Adaptive Detection of Human Skin in Color Images

Detection of human skin has several practical applications in image and video processing fields, such as face detection, image indexing, gesture recognition, and nudity detection. The purpose of a human skin detection method is to distinguish image portions between skin and non-skin regions. Many existing approaches in the literature are based on features that explore information of color, shape and texture in the images. Furthermore, several skin detection methods do not produce satisfactory results since they classify image regions based on fixed thresholds and are sensitive to illumination conditions. This paper describes a method for adaptive detection of human skin in images based on a normalized lookup table. The resulting probability map is used to detect skin and non-skin regions, which are refined through texture descriptors. Experimental results show the effectiveness of the proposed method.