提出一种人体运动跟踪算法,从无关节标记的单目视频中获取人体运动1 利用一个带外观模板的人体关节

模型,通过学习得到的运动模型及基于外观模型的相似性计算,巧妙地利用粒子滤波的概率密度传播策略鲁棒地跟

踪普通单目视频中的人体运动1 当出现跟踪丢失时,能在后续序列中自动恢复正确跟踪,且能较好地处理遮挡和自

遮挡问题1 实验表明,该算法鲁棒性好,跟踪结果令人满意-　In this paper , a novel approach is proposed for t racking markerless human motion in monocular

videos to capture the articulate motion data1 With an articulated human model const ructed , the new ap2

proach uses the probability density propagation of the particle filters through the learnt motion model and

likelihood computing with the appearance models to t rack the human motion1 The method is capable of auto2

matically recovering f rom t racking failures1 It can also process the occlusion and auto2occlusion problem cor2

rectly1 Experimental result s f rom real monocular videos show that the new approach is robust and the t rack2

ing result s are satisfactory1