基于IEK-PF的多传感器序贯融合跟踪

Based on Iterated Extend Kalman Particle Filter

针对粒子滤波中得到优化的重要性密度函数比较困难的问题,将迭代扩展卡尔曼滤波和序贯融合与粒子滤波相结合,应于雷达和红外多传感器目标融合跟踪。利用基于迭代扩展卡尔曼滤波的序贯融合算法得到的系统状态更新矩阵和误差协方差矩阵来构造粒子滤波的重要性密度函数,使重要性密度函数能够融入最新观测信息的同时,更加符合真实状态的后验概率分布。仿真结果表明基于序贯融合的迭代扩展卡尔曼粒子滤波(IEK-PF)能提高状态估计的精度。

A technique for fusing data from radar/infrared Multi-sensor was developed to track maneuvering target. Modified Iterated Extend Kalman Particle Filter is simple yet very effective in accounting for the measurement nonlinearities. The idea of fusion is to combine IEK-PF with pseudo sequential filter to obtain optimum state estimates. The main idea uses the system state transition matrix and the error covariance matrix which are gained from the IEKE and the sequential fusion to construct the importance density function of the particle filter. So the importance density function can integrate the latest observation into system state transition density, and the proposal distribution can approximate the posterior distribution reasonably well. The simulation results show that this technique can overcome the flaw that it is hard to get the optimization importance density function in the particle filter and significantly improve the accuracy of state estimation.