Hand tracking and motion detection from the sequence of stereo color image frames

This paper presents an approach for hand tracking which is based on color and motion. In particular, the YCbCr color system and the difference of the two consecutive frames of images are considered for detection of a moving hand. Since the luminance component is concentrated in Y, while the chrominance components are in Cb and Cr, the image region in which the luminance varies significantly is able to be effectively selected by the chrominance components only. All the objects and background that are not in motion are removed from the image by the difference operation. The head and the hand that may slightly move are also narrowed down by the difference operation. Thus, the region of the moving hand is extracted from a composite image. This approach is useful in detecting moving object of narrow banded color. The locations of the centers of hands or objects are obtained in a sequence of stereo images. The movement of a hand or an object is estimated with an equation that operates the set of locations obtained. A complete tracking of a hand or an object is described and presented.