—A novel human hand tracking algorithm using hand occlusion information based on a single-view camera is put forward. Firstly, the moving hand is tracked by the algorithm of edge tracking. Then the hand features of boundary are used to get the maximum inscribed circle of the palm. According to the circle, the hand is divided into two parts: the palm and fingers. Finally, using the proportion of the palm and fingers in the whole hand to guide particle filter tracking. The experiments show that this algorithm can correctly predict the position of the occluded parts, and has better tracking accuracy than some of the previous work.
展开▼