Abstract: Because of the lack of adequate characterization in existing models for the distribution of skin-tone,algorithm was proposed to detect skin-tone based on a look-up table storing parameters. This method takes skin-tone and non skin-tone as two types of mode. I3y means of statistics in YCbCr color space, we can obtain the probability distribution of different samples with same chroma along lum coordinate,then make classification using I3ayesian discriminant rules,among which look-up table is used to store the model parameters to achieve rapid search and calculation. During experiment,we evaluated the detection performance of algorithm from two aspects of the pixel samples classification and image segmentation, result shows our method has high detection rate and strong robustness.

Key words: Skin modeling, Look-up table, Chroma, Luminance,Bayes classification