Hand tracking and gesture gecogniton by anisotropic kernel mean shift

Mean shift algorithm is an iterative procedure that shifts each data point to the average of data points in its neighborhood. It been applied to object tracking. But traditional mean shift tracker by isotropic kernel often loses the object with the changing structure of object in video sequences, especially when object structure varies fast. This paper proposes a non-rigid object tracker with anisotropic kernel mean shift in which the shape, scale, and orientation of the kernels adapt to the changing object structure. The proposed tracker is used for hand tracking in video. Gesture recognition is implemented simultaneously with orientation histograms. Experimental results show that the new algorithm ensures the robust and real-time hand tracking and and accurate gesture recognition.