Research On Traditional Chinese Medicine Face Diagnosis Based On Image Recognition Processing

The objectification and modernization of TCM face color diagnosis is an inevitable choice for TCM to revitalize in modern society.The rapid development of cutting-edge technologies such as image processing and the strong support of national policies have brought opportunities for the development of objectification and modernization of TCM face-diagnosis,and provided the direction of development.Therefore,the objective research on face-to-face diagnosis is an important step to realize the development of TCM diagnosis.At present,most of the researches are aimed at the identification of facial color recognition methods and related instruments and equipment,but lack of a standard unified color measurement and evaluation system.The researchers' research is also carried out independently.Therefore,this paper combines the relevant techniques of image recognition processing to carry out the research on face-to-face objectification to verify the theoretical viewpoints of five-color diagnosis and facial five-stomach separation in TCM theory.This paper takes the theory of TCM face color diagnosis as the theoretical basis.Firstly,the segmentation extraction method of facial contour is studied.The clustering of several common face skin color models and face skin color models in color space is introduced.Then,based on YCb Cr color space The segmentation principle and segmentation effect of the three segmentation methods,such as fixed threshold segmentation method,Gaussian model segmentation method and maximum inter-class variance method,are compared.According to the effect of binary image segmentation,it is shown that for the sample collected in this paper,the segmentation of the Cb component of skin color by the maximum inter-class variance method will result in a better segmentation effect,and the maximum inter-class variance method will be automatically determined.The threshold range has been corrected.The facial contour segmentation image of all the samples is obtained based on the above segmentation method,and the noise points in the segmentation image are removed by morphological processing on the segmented face contour,and then a clear facial contour is extracted based on the edge detection,and the segmentation is obtained.The binary image is restored to an RGB spatial image without interference factors such as background and hair;Based on the Haar-like feature of the face,the facial image was segmented by the Viola-Jones algorithm.Then,based on the TCM theory,the a priori positional relationship of the five-soil feature region was obtained,and then the heart region,liver region,spleen region and lung were segmented.Five skin color blocks in the area and kidney area.Using Matlab programming,the RGB component mean of each skin patch of all samples is obtained in batches,and then the color distance between the skin patch and the face color is solved by weighted Euclidean distance,and then the skin color of the five-soil feature region and the five colors of TCM theory are studied.Similarity relationship.The results show that the color of the skin patch in the characteristic area has a certain tendency in the normal state compared with the five colors of TCM theory.The Chinese medicine viewpoint of the five color diagnosis and the facial five internal organs can support each other.