We introduced a computer-aider tongue examination system, which can reduce the large variation between the diagnosis results of different doctors and quantize the tongue properties automatically in traditional Chinese medicine (TCM) diagnosis. Several key issues and algorithms in this system are discussed, involving some newly proposed image analysis techniques: i) a new tongue color calibration scheme is proposed to overcome the drawbacks of the previous method; ii) a gradient vector flow (GVF) snake based model integrating the chromatic information of the tongue image is used to extract the tongue body; iii) an unsupervised segmentation method of color-texture regions is adopted to quantitatively analyze the distribution of the substance and the coat on the tongue surface. The experimental results show that the presented system has a great deal of potential for the tongue diagnosis.
展开▼
