摘要
针对异步采样下多红外传感器多目标跟踪问题,提出了一种基于概率假设密度粒子滤波的跟踪算法。该算法首先将一个融合周期内所有采样点在融合中心的坐标系中和时钟下进行统一映射,然后按照实际测量值到来的时间先后顺序,根据融合周期内相邻两个时刻之间状态的动态关系,建立相应采样时刻间的状态方程和量测方程,最后根据当前时刻测量对应的传感器的个数选择不同的滤波算法,对顺序到来的观测值依次进行状态估计和更新,从而得到目标数目和相应的状态估计值。仿真实验表明,所提的算法能较好地解决异步采样下多红外传感器多目标跟踪问题,具有较高的跟踪精度和较强的鲁棒性。
A multiple targets tracking algorithm based on probability hypothesis density particle filter is proposed to overcome the exiting problems in asynchronous multiple infrared sensors system. First, all sampling points are unified mapped in the reference frame and clock with fusion center by the algorithm. Then, based on the orderly measured values, a series of new state and measurement equations can be estabished through the dynamic equation of state between adjoining sampling time. Finally, the number of targets and comesponding state estimate can be obtained by selecting the filter algorithm based on the number of the sensors. Simulation results show that the presented algorithm can solve the problems in asynchronous multiple infrared sensor system effectively, and has high tracking accuracy and strong robustness.
出处
《测控技术》
CSCD
2015年第5期153-156,共4页
Measurement & Control Technology
基金
江西省高等学校省级教学改革项目(JXGJ-14-45-3)
关键词
红外传感器
概率假设密度
粒子滤波
异步采样
infrared sensor
probability hypothesis density
particle filter
asynchronous sampling