Research On Analysis Model Of Traditional Chinese Medicine Diagnosis Based On Visual Perception Fusion

Dialectical diagnosis of Traditional Chinese Medicine based on the four diagnostic methods,and in order to obtain the visual perception features from one patient,visual inspection of Traditional Chinese Medicine has to observe patient mental state,complexion,body,status and other external diagnostic information.In order to overcome the limitations of traditional subjective diagnosis and provide an effective solution to the reusability problem faced in the study of modernization of TCM.We start this topic with objectifying visual inspection and utilize one capture device with standardized visual perception measure to capture three types of visual perception images in the cooperative hospital,including tongue,face and sublingual vein.Then we give a quantitative processing on them with the help of digital image processing technology.After that,seven diagnostic features based on statistical features and pathological features are extracted to be analyzed and researched,they are tongue color features,tongue geometry features,tongue texture features,facial color features,facial texture features,sublingual vein color features and sublingual vein geometry features.According to the features of large-scale visual perception,this topic focuses on the data diagnosis,quantitative features and the complex relationship between different diseases.In this topic,pattern recognition method is used to fuse the multiple perception features extracted from patients under different healthy status,and explore the rules hidden behind a variety of diagnostic features in order to provide strong support for the clinical diagnosis.There are two kinds of diagnose fusion strategies in this topic,they are feature-level fusion and decision-level fusion.Feature-level fusion focuses on the analysis of a number of independent sub-characteristics of a disease and find out a group of features that show a strong performance in the disease relative to the health.Then we fuse them to integrate into one new feature with the use of canonical correlation analysis theory,so we can make the final diagnostic interpretation by identifying the new feature.At the same time,a fusion method based on the kernel canonical correlation analysis is used to fuse a group of diagnostic features in order to further analyze the relationship between diagnostic features.Decision-level fusion improves the performance of the whole recognition system by fusing out the final output of one pattern sample on the basis of a certain mathematical model with the supplement each other of the local recognition results of multi-feature and multi-classifier.Faced with different outputs from different classifiers,we make use of a fusion method based on Bayesian theory in multi-diagnosis feature analysis,and its performance is compared with another fusion method based on voting rule.Then,after a comprehensive performance analysis of SRC classifier and SVM classifier,a joint classifier based on SRC and SVM is proposed which aims at improving the recognition result of single classifier.The final binary classification experimental results between health and disease show that all the different fusion methods can play a different role in the diagnostic analysis of visual perception features,which indicates that the fusion analysis of visual perception is of great significance to the modern study of Chinese medicine.