摘要
针对粒子PHD滤波中最优采样分布解析式获取困难及聚类算法提取目标状态导致的滤波性能下降问题,论文提出了一种免聚类的最优粒子PHD滤波算法。论文研究发现,前一时刻的粒子和最新观测集中的某个观测存在最大关联,而与其它观测关联度很小,于是可以将最优采样分布近似为只与单个观测相关的形式,将系统的观测方程线性化,便可以得到最优采样分布的近似解析形式;由于粒子和观测的这种关联,使粒子具有了类别信息,不需要聚类算法提取目标状态。实验表明:该文提出的免聚类最优粒子PHD滤波算法的跟踪性能优于传统的粒子PHD滤波算法。
Analytical expression of the optimal sampling density for the particle PHD filter is difficult to obtain and the filter perform- ance degradation is caused by clustering to extract the targets state. A free-clustering optimal particle PHD filter algorithm is proposed in this paper. The study found that, the particles at previous time step in general has a great relationship with a measurement in the new measure- ment sets, while has little relationship with the rest measurements, so the optimal sampling density can approximate to the form related to a single measurement. Through the linearization of the measurement equation, the analytical form of the optimal sampling density can be got- ten. For the association between particles and measurements, the particle naturally has the class information, the clustering algorithm to ex- tract the target state is no longer needed. The experimental results show that the tracking performance of the free-cluster optimal particle PHD filter is superior to the traditional particle PHD filter.
出处
《计算机与数字工程》
2013年第11期1747-1750,1787,共5页
Computer & Digital Engineering
关键词
粒子PHD滤波
多目标跟踪
最优采样分布
免聚类
随机有限集
particle PHD filter, multi-target tracking, optimal sampling density, free-clustering, random finite sets