关键词: 手势跟踪, 范例集, 遮挡, HMM, ICAMSHIFT

Abstract: As a choice for tracker representations exemplars have been applied to gesture tracking and recognition. However exemplars based tracker is unable to acquire the accurate abstraction of hand silhouette in complex background, which makes the inaccurate predication of the hand motion，which degrades the gesture Tracking and recognition. This paper proposes a new exemplar based tracker i.e. CEE(CAMSHIFT Embedded Exemplar) tracker to implement the dynamic hand gesture tracking in complex background. In training phase, ICAMSHIFT(Improved CAMSHIFT) algorithm is used for extracting hand silhouette and generating exemplars. At the same time, the dynamic HMM modal is established. In tracking phase, firstly the hand motion is predicted according to the hand feather extracted using ICAMSHIFT algorithm and the HMM modal. Secondly the hand contour is obtained according to the learned exemplars. Experiments results show that the accurate gesture tracking can be acquired in complex background. What’s more, the tracking is effective on the occasion of occlusion.

Key words: Gesture tracking, Exemplars, Occasion, HMM, ICAMSHIFT