Random finite sets and sequential Monte Carlo methods in multi-target tracking

被引：25

作者：

Vo, BN ^{[1
]}

Singh, S ^{[1
]}

Doucet, A ^{[1
]}

机构：

[1] Univ Melbourne, Dept Elect & Elect Engn, Melbourne, Vic 3010, Australia

来源：

2003 PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON RADAR | 2003年

关键词：

D O I：

10.1109/RADAR.2003.1278790

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

Random finite set provides a rigorous foundation for optimat Bayes multi-target filtering. The major hurdle faced in Bayes multi-target filtering is the inherent computational intractability. Even the Probability Hypothesis Density (PHD) filter, which propagates only the first moment (or PHD) instead of the full multi-target posterior, still involves multiple integrals with no closed forms. In this paper, we highlight the relationship between Radon-Nikodym derivative and set derivative of random finite sets that enables a Sequential Monte Carlo (SMC) implementation of the optimal multitarget filter. In addition, a generalised SMC method to implement the PHD filter is also presented. The SMC PHD filter has an attractive feature-its computational complexity is independent of the (time-varying) number of targets.

引用

页码：486 / 491

页数：6