A novel approach for skin color detection in dynamically changing Illumination

This paper presents a novel approach for real-time skin detection. The proposed approach enables reliable skin detection in dynamically changing illumination and environmental conditions. Histograms are used to model the color distributions in the approach, and they are estimated and dynamically updated by Kalman filters (KFs) over time. In addition, the illumination condition is estimated during the tracking period, and with a fuzzy expert system, this approach is able to detect and correct the errors itself. Furthermore, the performance of this new approach is compared to those of a static color model [1]. The experimental results show that the skin classification rate increases significantly and becomes much more stable, when the approach is used.