摘要
针对群目标跟踪中的数据关联问题,本文提出一种应用超图匹配的多假设群目标跟踪方法。首先将每个群作为一个整体进行跟踪,通过引入延迟决策,利用延迟时间内产生的群航迹假设树,对群可能发生的分离与融合行为进行判断,实现对群整体的跟踪。接着考虑群内各目标通常在运动过程中将保持相对稳定的位置关系,应用超图匹配算法,由航迹与量测之间的相对位置信息辅助完成近距离群内目标的数据关联。仿真表明该多假设跟踪方法能够有效地对群结构进行估计。同时通过引入群内个体目标的相对位置信息,应用超图匹配算法能够获得更好的群内个体目标数据关联效果。
A multiple hypothesis group target tracking algorithm using hypergraph matching is proposed in this paper to achieve efficient data association in group target tracking. Firstly, the observations are clustered into different groups and each group is tracked as a whole to get the group estimation. By introducing the delay decision, the group splitting and merging behaviors are detected based on the group track hypothesis trees generating in the delay period. Then, based on the fact that the targets belong to a group are usually keep stable relative position, hypergraph matching algorithm is applied to make use of these relative spatial information to aid the data association of closely spaced targets in a group. The simulation results prove that the multiple hypothesis group target tracking algorithm proposed in this paper can obtain better estimation of the groups＇ states. Besides, the hypergraph matching algorithm also achieves better performance of data association among the targets in a group, because of the extra spatial information.
出处
《信号处理》
CSCD
北大核心
2017年第11期1497-1504,共8页
Journal of Signal Processing
基金
国家自然科学基金(61471019)
航空科学基金(20152051017)
国家留学基金(201606020013)
关键词
群目标跟踪
多假设跟踪
超图匹配
群分离与融合
相对位置信息
group target tracking
multiple hypothesis tracking
hypergraph matching
group splitting and merging
relative spatial information
