states other than the multi-target density. However, the PHD filter cannot cope with the pseudo missed-detection problem caused by the improper position distribution of target-originated measurements in scenarios. To address the problem, we propose a multi-target filtering algorithm by integrating a missed-detection renovation scheme and an improved component fusion scheme into the PHD filter. Specifically, the PHD filter is used to estimate time-varying number of targets and their states, and the missed-detection renovation scheme is used to redistribute the PHD of the pseudo missed detections from the multi-target posterior PHD. In addition, the improved component fusion scheme is used to reduce and optimize the components of targets in multi-target posterior PHD. Experiment results demonstrate that the proposed algorithm can achieve better estimation accuracy and reliability in possible pseudo missed-detection tracking scenarios when compared against the related existing multi-target PHD filters.