A new human hand-image tracking method

Hand gesture recognition has been widely used in surveillance systems. And human hand image tracking is the foundation of the recognition, and it is also an important step. This paper proposes a new human hand tracking method in video surveillance based on color and shape. In the hand detection step, it proposes a detection method based on chromatic adaptive thresholding and the circular degree. In the hand tracking step, it uses a modified formula of SAD (Sum of Absolute Differences) to implement the matching method in order to track the hand. The results suggest that the method is accurate and simple. Additionally, this method also has lower computational complexity and it can be uses in most scenes.