A hybrid approach to face detection under unconstrained environments

To detect faces in natural and unconstrained environments, we propose an approach which combines the advantages of both color and gray scale based methods. The idea consists of first preprocessing the images using a state-of-the-art approach for skin modeling in order to determine the potential skin regions. Thus, a scanning of the whole image when searching for faces is avoided. Then, in contrast to the existing methods, we consider the fact that the skin detection step still may produce unsatisfactory results or even fail and therefore we apply an exhaustive search in and around the detected skin regions using a new gray scale based approach. The experimental results show that the proposed approach inherits the speed from the color based methods and the efficiency from the gray scale based ones.