Research And Application Of Detection And Inpainting Algorithm For Tongue Image Glisten Points

Tongue diagnosis is one of the methods which have the most research and application value. In recent few years, with the rapid development of computer science, researchers begin to use lots of computer technologies to do medical image processing. With the application of image processing and pattern recognition in tongue diagnosis, it's urgent need to do preprocessing on tongue image before other works. The purpose of this dissertation is contributing to the solving of glisten points detection and filling problem by introducing the concept of Tatal Variation and Texture Synthesis.The major contributions of this dissertation are: improve,design and implement the self-adaptive data fusion glisten points detection algorithm, this algorithm can detect the glisten points of the tongue image more completely than before; introduce the image inpainting technique into the glisten points filling problem of the tongue image; at last, we verify the image inpainting technique can improve the effect of glisten points inpainting and is helpful to the follow-up study through lots of experiments.First, we do the research of glisten points detection, and propose a new self-adaptive data fusion glisten points detection algorithm, to point against the bad points of the original algorithm based on the data fusion theory, and show that the algorithm is better than original algorithm through a large number of comparative tests.Second, propose the idea that apply the image inpainting technique based on the Total Variation model into tongue image glisten points inpainting, and make this model more suitable for this problem, and at last, show that the method is better than original method through a large number of comparative tests.Third, propose the idea that apply the image inpainting techenique based on the texture synthesis into tongue image glisten points inpainting, and improve it when implement it, make it more suitable for this problem, at last proves the advantage of the method by experiments.Finally, we talk about when to use the aforementioned two kinds of algorithm, give the definition of big area and small area of tongue image, after that, we use the two algorithms to deal with different size of the area, and at last obtain a good result.