双目内窥镜三维重建方法的研究

发布时间：2018-02-24 02:19

本文关键词： 双目内窥镜系统摄像机标定手眼标定立体匹配三维重建　出处：《天津工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着科技的进步和发展,基于内窥镜的微创手术与人们的生活越来越息息相关,被广泛地应用于耳鼻咽喉科治疗、心脏瓣膜切换手术,腹腔疾病的检测等领域。微创手术相比传统手术,具有手术恢复时间短,减轻术后疼痛,减小组织的切口创伤的优点。先进的手术机器人平台可以使微创技术更安全,更容易成功完成手术,越来越成为科研工作者们研究的热点,是将图像处理、模式识别、信息传输与处理、自动控制等多学科知识进行融合应用于医学影像行业,为医生提供更安全可靠的三维数据,提高手术的成功率。本文基于双目视觉的原理,重点研究双目内窥镜中的三维重建的难题。研究内容围绕摄像机标定、机械臂和相机的手眼标定、图像预处理、特征匹配和三维重建等几个方面展开。首先,依据医学内窥镜的性能指标以及使用要求搭建了双目内窥镜系统,研究了系统的整体设计方案,包括硬件模块的选择和软件设计,其中软件设计主要包含两个模块,标定模块和三维重建模块。其次,进行相机的标定和机械臂的手眼标定。根据各个坐标系之间的转换关系以及摄像机的模型,在进行双目相机参数标定的时候选用张正友标定法,得到相机的内参数以及外参数矩阵,同时分析了双目视觉系统的立体几何关系,最后研究了机械臂和双目相机的手眼标定。再次,对采集到的目标物体的图像进行预处理和图像的校正,为立体匹配做准备。在图像的特征提取和立体匹配方面,选用了局部特征匹配的方法,并且对其中SURF算法做作了改进。通过实验验证,改进的SURF算法不仅提高了匹配的正确率,而且大大降低了匹配时间,保证了算法的实时性,具有更好的性能。最后,利用第三章标定好的相机的内外参数与第四章得到的正确的匹配点对数结合,由三维重建算法解算出目标物体的三维点坐标,再根据坐标转换关系,将相机坐标系下的三维点坐标转换到机械臂坐标系下,通过纹理映射的方式,重建出不同坐标系下具有可视性和真实感的模型结构。
[Abstract]:With the progress and development of science and technology, minimally invasive surgery based on endoscope is more and more closely related to people's lives. It has been widely used in the treatment of otorhinolaryngology, heart valve switching surgery. Detection of celiac diseases. Compared with traditional surgery, minimally invasive surgery has the advantages of shorter recovery time, less postoperative pain and less wound wound. Advanced surgical robot platform can make minimally invasive technology safer. It is more and more easy to complete the operation successfully, which has become the research hotspot of researchers. It is the fusion of image processing, pattern recognition, information transmission and processing, automatic control and so on, which is applied to the medical imaging industry. In this paper, based on the principle of binocular vision, we focus on the problem of 3D reconstruction in binocular endoscope. The hand-eye calibration, image preprocessing, feature matching and 3D reconstruction of the manipulator and camera are carried out. Firstly, a binocular endoscope system is built according to the performance index of medical endoscope and the requirement of application. The whole design scheme of the system is studied, including the choice of hardware module and software design. The software design mainly includes two modules, calibration module and 3D reconstruction module. Secondly, According to the conversion relationship between coordinate system and camera model, the calibration method of Zhang Zhengyou is used to calibrate the parameters of binocular camera. The inner and outer parameter matrices of the camera are obtained, and the stereo geometry of the binocular vision system is analyzed. Finally, the hand-eye calibration of the robotic arm and the binocular camera is studied. The image of the object is preprocessed and corrected to prepare for stereo matching. In the aspect of feature extraction and stereo matching, the method of local feature matching is used. The improved SURF algorithm not only improves the accuracy of the matching, but also greatly reduces the matching time, ensures the real-time performance of the algorithm, and has better performance, through the experimental verification, the improved SURF algorithm not only improves the matching accuracy, but also greatly reduces the matching time. By combining the internal and external parameters of the camera calibrated in Chapter 3 with the correct matching point logarithm obtained in Chapter 4th, the 3D reconstruction algorithm is used to solve the 3D coordinates of the target object, and then according to the coordinate transformation relationship, The 3D point coordinates in camera coordinate system are converted to the manipulator coordinate system, and the visual and realistic model structures in different coordinate systems are reconstructed by texture mapping.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R616;TP391.41

【参考文献】