Face and hands segmentation in color images and initial matching with a biomechanical model

In this paper we describe a robust and efficient procedure to detect skin region with homogeneous color values in monocular indoor images. The mathematical background is a functional segmentation process adapted to color images. The idea is to obtain a system to detect the hands and face in a sequence of monocular or binocular images. The main criteria are to select the best probabilistic skin regions in the previous segmentation process. In particular we are focuses in hands and faces because are visible in any avatar in general. We also use a biomechanical restriction to reach this initial estimation, so the system is to be able to detect hand-face-hand configuration to be used as an initialization procedure to human detection in a more general human recognition system. Several analytical rules are applied to reach these specific configurations. We introduce the segmentation process and region classification criteria (hands, face, head and upper-torso). Finally, we present some results over a significant statistical number of potential users. Also we implement the algorithm efficiently in order to obtain real time results processing standard video format