基于随机有限集的多目标跟踪算法综述  

Multi-target Tracking Algorithms Based on Random Finite Set: a Survey

作　　者：吴卫华[1] 江晶[1] 冯讯[1] 刘重阳[1] 秦星[1] 

机构地区：[1]空军预警学院,武汉430019

出　　处：《电光与控制》2016年第3期1-6,共6页Electronics Optics & Control

基　　金：国家自然科学基金(61102168)

摘　　要：基于随机有限集(RFS)的多目标跟踪算法是近年来国际上多传感器信息融合领域的研究热点。分别论述了RFS框架下3种重要的近似滤波器即概率假设密度(PHD)、势化PHD(CPHD)和多目标多伯努利(Me MBer)滤波器的研究进展,并对它们进行了比较和总结;然后,着重梳理了它们在机动目标跟踪、非标准目标跟踪、多传感器融合、多目标跟踪性能评估等方面的研究现状;最后,对相关领域的未来研究方向进行了分析和展望。The multi-target tracking algorithms based on Random Finite Set（ RFS） recently become a research hotspot for multi-sensor information fusion in the world. The research progresses of three kinds of important approximate filters under the RFS framework are discussed and summarized, including Probability Hypothesis Density（ PHD）, cardinalized PHD and multi-target multi-Bernoulli filter, and comparison is made to them. Then, the research status about the following aspects as maneuvering target tracking,nonstandard target tracking, multi-sensor fusion, multi-target tracking performance evaluation, etc. is presented. Finally, the future research directions in related areas are given.

关 键 词：多传感器融合 多目标跟踪 随机有限集 概率假设密度 多目标多伯努利滤波器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]