目标跟踪状态估计排序粒子群算法研究  

Ground target tracking state estimation using sort-based particle filter algorithm

作　　者：刘华伟[1] 李湘清[1] 谢文俊[1] 

机构地区：[1]空军工程大学工程学院,西安710038

出　　处：《计算机工程与应用》2012年第15期21-23,28,共4页Computer Engineering and Applications

基　　金：航空科学基金(No.20080896009)

摘　　要：基于视觉的无人机地面目标跟踪状态估计为非线性滤波问题,针对使用一般粒子滤波算法存在粒子退化和计算量大的缺陷问题,提出了一种基于排序的粒子滤波算法,对粒子依误差大小进行排序并计算粒子权重。仿真试验表明,该方法减轻了粒子贫化的影响,提高了状态估计精度。Vision-based ground target tracking state estimation is a non-linear filter problem. For particle degeneracy and large amount of computation phenomenon exist in the standard particle filter algorithm, a sort-based particle filter algorithm is proposed. In this algorithm, all the particles are sorted by measurement update error during weighting update process and the weight of the particles is calculated based on sequence. Compared to standard particle filter, the sort-based particle filter has lightened particle degeneracy, reduced the amount of computation and improved the precision.

关 键 词：目标跟踪状态估计 粒子滤波算法 排序 粒子权重 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]