一种新的概率数据关联滤波算法

在密集多回波条件下对单个机动目标跟踪的情况下,综合利用"最近邻"法和概率数据关联滤波算法,推出了一种基于"最近邻"方法的概率数据关联滤波算法,采用关联区域内总数固定的候选回波来更新被跟踪目标的状态,并进行了计算机仿真.结果表明,新滤波算法的跟踪性能明显要优于概率数据关联滤波算法,而且降低了概率数据关联滤波算法的计算量.因此,可以通过回波的残差协方差矩阵进行衡量,挑选总数固定的"最近邻"回波,取这些回波的加权和作为目标回波用于概率数据关联滤波算法中更新被跟踪目标的状态.

A New Probabilistic Data Association Filter

Tracking a single target in a cluttered environment, a new probabilistic data association filter (PDAF) is brought forward, which is based on using nearest neighbours approach to improve PDAF. A fixed number of nearest neighbouring measures that is selected by the innovation covariance matrix is used for updating the filter, and the result is given to testify its efficiency through Monte Carlo. The filter can not only get better performance for tracking targets than that of PDAF, but also give an overall lower processing load for the nearest neighbours approach.