杂波干扰环境下的变分贝叶斯目标跟踪算法

Variational Bayesian Target Tracking Algorithm in Clutter Interference Environment

为了提高杂波干扰环境下的目标跟踪精度,将概率数据关联与变分贝叶斯理论相结合,提出了一种变分贝叶斯概率数据关联卡尔曼滤波(PDA-VB-KF)跟踪算法。算法首先采用概率数据关联(PDA)技术获取目标的有效量测,随后基于伽马分布设计相应的自适应因子以修正有效量测的噪声协方差。最后在变分贝叶斯(VB)框架下,实现自适应因子和目标状态信息的联合估计。为验证提出算法的跟踪性能,通过两种仿真案例对其进行测试。仿真结果表明,相比于传统的最近邻卡尔曼滤波(NN-KF)算法,k最近邻卡尔曼滤波(KNN-KF)算法和概率数据关联卡尔曼滤波(PDA-KF)算法,提出的算法具有更高的跟踪精度。

To improve the target tracking accuracy in clutter,this paper combines probabilistic data association with variational Bayesian and proposes the variational Bayesian probabilistic data association Kalman filter(PDA-VB-KF)tracking algorithm.Firstly,the effective measurement of the target is obtained by using probability data associa-tion(PDA),and then the adaptive factor is designed based on gamma distribution to correct the effective measurement noise covariance.Finally,the joint estimation of adaptive factor and state is realized under the framework of variational Bayesian(VB).In order to verify the tracking performance of proposed filter,this paper tests it in two simulation ca-ses.The simulation results show that the proposed filter has higher tracking accuracy than the nearest neighbor Kal-man filter(NN-KF),the k-nearest neighbor Kalman filter(KNN-KF)and the probability data association Kalman filter(PDA-KF).