基于辅助光源的舌诊客观化研究

发布时间：2018-04-14 02:30

本文选题：舌诊客观化 + 辅助光源　；参考：《天津大学》2016年硕士论文

【摘要】：舌诊是中医诊病的重要方法,传统的舌诊检测主要依靠医生的眼睛观察、文字描述,诊断结果往往受到医生经验、患者情绪和环境变化的影响,难以量化描述和精确诊断,舌象不能保留存档,阻碍了舌诊技术水平的进一步提高和中医学术交流。因此实现舌象诊断的定量化、客观化是舌诊发展的必然要求。本文从舌象诊断需求以及舌质舌苔对不同光源敏感性差异入手,设计了一套基于辅助光源的舌诊客观化系统,实现了舌象的自动采集、分割、质苔分离以及点刺的识别与定位。其中包括舌象采集系统和舌诊分析系统两个部分。本文首先搭建了一个由类标准光源、辅助光源、暗箱、数码相机、红外测距传感器、单片机、计算机等多个部分组成的舌象采集系统,能够在舌伸出度达到要求时自动拍照,得到两幅舌象——彩色舌象和单色光舌象。各硬件部分的功能由C语言和C#编程实现,其中包括单片机对红外测距传感器的AD采集控制,单片机对两种光源的转换控制,计算机与单片机的串口通讯,以及计算机对数码相机的快门控制。同时本文根据舌诊客观化需求,利用图像处理方法和OpenCV图像处理开发工具,编程实现了一个包括舌体分割、质苔分离、点刺识别的舌诊分析系统。首先采用了不同的初始轮廓提取方法对类标准光源以及辅助光源下舌图像进行Snakes算法分割;之后由图像配准以及K-means聚类实现了质苔分离并对两种光源下的分割结果进行了比较;最后通过图像预处理、边缘检测、轮廓提取、中轴线确定、区域划分等过程完成了舌点刺的识别与区域计数。通过图像配准以及辅助光源下图像的K-means聚类,进一步克服了舌质舌苔分离不理想的问题,准确率达到了86.7%;同时本文不仅实现了舌点刺的提取也对舌体区域进行了自动划分;并对不同区域的点刺进行计数,准确率能达到70%以上,在舌中和舌根的准确率更高。
[Abstract]:Tongue diagnosis is an important method of TCM diagnosis. The traditional tongue examination mainly depends on the doctor's eye observation and written description. The diagnosis result is often affected by the doctor's experience, the changes of patients' mood and environment, and it is difficult to quantify and accurately diagnose.Tongue image can not be kept on file, which hinders the further improvement of tongue diagnosis technology and academic exchange of TCM.Therefore, it is necessary for the development of tongue diagnosis to realize the quantification and objectification of tongue diagnosis.Starting from the needs of tongue image diagnosis and the difference of tongue coating sensitivity to different light sources, this paper designs an objective tongue diagnosis system based on auxiliary light source, which realizes the automatic collection, segmentation, separation of coating and identification and localization of tongue image.It includes two parts: tongue image collection system and tongue diagnosis and analysis system.In this paper, a tongue image acquisition system consisting of standard light source, auxiliary light source, dark box, digital camera, infrared ranging sensor, single chip microcomputer, computer and so on is set up, which can take pictures automatically when the tongue reaches the requirement.Two pictures of tongue-color and monochromatic light are obtained.The function of each hardware part is realized by C language and C # programming, including AD acquisition control of infrared ranging sensor by single chip microcomputer, conversion control of two kinds of light source by single chip microcomputer, serial communication between computer and single chip microcomputer.And the computer to digital camera shutter control.At the same time, according to the needs of objectification of tongue diagnosis, using image processing method and OpenCV image processing development tool, a tongue diagnosis and analysis system including tongue body segmentation, quality and fur separation and prick recognition is realized.At first, different initial contour extraction methods are used to segment the tongue image under the standard light source and the auxiliary light source, then the image registration and K-means clustering are used to realize the separation of texture and fur, and the segmentation results under the two kinds of light source are compared.Finally, tongue prick recognition and region counting are completed by image preprocessing, edge detection, contour extraction, central axis determination and region division.Through image registration and K-means clustering of images under auxiliary light source, the problem of tongue coating separation is further overcome, and the accuracy is up to 86.7. At the same time, not only tongue prick extraction is realized, but also tongue body area is automatically divided.The accuracy of spot counting in different regions is more than 70%, and the accuracy is higher in tongue and tongue root.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP274.2;R241;TP391.41

【相似文献】