基于数字摄影测量系统的核线影像模块研究

发布时间：2019-02-16 09:21

【摘要】：纵观摄影测量和计算机视觉技术的发展，影像处理的格式日渐繁多，而且自动化程度不断提高。无论是近景摄影测量还是计算机视觉领域都避免不了影像匹配的问题，影像匹配是一个瓶颈问题，怎样提高影像匹配精度以及如何缩短影像匹配时间是一个关键性问题。在20世纪70年代，自从核线概念被提出来后，核线相关在影像匹配上起到了关键作用。核线是摄影测量的一个基本概念，在立体像对上，同名像点一定位于同名核线上，而且同名核线对上的像点是一一对应的，因而如果能在立体像对上确定同名核线，那么就可以根据同名核线的性质，将二维的影像匹配转变为一维匹配，大大的提高了计算效率和可靠性。为了方便论文的展开以及对数字影像的处理，首先设计了一个数字摄影测量系统框架，包含工程的创建、影像的读取、相机文件、控制点文件的读取和创建等基本功能。然后只对核线影像生成模块展开了讨论，分别采用基于数字纠正的方法、基于共面条件的方法以及求解相对定向直接解的方法生成核线影像，因为实验的影像数据不包含内方位元素等信息，而且前两种方法都涉及到共线条件方程的解算，所以采用第三种方法编程实现，并利用相对定向的直接解进行核线排列，暂不考虑整幅影像生成核线影像，只用单个像点进行测试，，结果达到一定精度。论文的最后一部分简单介绍了基于射影几何生成影像的方法，它和摄影测量中基于共面条件的方法可以相互转换，这部分的讨论不作为本文的重点。该论文是一篇工程应用型文章，详细的讨论了核线影像生成部分，可以为下一步影像匹配作铺垫。
[Abstract]:With the development of photogrammetry and computer vision, the format of image processing is increasing and the degree of automation is increasing. Neither close-range photogrammetry nor computer vision can avoid the problem of image matching. Image matching is a bottleneck problem. How to improve the accuracy of image matching and how to shorten the time of image matching is a key problem. Since the concept of nuclear line was proposed in 1970s, nuclear line correlation has played a key role in image matching. Nuclear line is a basic concept of photogrammetry. In stereo image, the same name image point must be located on the nuclear line of the same name, and the image point on the pair of the same name nuclear line is one-to-one correspondence, so if the nuclear line of the same name can be determined on the stereo image pair, Then according to the properties of the kernel line of the same name, the two-dimensional image matching can be transformed into one-dimensional matching, which greatly improves the computational efficiency and reliability. In order to facilitate the development of the thesis and the processing of digital image, a digital photogrammetry system framework is designed firstly, which includes the basic functions of engineering creation, image reading, camera file, control point file reading and creating and so on. Then only the kernel line image generation module is discussed. The method based on digital correction, the method based on coplanar condition and the method of solving the relative directional direct solution are used to generate the core line image, respectively. Because the experimental image data does not contain information such as internal azimuth elements, and the first two methods involve the solution of collinear conditional equations, the third method is used to program and arrange the kernel lines using the direct solution of relative orientation. The kernel line image is not considered for the whole image, but only a single image point is used to test, and the result reaches a certain accuracy. In the last part of the paper, the method of image generation based on projective geometry is briefly introduced, which can be converted to the method based on coplanar condition in photogrammetry. This part of the discussion is not the focus of this paper. This paper is an engineering application article, the core line image generation part is discussed in detail, which can pave the way for the next image matching.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P231.5

【参考文献】