基于多次卡尔曼滤波的目标自适应跟踪算法与仿真分析  

Adaptive Algorithm Based on Multi-Kalman Filter for Target Tracking and Simulation Analyses

作　　者：王向华[1] 覃征[1] 杨新宇[2] 杨慧杰[2] 

机构地区：[1]清华大学计算机科学与技术系,北京100084 [2]西安交通大学计算机科学与技术系,西安710049

出　　处：《系统仿真学报》2008年第23期6458-6460,6465,共4页Journal of System Simulation

摘　　要：标准的卡尔曼滤波算法由于使用了固定单一的状态噪声模型,因此当目标运动状态经常发生剧烈变化时,跟踪效果不是很理想。为了提高对目标的跟踪精度和跟踪收敛速度,提出了一种新的算法,通过多次步长不同的卡尔曼滤波算法来判断机动目标的运动状态,进而使得系统状态噪声协方差能够随着目标机动情况自适应调整。最后的蒙特卡罗仿真实验验证了此算法的有效性。Because the state noise model is fixed and simplex in kalman filter, the result of tracking is unsatisfactory while the moving state of target changing acutely. In order to improve the tracking precision and convergence rate, a novel algorithm was proposed, multi-kalman filter with different sample time were used to judge the target moving state, and then adjust the state noise covariance adaptively. From the simulation, it can be seen that the algorithm proposed can improve the result of tracking.

关 键 词：目标跟踪 卡尔曼滤波 收敛性 自适应 仿真分析 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]