Lip color classification based on support vector machine and histogram

According to observe diagnostics theory of Traditional Chinese Medicine (TCM), the lip color of one person is considered as symptoms and signs to diagnose whether his spleen or stomach is healthy or not. In traditional diagnostics method, the lip color is recognized by TCM master or veteran practitioner. The diagnostic result is affected not only by the doctor's knowledge and experience, but also by the light, temperature and other environmental condition. Developing the new objective method to diagnose the lip color becomes urgent and important. This paper describes an automatic, low-cost method based on support vector machine and histogram for classifying lip color, independent of lighting and imaging device characteristics, using consumer digital cameras in a close chamber with stable light source. In HSI color space, the Hue, Saturation, Intensity channel were divided into different number of bins according to probability distribution of all pixels. Histogram of three channels are computed and combined together as feature vector for classifying lip color. A set of lip images with assigned color labels including five classes by TCM master practitioner was used to train multi-classes SVM classifier. The accuracy rate of classification on testing set is close to 85%. Experimental results show that the proposed method is effective and available for the lip color classification.