Research On Methods Of Image Processing In The Qualification Of Tongue Diagnosis

Tongue diagnosis is the "Four Diagnosis Methods" in Traditional Chinese Medicine (TCM). And it plays an important role in the clinical diagnosis procedure. It has many advantages over the Western Medicine (WM), such as painless, noninvasive, simple yet efficient, low cost and has no side effect. It is believed that the tongue has a close relationships and connections with the meridians and the inner organs of human body. Thus, the TCM doctors (TCMDs) can analysis and predict the condition of human body by observing the changes of the tongue body, tongue color and tongue back.However, TCMDs usually carry out diagnoses only by observation. So, the diagnosis result is lack of quantitative and standard representation and subjective. With the development of digital image processing techniques, it reveals us a new idea of achieving the qualification and standardization of TCM by combining the digital image processing techniques with it. This may lead the TCM to a new level.The research of this paper is focused on the tongue crack segmentation, tongue body segmentation and tongue color classification, which are the three of the most important techniques in this field. The major works have been done is enumerated as follows:First, a brief introduction of tongue inspection and the development of tongue crack segmentation, tongue body segmentation and tongue color classification, which are the key techniques of tongue inspection qualification is gave in the first chapter. The meanings of tongue diagnosis qualification and the contents of my work are also enumerated.Second, in order to make a progress on the accuracy of tongue cracks qualification and to change the situation that TCMDs are not willing to make their diagnoses by observing the tongue cracks, a segmentation method of tongue cracks segmentation based on the combination of modified LBP(Local Binary Pattern) operator and Otsu’s threshold technique is proposed in the second chapter to achieve the correct and complete segmentation result. Third, an active contour model and LBP operator based method for tongue body segmentation is proposed in the third chapter for overcoming the disadvantage of several popular tongue body segmentation methods. And it can solve the problem of the inclusion of dark region in mouth cavity, and achieve a more accurate result of tongue body with weak edge.In addition, on the basis of tongue cracks segmentation and tongue body segmentation a method for selecting the standard tongue image with normal color is proposed based on the histogram Bhattacharyya Distance in the forth chapter. And then, the tongue color classification is implemented based on the histogram Bhattacharyya Distance and the minimum distance classifier.Finally, the major works and innovations are summarized, and the prospects about the disadvantages of the tongue diagnosis qualification and the works that have been done in this paper are listed in detail.