Study On The Extraction And Classification Of Tongue Features In The Tongue Diagnosis Of Traditional Chinese Medicine

The use of tongue diagnosis in traditional Chinese medicine can be traced back to thousands of years ago.However,due to the limitation of conditions,the traditional tongue diagnosis in traditional Chinese medicine has been lack of theoretical basis of modern medical science.Influenced by its subjectivity,the diagnostic method is determined by the prior knowledge of traditional Chinese medicine and the macroscopic performance of the patients,lacking ideal objective reference standards.In addition,traditional Chinese medicine and Western medicine are different in concept,so that the clinical experience and conclusions of traditional Chinese medicine can not be reasonably applied.Therefore,the development of tongue diagnosis in traditional Chinese medicine has been restricted in many aspects.In order to overcome the shortcomings and deficiencies of traditional Chinese medicine tongue diagnosis and continue to develop modern Chinese medicine tongue diagnosis,this paper carried out a variety of studies.According to the development of tongue diagnosis,the extraction of representative color of tongue image,the extraction of color and texture features of tongue image,the pretreatment of tongue image samples and the segmentation of fur of tongue image are studied.Tongue image color analysis: By analyzing the extraction method of tongue image color,the relationship between tongue image color and symptoms,it is found that the clustering algorithm can obtain better representative color of tongue image.In addition,the representative color obtained by clustering can be used as a reference standard for fur segmentation experiment and tongue image classification model establishment,which helps to solve the problem of lack of reference object in classification experiment.The method of extracting tongue features is mainly to extract the color and texture characteristics of tongue.Color histogram,color moment and tongue color threshold algorithm are the main methods to extract color features of tongue.Establishment and performance evaluation of tongue image fur segmentation and classification model: Through studying and analyzing the advantages and disadvantages of several commonly used single-kernel and multiple-kernel SVM segmentation algorithms in tongue image fur segmentation,and analyzing the performance of various classification algorithms(such as KNN,single-kernel SVM and PSO multiple-kernel SVM)in tongue image classification,an improved PSO optimized hybrid algorithm is proposed.A kernel SVM classification algorithm.Finally,the experiments of fur segmentation on different tongue images and the classification experiments on different tongue images are carried out.The experimental results show that the algorithm achieves the best results in fur segmentation experiments and the establishment of classification model of tongue images.In this paper,the pretreatment of tongue image,the exact extraction of tongue image features and the segmentation of tongue image fur have been studied deeply.On this basis,a classification model has been designed,which has made a beneficial contribution to the development of tongue diagnosis in modern Chinese medicine.