Quantitative analysis method of tongue color and fur color in traditional Chinese medicine based on machine learning

Abstract

The present invention relates to the identification of tongue color and fur color, in particular to a machine learning-based quantitative analysis method for tongue color and fur color in traditional Chinese medicine. The tongue image survey and image collection are carried out on people in different regions and ages, and the collected tongue images are analyzed by a consistent review method. Tongue color and fur color are judged on the image, and TCM tongue image big data is constructed. The tongue body area is extracted from the tongue image image in the TCM tongue image big data. Through HSV space clustering, it is divided into sub-graphs, and the clustering results are automatically generated according to the hue. Tongue color color card, moss color color card, construct pixel color attribute classifier X based on xgboost machine learning algorithm, respectively build whole tongue tongue color classification model s and whole tongue fur color classification model t based on xgboost machine learning algorithm, the present invention provides The technical solution can effectively overcome the defects of the existing technology that the accuracy of the recognition result is low and the recognition results are relatively dependent on sample data.

Description

technical field [0001] The invention relates to identification of tongue color and fur color, in particular to a machine learning-based quantitative analysis method for tongue color and fur color in traditional Chinese medicine. Background technique [0002] Traditional tongue inspection is mainly based on naked eye observation, and its accuracy depends on the doctor's experience, limited by environmental factors, lack of objective and unified identification indicators, and cannot meet the reproducibility requirements in research. Doctors may make mistakes in judging tongue color and fur color due to factors such as light and environment. Therefore, it is urgent to use machine quantitative analysis and identification methods. There are mainly two methods in the prior art: [0003] Convolutional Neural Network Classification Model [0004] Directly use the neural network to classify and distinguish the color of the tongue coating. Taking the Inception V4 classification netw...

AI Technical Summary

Problems solved by technology

The above method is indeed improved compared with the first method, but there are also problems. For example, using the sliding window method to take blocks is random and unstable. Often a picture has many other colors, which leads to deviations in marking, resulting in the final inaccurate result

Images