基于舌象特征的中医体质自动辨别系统研究

An automatic discrimination system for Traditional Chinese Medicine constitution based on tongue image features

采用基于TensorFlow的深度学习技术对舌象进行精准分割,提取舌体的颜色特征、形状特征和纹理特征,根据中医名家知识库与中医体质诊断规则,找出舌象各项特征与中医体质的对应关系,构建“舌象-体质”关系模型,自动辨别其可能的中医体质类型:平和质、气虚质、阴虚质、阳虚质、痰湿质、湿热质、淤血质、气郁质、特禀质,最后实现用户通过微信小程序拍摄舌像照片,上传照片,识别中医体质类型,填写问卷,生成相应的特质报告和养生建议。

In this study,tensor flow based deep learning was used to accurately segment the tongue image,extract the color characteristics,shape characteristics and texture characteristics of tongue body,find the correspondence between each characteristic of tongue image and Traditional Chinese Medicine(TCM)constitution according to the knowledge base of famous TCM and the diagnostic rule of TCM Constitution,construct the“Tongue-Constitution”relationship model and automatically discern its possible TCM constitution types:smooth sum quality,Qi deficiency,yin deficiency,Yang deficiency,phlegm and dampness,dampness and heat,congestion,Qiyu and idiosyncrasy,and finally to realize the users to take tongue image photographs through Wechat applet,upload photographs,recognize the type of TCM body,fill in questionnaires and generate corresponding trait reports and health advice.