摘要
目的舌苔、舌质分离对后续肿瘤患者舌象的客观化辨证具有重要的意义。常用的算法是基于颜色空间通道舌图像的K-means聚类算法。CIELAB颜色空间的a*通道舌图像相较于其他颜色空间通道的舌图像分割结果稳定,常用于后续的分割。对于部分舌图像,a*通道舌图像的舌苔、舌质虽然具有一定的区分度,但二者的区分度并不是十分明显,影响后续的分割结果。因此,本文提出一种基于直方图均衡化的伽马校正和K-means聚类的舌苔、舌质分离方法。方法采用200幅肿瘤患者的舌图像作为实验样本。首先将舌图像从RGB颜色空间转换到CIELAB颜色空间,对a*通道舌图像进行直方图均衡化增强以及伽马校正,然后利用K-means聚类方法对增强后的舌图像舌苔、舌质分离,得到直方图均衡化以及伽马校正后的a*通道舌图像和分割后的舌苔、舌质图像。为了验证算法的可行性,请5位专业中医医生对200例肿瘤患者的舌图像舌苔、舌质分割效果进行辨析。结果进行直方图均衡化以及伽马校正后的a*通道舌图像舌苔、舌质分割结果明显强于未经处理的a*通道舌图像分割结果。经辨析,分割合格率为97%。结论该方法可以很好地实现舌苔、舌质分离,具有一定的应用价值。
Objective Tongue coating and body separation has an important significance to the subsequent objective syndrome differentiation of tongue images of tumor patients.The common algorithms are K-means clustering algorithm based on tongue image of color space channels.The a^*channel tongue image in CIELAB color space is stable compared with other color space channels in tongue segmentation.It is often used for subsequent segmentation.For some tongue images,tongue coating and body of the a^*channel have a certain degree of differentiation,but the degree of discrimination between the two is not very obvious,which affects the subsequent segmentation results.Therefore,we designed a separation method of tongue coating and body based on histogram equalization and gamma correction and K-means clustering.Methods 200 tongue images of tumour patients were used as experimental samples.Firstly,the tongue image was transformed from RGB color space to CIELAB color space.The tongue image of the a^*channel was enhanced by histogram equalization and gamma correction.Secondly,tongue coating and body of enhanced tongue image was separated by the K-means clustering method.The histogram equalization and the gamma corrected a^*channel tongue image and the segmented tongue coating and tongue body images were obtained.In order to verify the feasibility of the algorithm,five professional Chinese medicine doctors were asked to discriminate results of tongue coating and tongue body segmentation on 200 patients with tumors.Results The tongue coating and body segmentation results of the a^*channel tongue image after histogram equalization and gamma correction are obviously better than unprocessed a^*channel tongue image segmentation results.After analysis,the qualified rate of segmentation is 97%.Conclusions This method can achieve good separation of tongue coating and body and has a certain application value.
作者
韩立博
胡广芹
张新峰
冯利
李泉旺
蔡轶珩
HAN Libo;HU Guangqin;ZHANG Xinfeng;FENG Li;LI Quanwang;CAI Yiheng(Faculty of Information Technology,Beijing University of Technology,Beijing 100124;College of Life Science and Bioengineering,Beijing University of Technology,Beijing 100124;Cancer Hospital Chinese Academy of Medical Sciences,Beijing 100021;Dongfang Hospital,Beijing University of Chinese Medicine,Beijing 100078)
出处
《北京生物医学工程》
2019年第1期1-6,共6页
Beijing Biomedical Engineering
基金
国家中医药管理局资助项目(GZY-FJS-2018-023)
国家重点研发计划中医药现代化研究专项项目(2017YFC1703300)资助
作者简介
通信作者:胡广芹。E-mail:hdmh@163.com
