摘要
望眼神是中医望诊的重要内容,能直接反映患者的病理生理状态。传统的望眼神主要由医生的直接目测进行判断,主观性较强,精确性及一致性较差。本研究运用现代图像处理技术和机器学习等方法对中医眼神的计算机自动识别进行了研究,提取了眨眼频率、单次眨眼时间、长眨眼次数等6个能够反映眼神的数字特征,并通过特征选择筛选出其中5个特征组成了眼神识别特征集,建立了中医眼神特征识别方法,为辅助临床辨证诊断提供了客观依据。
Eyes spirit diagnosis is an important part of inspection of traditional Chinese medicine, which can directly reflect pathophysiological state of patients. The traditional eyes spirit diagnosis is mainly conducted in direct visual judgment by the doctor, so it is some extent subjective and less of accuracy and consistency. This research aims to study computer automatic recognition of eyes spirit through the modern image processing technology and machine learning. It extracts 6 digital indexes, such as blink frequency, single blink time, long blink times and so on, which can reflect the eyes spirit. Five of the features are selected to form the eye recognition feature set, and the eye feature recognition method is established to realize the automatic recognition of the eyes of the Chinese medicine practitioner.
出处
《中华中医药杂志》
CAS
CSCD
北大核心
2018年第3期886-889,共4页
China Journal of Traditional Chinese Medicine and Pharmacy
基金
国家自然科学基金面上项目(No.81373555)
张江高科技园区中医药发展扶持项目(No.PZY2015-2)~~
关键词
眼神
特征
图像处理
机器学习
计算机识别
分类
Eyes spirit
Characteristic
Image processing
Machine learning
Computer automatic recognition
Classification
