Application And Research Of Traditional Chinese Medicine Tongue Image Retrieval Techniques Based On Color And Texture

Traditional Chinese medicine tongue diagnosis is one of the important topics at present, With the deepening of the Chinese medicine tongue diagnosis research, Tongue diagnosis digital imaging is also applied constantly in clinical, and many tongue images will produced every day in clinical. However, How to effectively retrieval and manage of the increasingly large database of tongue image and offer clinical services for the tongue diagnosis has become complex and challenging issues. The traditional management of the tongue image is often used as simple method based on artificial mark to describe the tongue image information, and according to the described information to retrieve on the tongue image, This approach has been unable to meet the large-scale tongue image database retrieval needs, and seriously affected the tongue image data play a significant role in the process of traditional Chinese medicine tongue diagnosis.So, in this paper, the content-based image retrieval technology is applied to traditional Chinese medicine tongue image retrieval is proposed, explore the retrieval method of consistent with traditional Chinese medicine tongue image. Provides a convenient and accurate query tool for the mass of the tongue image management and query, and also provide a new method for the traditional Chinese medicine tongue diagnosis.First, the paper describes the image retrieval research status at home and abroad, discusses the key technologies of the content-based image retrieval involved, including content-based image feature description and extraction, feature similarity computation and retrieval efficiency evaluation.Second, Combination of the content-based image retrieval, mainly studies the feature extraction algorithm of the tongue image low-level features such as color and texture. In tongue image color feature extraction algorithm, through research and analysis of different color space and color features, we use HSV color space and reasonable quantitative color, and then calculate the color histogram as the tongue image feature to retrieval. For the color histogram existence of the problem of zero values, we used cumulative histogram based on HSV color space, Experiments show that this methods can greatly improve the accuracy of tongue image retrieval.Third, Owing to the color feature is not enough express the tongue image information, this paper focuses on the tongue image texture features for further analysis. Proposed the tongue image texture feature extraction algorithm, we analysis and experimental comparison the GLCM based on spatial domain and wavelet transform based on frequency domain to extract tongue image feature respectively. Experiments show that the wavelet transform to extract tongue image texture features have higher accuracy than the GLCM to extract tongue image texture features. Based on in-depth study of single feature retrieval, this paper provides a retrieval methods which can comprehensive use of both color and texture to realize tongue image retrieval with multi-feature. Experiments based on tongue image database retrieval show that general features to retrieve is more in line with human perception than single feature to retrieve.Finally, we use of Visual C++6.0 design and implementation of a tongue image retrieval system prototype based on color and texture. This system provides color and texture of the single feature to search, and also provides the integrated color and texture of the multi-feature retrieval. The system overcomes the text-based image retrieval deficiency, can take advantage of the abundant information of the tongue image contained, to achieve efficient, accurate and convenient tongue image retrieval.