基于粒子滤波器和风险决策跟踪遮挡目标的方法  

Tracking Occlusive Target Method Based on Bayesian Decision Theory and Particle Filter

作　　者：周妍[1] 胡波[1] 张建秋[1] 

机构地区：[1]复旦大学电子工程系,上海200437

出　　处：《电子学报》2007年第2期350-353,共4页Acta Electronica Sinica

基　　金：国家973重点基础研究发展规划(No.2006CB705700)

摘　　要：为了跟踪遮挡目标,本文通过赋予外界遮挡和目标自身外观变化不同的风险度,利用风险决策的方法判断遮挡,设计了新的目标模板和模板更新规则,使目标模板既能随外观的缓慢变化自动更新,又能在全遮挡出现时维持模板不变.当这个新的目标模板和模板更新规则结合粒子滤波器的跟踪算法时,可实现遮挡目标的跟踪.实际视频流的测试结果表明,本文提出的方法可以有效判断遮挡,实现了长时间全遮挡情况下的目标跟踪.In order to track an occluded target in an image sequence,the Bayesian decision theory is,here,introduced to the problem of distinguishing occlusions and appearance changes according to their different risk possibilities. A new target template combining image intensity and histogram is designed. The ＂corresponding updating method is also derived based on particle filter. If the target is totally occluded by another target, the template can be kept unchanged. The occlusion of a target will not influence tracking. Simulation results show that the presented method can efficiently justify whether the occlusion occurs and realize target tracking in image sequences even though the tracked target is totally occluded with long time.

关 键 词：目标跟踪 粒子滤波器 风险决策 目标遮挡 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]