一种高效粒子滤波检测前跟踪算法的仿真分析  

Analysis on efficient particle filter based track-before-detect

作　　者：梁新华[1] 潘泉[1] 杨峰[1] 黄冬民[1] 

机构地区：[1]西北工业大学自动化学院,西安710072

出　　处：《计算机应用研究》2012年第1期67-71,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61075029;61074179);航空科学基金资助项目(20090853013);西北工业大学基础研究基金资助项目(JC201015)

摘　　要：分析了一种高效粒子滤波检测前跟踪算法的独特结构。其算法中存在两类粒子互相竞争与协作的复杂关系。针对其两类粒子数量选择的困难,通过大量仿真对比分析了两类粒子五种数量比例在不同总粒子数和不同过程噪声水平下的算法性能。仿真结果表明在两类粒子数量相等附近时,算法综合性能较好。This paper investigated the particular architecture of an efficient particle filter based track-before-detect.It was found that there was a complex competitive and cooperative relationship between these two types of particles in this algorithm.Made a lot of simulations to solve the selection problem of the quantitative relationship of these two types＇.Considered five kinds of particle number ratio of these two types＇ in these simulations,which were designed to compare performance using different total particle numbers and process noise levels.Simulation results show when these two numbers is equal or nearly equal,can achieve a good performance.

关 键 词：检测前跟踪 粒子滤波 粒子个数 红外弱目标 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]