Gesture segmentation based on YCb´Cr´ color space ellipse fitting skin color modeling

Color feature is zooming, rotation, affine and deformation good invariance. At present, the color models are established based on extensive statistical information, but not effective for specific imaging equipments. According to this problem, this paper presents a YCb´Cr´ space ellipse fitting under the skin modeling method based on the specific statistical properties of the color distribution, and use it for the gesture segmentation. Experimental results show that the method has good accuracy, eliminating some of the color dependence of the brightness, and has better versatility and environmental adaptability for some specific imaging equipment.