一种改进的多关节目标跟踪算法

Improved multi-articulated object tracking algorithm

为提高多关节物体的跟踪效率和精确度,提出了一种目标跟踪算法。首先建立多关节物体的团结构模型,模型的建立结合团势函数等有效降低了传统图模型的冗余度,并结合目标物理的深度、边缘、颜色等信息,利用K-means、Mean-shift算法、粒子滤波实现目标跟踪。实验结果表明,该推理算法能够提高跟踪效率,有较强的鲁棒性,能够满足多关节目标的跟踪要求。

In order to improve the tracking efficiency and precise of multi-articulated object tracking,this paper proposed an algorithm of object tracking.Firstly constructed a graphical model that employed clique potentials which to denote multi-articulated object,since it made use of clique potential function etc.To reduce the complexity of graphical model such as traditional graphical model,and it also combined some information such as depth,edges,colors.Then it ultilized the K-means,Mean-shift algorithm and particle filtering to track mobile.Experiment results show that the proposed algorithm maintains the high precision and the ability to control the occlusion,and it can meet requirements of mutil-articulated object tracking.