Chinese Journal of Computers   Full Text
Title	Method of Face Locating and Tracking Based on Chromatic Coordinates Transformation of Color Images
Authors	YAO Hong-Xun LIU Ming-Bao GAO Wen FAN Xu-Tong ZHANG Hong-Ming LU Ya-Juan
Address	(Department of Computer Science and Engineering, Harbin Institute of Technology, Harbin 150001)
Year	2000
Issue	No.2(158-165)
Abstract &   Background	On basis of color factors of images, the relationship between chroma and color coordinates is obtained, and a method of color coordinates transformation applicable to skin-chromatic, lip-chromatic enhancement is established in this paper. Based on the coordinates, an approach to human face locating and tracking in color images is presented by simply giving a threshold. The method is effective and robust in detecting the positions of objects, not impacted by either the pose of objects or the background. The influence of illumination variance and camera device difference is overcome by on-line learning. After the intensity of color images is normalized, the remained relatively stable chromatic information is utilized to detect, locate and track the objects. The intensity information of color images is employed to enhance the contrast between tiny-difference objects, thus to give prominence to the particulars and describe them. keywords chroma, intensity of light, normalization, principal component analysis, color space representation