Hand Tracking Algorithm Based on SuperPixels Feature

There have been considerable research efforts to use the hand as an input device for HCI in recent years. Hand tracking is the most important procedure for HCI, it is essential of tracking stability and efficiency for hand manipulation. This paper proposed a novel hand tracking algorithm which can track a hand stable and is real time, and the proposed algorithm can work on normal CCD cameral. Our algorithm is based on mean-shift and we improved it to fit for robust hand tracking by using super pixel cluster, integrated GIH and skin color mask, the skin color mask was extracted using online learning. The proposed improved algorithm can track hand reliably even in clutter environments comparing to the existing traditional algorithms.