基于两级采样的粒子滤波检测前跟踪算法  

Particle filter track-before-detect algorithm based on two-hierarchy sampling

作　　者：梁新华[1] 潘泉[1] 杨峰[1] 王增福[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《系统工程与电子技术》2011年第9期1921-1926,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(60634030;60702066);航空科学基金(20090853013);西北工业大学基础研究基金(JC201015);西北工业大学"翱翔之星"计划资助课题

摘　　要：针对粒子滤波检测前跟踪算法中高维状态采样效率低的问题,提出了一种基于两级采样的粒子滤波检测前跟踪算法。根据状态向量中各分量对量测有无直接影响,把粒子的高维状态采样转换为两个序列低维采样,提高了粒子采样效率,进而可获得更高的检测概率和估计精度。仿真结果验证了算法的有效性。A particle filter track-before-detect （TBD） algorithm based on two-hierarchy sampling is proposed to deal with low sampling efficiency in high dimension state space. According to each component＇s impacting on measurement directly or not, the higher-dimension state sampling process is converted to two successive lower dimension state sampling processes. This algorithm improves the efficiency of state sampling for particles, and a better detection and tracking performance is obtained. Simulation results show the effectiveness of the proposed algorithm.

关 键 词：检测前跟踪 粒子滤波 两级采样 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]