Adaptive Skin Color Prediction Using Multi Skin Color Models

In this paper, a new skin color detection method, in which a skin color model for a given image is adequately selected from a set of models to realize adaptive detection, is proposed. In the proposed method, multiple skin color models are tuned by a learning based on a concept of self-organizing adaptive controller. The skin color models for various lighting conditions can be obtained from small number of images. The effectiveness of the proposed method is verified by simulation results.