Real-time hand gesture recognition for augmented screen using average background and camshift

This paper proposes a hand tracking and gesture algorithm for HCI (Human Computer Interaction) using real-time image sequence. The hand region is extracted the background subtraction technique and is surrounded by window. For the hand region segmentation the minimum window size was used with a blob labeling. Fingertip is defined as the point from which has the longest distance from the center point to outlines of the hand region using the convex hull. Once fingertip is detected, CAMshift (Continuously Adaptive Mean shift) is started for tracking a hand. We propose a more accurate tracking method to combine background subtraction and CAMshift. Experimental results show that the success rate of finger's grip detection reaches 96.77%.