采用粒子滤波和模糊聚类法的非线性多目标跟踪

提出一种新的非线性多目标跟踪方法,用模糊聚类算法实现数据关联,采用粒子滤波实现对各目标的独立跟踪.首先利用最大熵模糊聚类对目标和观测数据进行关联,采用模糊隶属度重建多目标滤波中的联合关联概率矩阵.然后利用粒子滤波适于处理非线性问题的特点,通过联合关联信息,采用粒子滤波独立对各目标进行滤波,实现对目标状态的更新.最后,将该算法应用于多传感器多目标纯方位角跟踪.仿真结果表明,相比于联合概率数据关联算法及MEF-JPDAF,新算法具有更高的跟踪精度.

Passive multi-target tracking based on independent particle filtering and fuzzy clustering

A novel method based on fuzzy clustering and independent particle filtering is proposed for nonlinear multi-target tracking. Firstly, the association of target with measurement is carried out by the use of the maximum entropy fuzzy clustering. Then the joint association probability matrix is reconstructed by utilizing the fuzzy membership degree of the target and measurement. Since particle filtering performs well in the nonlinear tracking system, this paper employs it and the joint association innovations to update each target state independently. Finally, the proposed method is applied to multi-sensor multi-target bearings-only tracking. Simulation results show that the method can obtain a higher tracking precision than JPDAF and MEF-JPDAF.