摘要
针对概率假设密度(PHD)滤波使用聚类方法提取目标状态时,会出现结果不准确,且PHD滤波无法给出状态到航迹关联的问题,提出一种在目标状态中加入标签的方法来实现状态到航迹的关联.该方法对权值较大的标签,通过两次确认来剔除杂波干扰,得到比传统PHD滤波更精确的状态估计.在提取目标状态时,只对相同标签的粒子进行处理,避免使用聚类方法.通过与传统PHD算法的仿真对比表明,改进算法具有较好的跟踪性能.
To investigate the problem of poor result when the probability hypothesis density（PHD） filter uses clustering technique to extract the target states and the PHD filter keeps no track association, an improved method of the PHD filter is proposed, which inserts a tracking label in the target state. The improved method confirms the label with biggish weight two times to eliminate the influence of clutter, which provides more exact target states than the standard PHD filter. In the states extract step, the improved method only deals with the particle with the same label to avoid using clustering technique. Simulations are presented to compare the performance of the improved method with that of the standard PHD filter. The results show the better tracking performance of the improved method.
出处
《控制与决策》
EI
CSCD
北大核心
2011年第9期1367-1372,共6页
Control and Decision
关键词
随机有限集统计理论
多目标跟踪
概率假设密度滤波
粒子滤波
数据关联
finite set statistics theory
multi-target tracking
probability hypothesis density filter
particle filter
data association
