复杂背景下目标模板更新算法研究  

Study of Target Template-Update Algorithm in Complex Background

作　　者：应陈克[1] 胡文江[1] 赵亦工[1] 

机构地区：[1]西安电子科技大学模式识别与智能控制研究所,西安710071

出　　处：《遥测遥控》2006年第6期34-39,共6页Journal of Telemetry,Tracking and Command

摘　　要：在匹配跟踪算法中,模板更新策略是跟踪成败的关键。从复杂背景下的目标跟踪出发,指出固定模板和逐帧模板更新的不足之处,介绍几种性能较好的模板更新策略,并在此基础上提出改进的模板更新方法。同时,为了更好地克服背景光照带来的影响,对普通最小平均绝对差值函数MAD(M in imum Absolute D ifference)方法进行了改进。仿真实验结果表明,该算法能克服诸如烟尘干扰、背景光照变化、地貌起伏等不利因素,能在复杂背景中稳定和精确地进行目标跟踪。Template update method is the key of the success of the matching and tracking algorithm.The inadequacy of fixed template update and each-frame template update for the target tracking in complex background is presented in this paper,which descripes some good template update strategies and presents an improving template update algorithm.Meanwhile,with the background light changing,we have improved the Minimum Absolute Difference(MAD) algorithm.The result shows that the new algorithm can track target accurately and stably in complex scene,ignoring the smoke and dust disturbing,background light changing and terrain waving.

关 键 词：目标跟踪 模板更新 最小平均绝对差值函数 卡尔曼滤波 

分 类 号：TN953[电子电信—信号与信息处理]