Researching On The Key Technologies Of Tongue Segmentation And Separation Of The Substance And The Coating Of Tongue In Traditional Chinese Medicine

Traditional tongue diagnosis has inevitable limitations that impede its medical applications. Firstly, the clinical competence of tongue diagnosis is determined by the experience and knowledge of the practitioners. Secondly, tongue diagnosis is usually based on the detailed visual discrimination. Therefore, it depends on the subjective analysis of the examiners, and the diagnostic results may be unreliable and inconsistent. Therefore, it is necessary to build an objective and quantitative diagnostic standard for tongue diagnosis, and explore the relations between features and diseases. And automatic segmentation of the body of tongue and separation of the substance and the coating of tongue are basis work for this. In this dissertation, two key technologies of image analysis for the characterization of tongue diagnosis in Traditional Chinese Medicine (TCM) are studied.The correct segmentation of the body of tongue is premise to the information on diagnosis by the feature of tongue in TCM. Through researching and summering the other researchers'algorithms, Where is difficult for the segmentation of the body of tongue is concluded. And an improved algorithm is advanced in the paper. To the first tongue image database coming from Shanghai University of Medicine, an improved algorithm is advanced to make the segmentation better, and to the second tongue image database coming from Xiamen University which is acquired by different environment, the new segmentation algorithm is advanced. From the experiment, two important views are concluded: firstly, the HSV color space is important for the segmentation; secondly, the Geometric Model is advanced to correct the edges influenced by lips and that of coming from the similar Hue between tongue and face. And our algorithm has higher accuracy. The accuracy for the second tongue image database is about 95%.Physicians mainly inspect the substance of tongue and the coating of tongue when doing tongue diagnosis for examiners; the segmentation of them is basis of the tongue diagnosis in computerized tongue system. In the paper, a new algorithm is advanced. The L*a*b* color space is used. The vector a* and b*, without L vector which is influenced by light very much, are clustered to two classes by K-Means algorithm. Finally, we compare the algorithm with the Split-Combining Algorithm to show that the algorithm has more fitness and higher correctness.Finally, the main content of this dissertation is summarized and the related research in the future is prospected.