一种自动提取彩色舌图像的算法  

Algorithm of automatic extraction for color tongue images

作　　者：刘媛[1] 赵鹏程 李甜 田飞[1] 陆小左[1] 

机构地区：[1]天津中医药大学中医药工程学院,天津300193 [2]天津市建筑设计院,天津300074 [3]天津市东新街社区卫生服务中心,天津300162

出　　处：《长春中医药大学学报》2017年第5期749-752,共4页Journal of Changchun University of Chinese Medicine

基　　金：国家科技支撑计划课题(2012BAI25B05);国家重点基础研究发展计划("973"计划)课题(2011CB505406);国家重点基础研究发展计划("973"计划)课题(2014CB542902)

摘　　要：彩色舌体的自动提取技术为中医舌诊客观化提供了更加便捷的操作手段。传统舌图像的提取方法不能较精确地提取所需的舌体部分,且对于舌体细节(如舌体伪轮廓和点刺)的处理也不够理想。采用首先提取彩色舌图像在RGB空间的亮度特征信息,然后使用平滑、图像增强的方法对舌图像进行处理,再使用最大类间方差法进行自动分割,以提取出彩色舌体的初始轮廓。通过提取舌图像的最大连通区域以去除其他非舌体区域,进行负向处理后再次提取最大连通区域以去掉舌体内的孔洞,最终通过数学形态学及逻辑与运算提取出所需的舌体部分。实验证明,本研究具有一定的分割效果,满足后续舌体内部感兴趣区域再提取和分析的需要。The automatic extraction of color tongue body provides a more convenient means for the tongue diagnostic objectification of traditional Chinese medicine. The traditional tongue image extraction method can not extract the required part of the tongue more accurately, and the treatment of tongue body details （such as tongue false contour and prick） are not ideal. In this paper, in order to extract the initial contour of color tongue, the luminance feature information of color tongue image in RGB space was extracted firstly, the tongue image was processed by the method of smoothing, enhancement and OTSU algorithm secondly. In order to extract the final contour of color tongue, the maximum connected area of the tongue image was extracted to remove the other non-tongue area, the maximum connected area after the negative processing was extracted again to remove the holes in the tongue, and the tongue was extracted by the method of mathematical morphology and boolean AND operator lastly. The experiments show that this research has a certain segmentation effect to meet the needs of the extraction and analysis for the following interested area in tongue body.

关 键 词：舌诊客观化 舌体提取 亮度 直方图均衡 数学形态学 

分 类 号：R241.25[医药卫生—中医诊断学]