基于双目立体视觉系统的大视域图像拼接和三维重建及测量的应用研究

发布时间：2018-03-12 18:49

本文选题：双目视觉视觉系统　切入点：大视域　出处：《河南工业大学》2013年硕士论文　论文类型：学位论文

【摘要】：双目立体视觉系统是计算机视觉的一个重要的分支,是由不同位置的相机经过移动或旋转拍摄同一景物,通过计算空间物点在左右图像对中的视差,获取该点的三维信息的过程。主要过程有：图像获取、相机标定、图像匹配和三维实体模型的重建。本文基于人和动物的双目视觉,设计了一套双目立体视觉系统,用两个相机模仿人和动物的双眼,获取场景的图像信息。采用基于标定板的相机标定方法对双目立体视觉系统进行标定,在对双目视觉系统光学分析的基础上,提出一种直接图像拼接的方法,即不需要通过相关的算法搜寻相邻两幅图像对中相同的特征来确定他们的相对位置,对初始图像直接剪切并拼接成一幅大视域的图像拼接方法,实现了大视域图像的快速拼接。对由系统误差造成的相邻图像对之间的错位,通过计算拼接处的相关性,得到图像拼接边缘的最佳匹配位置,并获得图像之间的错位值,对图像拼接产生的错位进行了校正。针对相机问光的强度不同造成的拼接缝,采用改进的加权法消除图像拼接缝处的人为痕迹。最后,本文相机标定算法的实现应用HALCON软件,借助HALCON强大的图像处理功能来完成相机标定。对拼接大视域图像中的标记点进行三维坐标提取,实现三维数字测量,并用Delaunay三角剖分的方法拟合三维曲面。然后运用Visual Studio2010和函数库AutoCAD二次开发工具ObjectARX2013进行二次开发,运用面向对象的技术实现三维重建。
[Abstract]:Binocular stereoscopic vision system is an important branch of computer vision. It is used by cameras in different positions to photograph the same scene through moving or rotating, and the parallax between the objects in the left and right image pairs is calculated by calculating the parallax of the spatial object points in the left and right image pairs. The process of obtaining 3D information of this point includes image acquisition, camera calibration, image matching and 3D solid model reconstruction. Based on the binocular vision of human and animal, a binocular stereo vision system is designed in this paper. Two cameras are used to imitate the eyes of human and animal. The camera calibration method based on calibration board is used to calibrate the binocular stereo vision system. Based on the optical analysis of the binocular vision system, a direct image stitching method is proposed. That is to say, it is not necessary to search for the same features of two adjacent image pairs to determine their relative position, and to directly cut and splice the initial image into a large horizon image mosaic method. The fast stitching of large horizon images is realized. By calculating the correlation between adjacent image pairs caused by system error, the best matching position of image stitching edge is obtained, and the dislocation value between images is obtained. The misalignment caused by image stitching is corrected and the artificial traces of the stitches in the image stitches are eliminated by the improved weighting method for the stitches caused by the different intensity of the camera light. Finally, the realization of camera calibration algorithm is realized by using HALCON software, and the powerful image processing function of HALCON is used to complete camera calibration. The 3D surface is fitted with Delaunay triangulation method, and then Visual Studio2010 and AutoCAD secondary development tool ObjectARX2013 are used to carry out the secondary development, and the object oriented technology is used to realize 3D reconstruction.
【学位授予单位】：河南工业大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P23;TP391.41

【参考文献】