Color Correction Of Face Image And Its Application In Hepatopathy Diagnosis

With the development of modern information science and technology, the modernization of the chinese medicine gains more and more attention and research from Chinese medicine scholars and departments, because using modern science and technology makes Chinese medicine spread widly and develop quickly. The face diagnosis is a main method of Chinese medicine, it mainly through the color and state of face to get the diseases information. The modernization of face diagnosis is that using computer automatic analysis color features of face and give general diagnosis results. However, there exists many problems in the modernization of face diagnosis, such as the environment of face image signal acquisition is unstable which makes the color of face image is inconsistent, except that, the color feature extraction of face and classification method is not suitable which makes the classification accuracy in liver diagnosis very low. So this paper proposes to get a solution for those problems, as follows.(1) Color correction strategy: For the face image color inconsistencies problem, we propose color correction strategy to project all color spaces to a standard color space. There has many color correction methods, we mainly compare several methods, such as, polynomial regression correction, BP neural network correction and support vector regression correction. Besides that, we came up with evaluation criteria to measure the performace of color correction method.(2) Segmentation of pathological region from face: Mainly designed segmentation method for the face images which acquired by the third generation device. The segmentation method successfully solves the problem of hair interference, so that all images can be precise positioning and segmentation.(3) Color feature extraction algorithms: We designed a series of algorithms about color feature extraction and classification, such as comparison and selection of different color spaces, different color features, different distance measurement methods, different classification methods. We use these strategies in liver disease classification. Finally, we came up with the best combination of color space, color feature, distance method and classification algorithm.(4) The face diagnosis system: Based on the research and realization of formal algorithms, finally we realize the face diagnosis system, it can get pathological regions of face automatically, do color correction and extract color feature, finally it can give out the general disease diagnostic information.