A fuzzy approach to skin color detection

Abstract

Skin detection is a useful tool for many subsequent image processing tasks, such as face detection, face tracking or human-computer interfaces. In many cases, using a skin color detection scheme is related to a proper representation of the color space chosen to interpret image information. In this work, we propose a fuzzy system for detecting skin in the RGB color space, so that each color plane is modeled using fuzzy sets. As a consequence, the development of the membership functions, the inference system and the defuzzifying process are described throughout this paper. The experiments have been performed with the XM2VTS and VALID face databases and the results show that our system has high detection rates and low false positive rates compared to other similar skin color detectors.