Skin Region Segmentation based on the Average Preprocessed Image of Multicolor Face Image Sequence

Abstract

For face images with unbalance color and light, this paper proposed a method of skin region segmentation based on preprocessing of multicolor face image sequence. In this method, the multicolor face images of the sequence were preprocessed to remove color offset and uneven brightness. Then the color informations of preprocessed face images were jointly used for taking the average to remove noise and smooth the face image. Finally, for the average preprocessed face image, based on the established ellipse skin model, the skin region segmentation was realized. Experiments show that, this method had good performance.