Face Detection and Face Direction Estimation Using Color and Shape Features

Abstract

This paper describes the detection of faces and the estimation of the face direction from an image where their sizes, positions and poses are unknown. As the candidate regions of the face, the region whose upper part is hair color and lower part is skin color was first extracted. To make sure that there is a face actually in the face candidate region, facial features, pupils, a nostril and a mouth, using color, shape and the geometrical conditions of the facial features were extracted next. After the face detection, the face direction is estimated on the basis of the position of the pupils and face contour. It is seen that the effectiveness of this method by experimental results. This paper has demonstrated the new method of the detection of the face whose size, position and pose are unknown and the estimation of the face direction. Future face detection studies would be done without any facial assumptions.