基于立体视觉的三维舌象重建算法的研究与实现

发布时间：2018-02-22 13:08

本文关键词： 立体匹配 3D舌象三维重构立体视觉方向权值　出处：《天津大学》2016年硕士论文　论文类型：学位论文

【摘要】：舌诊,又称望舌,是观察病人舌质和舌苔的变化以诊察疾病的方法,在中医传统的“四诊”之中,属于望诊的范畴,是中医有别于西医的诊法之一。由于它在中医学中具有极为重要的诊断价值,基于舌象的智能研究也越来越广泛。多数基于舌象的研究都是对二维图像,根据舌苔的颜色信息来判别病人的生理状况,而舌苔的裂纹长度、深度、以及舌苔的厚度等信息也对病情判断有很大的辅助作用,因此对于3D舌象的需求迫在眉睫。本文采用立体视觉的方法对舌象表面进行了重建,详细阐述了立体视觉理论的原理和实现过程,并对舌象实现了较好的三维复原。论文的主要工作体现在以下几个方面:1研究了相机模型及成像机理,并编写了相机采集及自动化标定软件。2研究和总结了立体匹配策略和常用的立体匹配算法,并提出了基于方向权值的立体匹配算法,实现了快速的局部立体匹配,得到了较为精确的稠密视差图。3编写了立体舌象三维重构软件,该软件集成了相机的畸变矫正,舌象分割、两幅舌象的极线校正,多种立体匹配算法,点云去噪、点云精简、点云剖分和三维显示等功能。4研究了常用的舌象分割算法,并在snakes主动轮廓模型的基础上开发了舌象半自动提取算法,经实验验证该算法的分割效果较好。5搭建了立体重构系统,并对系统的理论精度和实际精度进行了理论分析和实验验证。
[Abstract]:Tongue diagnosis, also called "gazing tongue", is a method of observing the changes of tongue quality and tongue coating of patients to diagnose diseases. In the traditional "four diagnoses" of traditional Chinese medicine, it belongs to the category of inspection. It is one of the diagnostic methods of traditional Chinese medicine, which is different from western medicine. Because of its very important diagnostic value in traditional Chinese medicine, intelligence research based on tongue image is more and more extensive. According to the color information of tongue coating to judge the physiological condition of the patient, the length, depth and thickness of the crack of tongue coating are also helpful to judge the condition of the patient. Therefore, the demand for 3D tongue image is urgent. In this paper, the method of stereo vision is used to reconstruct the surface of tongue image, and the principle and realization process of stereo vision theory are described in detail. The main work of this paper is to study the camera model and imaging mechanism in the following aspects: 1. The camera acquisition and automatic calibration software .2 has been developed to study and summarize the stereo matching strategy and the commonly used stereo matching algorithm, and a stereo matching algorithm based on directional weight is proposed, which realizes fast local stereo matching. A more accurate 3D reconstruction software for stereo tongue image is developed in dense parallax figure .3. The software integrates camera distortion correction, tongue image segmentation, polar line correction of two tongue images, various stereo matching algorithms, point cloud denoising and point cloud reduction. On the basis of snakes active contour model, a semi-automatic tongue image extraction algorithm is developed. Experimental results show that the segmentation effect of the algorithm is good. 5. The stereo reconstruction system is built, and the theoretical and practical accuracy of the system is analyzed and verified by experiments.
【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】：R241;TP391.41

【相似文献】