基于分块EnKF的非线性目标跟踪算法

针对滤波航迹的相关性以及初始状态的选择会对跟踪性能产生影响的问题,将集合卡尔曼滤波算法引入到非线性目标跟踪领域,验证了其可行性和有效性,提出了基于分块集合卡尔曼滤波的非线性目标跟踪算法.采用分块思想生成初始集合,使用协方差矩阵加权方法解决分块间的航迹相关问题.仿真结果表明基于分块集合卡尔曼滤波的目标跟踪算法计算复杂度和以往的集合卡尔曼滤波算法同阶的情况下可以提供更高的运动参数估计精度,解决了粒子滤波算法计算量大难以进行实时跟踪的问题. 

Nonlinear Target Tracking Algorithm Based on Block Ensemble Kalman Filter

As target tracking performance depends on correlated tracks and the selection of filter initial states, ensemble Kalman filter was introduced to nonlinear target tracking system, and the feasibility and validity of the system were verified. A new target tracking algorithm based on block ensemble Kalman filter was proposed, where initial ensemble was produced by the block method, and covariance matrix weighting was used for all the blocks in the target tracking process. The simulation results show that the algorithm based on block ensemble Kalman filter has the same computational complexity as previous ensemble Kalman filter while offers higher estimation accuracy for motion parameters, and can fulfill real-time tracking in contrast to high computational complexity of particle filter.