自适应粒子滤波在人体跟踪过程中的应用

人体运动跟踪技术近年来在图像处理与计算机视觉领域引起很多关注,在当前一些重要研究和应用领域有着广泛的需求。在以往跟踪方法的基础上提出了基于决策规则的自适应粒子滤波的无标记运动目标跟踪方法。利用一个带外观模板的人体关节模型,通过学习得到运动模型及基于关节模型的相似性计算,巧妙地利用自适应粒子滤波对运动目标进行实时跟踪,使得在粒子滤波过程中,可以根据实际滤波情况在线调节粒子数。实验表明,提出的算法鲁棒性好,跟踪速度比基于传统粒子滤波的快。

Markerless Human Motion Tracking Based on Adaptive Particle Filtering

In this paper,an adaptive particle filtering algorithm based on decision-making rule is proposed for tracking non-mark-ing human motion.With an articulated human model constructed,the new approach uses the adaptive particle filtering algo-rithm through the learnt motion model and likelihood computing with the appearance models to track the human motion and achieve the goal of movement of real-time tracking,In the process of particle filtering,according to the actual situation particle count can be adjusted on-line particle filter.Experimental results from real monocular videos show that the new ap-proach is robust and the tracking results are better than traditional particle filter.