基于随机集理论的模糊观测多目标跟踪方法

Multi-target tracking with ambiguous measurements based on random set theory

为解决传感器观测数据具有不确定性和模糊性的多目标跟踪问题,首先给出了模糊观测的随机子集表示及其似然函数构造方法:然后利用所构造的似然函数,并结合概率假设密度(PHD)滤波器来实现模糊观测的多目标跟踪.仿真结果显示,标准PHD滤波器在模糊观测下会出现目标数目估计不准确的问题.针对这一问题,在分析了该问题产生原因的基础上,通过改进PHD滤波器的更新过程,提出了一种单量测独立更新的PHD滤波方法.仿真结果表明,在模糊观测下,改进算法能得到比标准PHD滤波方法更准确的目标数目估计和更高的跟踪精度.

In order to deal with the problem of multi-target tracking with vagueness and ambiguous measurements, firstly, this paper discusses how to model ambiguous measurements as a random subset and construct its likelihood function. Then, the paper uses probability hypothesis density （PHD） particle filter to deal with multi-target tracking with ambiguous likelihood function. Simulation results show that the standard PHD filter provides poor estimate result of target number when using ambiguous measurements. After investigating the causes of the problem, the paper proposes an improved PHD filter, which uses each measurement to update particles. Simulation results show that the proposed method can enhance target number estimate and tracking accuracy.