CASIA OpenIR  > 毕业生  > 博士学位论文
基于图像的测量及三维重建研究
其他题名Image-Based Metrology and 3D Reconstruction
王光辉
学位类型工学博士
导师胡占义
2004-05-01
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业模式识别与智能系统
关键词单视测量 摄像机标定 三维重建 结构化场景 仿射重建 单应矩阵 Single View Metrology Camera Calibration 3d Reconstruction Structured Scenes Affine Reconstruction Homography
摘要主要以结构化场景的图像为对象,围绕单视测量、摄像机标定以及三 维重建等方面的实际问题,进行了比较深入的研究。主要内容包括:1.提出了通过单幅图像进行平面测量的三种新方法并比较了这些方法的测量 精度。第一种方法直接利用直线的对应关系来计算空间平面与图像间的单应矩阵; 第二种方法先利用两个相互垂直方向上的消隐点构造一个仿射变换矩阵,然后通 过场景中的参考距离获得欧氏度量信息;第三种方法是利用射影变换保持交比不 变的性质来计算空间距离与图像线段长度间的变换关系。此外,还给出了从单幅 图像测量点到直线的距离、直线间的夹角、平面图形的面积等的方法。2.提出了两种由参考平面单应矩阵和垂直方向消隐点恢复摄像机投影矩阵的 方法,并在此基础上提出了三种恢复场景欧氏度量信息的新方法。一是参考平面 上物体高度的测量方法;二是垂直于参考平面的平面内几何信息的度量方法;三 是任意平面内几何信息的度量方法。此外,还首次提出了由单幅图像测量某些规 则物体的体积和表面积,以及恢复场景中其它几何参数的方法。 3.证明了当场景中存在长度相等或比值已知的直线段时,可以提供新的关于 摄像机内参数的约束,并从消隐点和圆环点的角度对此进行了深入分析。提出了 一种求解摄像机外部参数和投影矩阵的简便方法,并由此恢复出场景的三维几何 结构,同时给出了一个完整场景的三维重建结果。 4.提出了一种由矩形图像线性求解摄像机内外参数的方法。该方法不需要知 道矩形的几何信息,也不涉及图像匹配问题。另外,还提出了在相差一个比例因 子意义下,恢复矩形欧氏度量的线性方法。 5.提出并证明了下述命题:一是在变参数摄像机模型下,如果没有任何可以 利用的场景信息,则无法通过两幅纯平移图像进行仿射重建;二是如果场景中含 有一张平面和一对平行直线,或者场景中含有两张平行平面,则从两个平移视点 下的图像可以线性地对场景进行仿射重建。 6.提出了一种利用场景中的几何约束关系,通过两幅图像对结构化场景进行 重建的方法。该方法的主要特点在于:一是通过一种RANSAC机制从二维图像直 接检测场景中的主平面;二是提出了一种简单的、由单应矩阵引导的直线匹配方 法,并结合点和直线的匹配来优化单应矩阵;三是在畸变因子为零的情况下,提 出了由两幅图像标定两个变参数摄像机的方法;四是本方法可以重建出在任何一 幅图像中可见的物体表面,而非仅仅重建出两幅图像的重
其他摘要The study is focused on single view metrology, camera calibration and 3D reconstruction of structured scenes. The main work is summarized as follows: 1. Chapter 2 is focused on plane metrology using a single un-calibrated image, and three novel methods are proposed. The first one uses line correspondences directly to compute homography between the space plane and its image so as to increase the computational accuracy. The second one is based on two orthogonal vanishing points. which first maps the image points to an affine space via a transformation constructed from the vanishing points, and then computes the metric distance according to the relationship between the affine space and the Euclidean space. The third one is based on the invariance of cross ratios to retrieve the lengths of space segments. In addition, a brief description on how to retrieve other geometrical entities on the space plane, such as distance from a point to a line, angle formed by two lines, area of planar objects, etc. is also presented in the chapter. 2. Chapter 3 is focused on single view metrology of 3D scenes. At first, two new methods for the determination of camera projection matrix through the knowledge of the homography of a space reference plane and its vertical vanishing point are presented. Then, three novel approaches to geometric information retrieval of the scene directly from the recovered projection matrix are proposed. The first one is to measure the height of an object on the reference plane. The second one is to take measurement on a vertical plane. The third one is to take measurement on an arbitrary plane with respect to the reference plane. In addition, the approaches to computing the volume and surface area of some certain regular and symmetric objects on the reference plane and to retrieving other geometrical entities are also presented in the chapter. 3. Chapter 4 is mainly focused on the problem of camera calibration and 3D reconstruction from a single view of structured scene. It is proved that two line segments with equal length or known length ratio in the scene can provide a new independent constraint to the image of the absolute conic. The constraint is further studied both in terms of the vanishing points and the images of circular points, and the camera can be calibrated under the widely accepted assumption of zero-skew. This chapter also presents a simple method for the recovery of camera extrinsic parameters and projection matrix with respect to a given world coordinate system. Thus, a scene structure can be reconstructed by combining planar patches. A reconstruction result of a whole building site is presented in the end of this chapter. 4. In chapter 5, a linear approach is proposed to determine the camera's intrinsic and extrinsic parameters. The method is based on the circular points derived from the images of two unparallel coplanar rectangles in space, an
馆藏号XWLW842
其他标识符842
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/5811
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
王光辉. 基于图像的测量及三维重建研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2004.
条目包含的文件
条目无相关文件。
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[王光辉]的文章
百度学术
百度学术中相似的文章
[王光辉]的文章
必应学术
必应学术中相似的文章
[王光辉]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。