基于标志点的三维云拼接技术【本科毕业论文】.doc

基于标志点的三维点云拼接技术 摘 要 多视角点云数据拼接是三维形貌测量、三维重构中不可或缺的环节，本文针对实验室已组建的编码结构光三维测量系统要求的拼接速度快、拼接准确度高，研究了多视角点云数据的光学拼接方法及其关键技术。 本文在总结分析三维点云拼接技术和算法的国内外现状的基础上，从理论上分析了基于特征标志点的拼接方法，设计了适用于拼接方法的特征标志点图像处理方案，并且编写程序实现标志点二维坐标的提取；再分别从视觉效果和量化效果比较三点法和SVD。具体研究内容包括： 1.首先介绍了标志点的选择和匹配。本文说明了标志点的选择、灰化、平滑、阈值分割、边缘的检测与追踪、圆心定位等方法。并编写程序对标志点的二维坐标进行确定，得到标志点的二维坐标。 2.介绍了拼接原理和两种快速拼接算法的基本原理。并从理论上分析了三点法、SVD法两种基于标志点的拼接方法的准确度和速度。通过编程实现算法的计算过程，评价了上述拼接方法的优缺点和精度。 3.介绍了拼接系统的主要结构和拼接实验。拼接系统的主要结构分为硬件结构和软件结构，并说了各部分的作用；重点分析了拼接实验的过程，通过视觉效果和量化效果两方面的实验结果比较三点法和SVD法，以标志点重合度量化评价了拼接准确度。 关键词　视觉测量；拼接；SVD；标志点定位 Marker-based Three-dimensional Cloud Registration Techniques Abstract Multiple-views registration point cloud data is one of the indispensible parts of 3D topography measurement and 3D reconstruction. So in this paper, according to the characteristics of encoded structured-light of 3D measurement system set up in the lab which is noted for high accuracy, fast registration speed, applying for integral measurement of industrial component, it researches on optical registration method of point cloud data and key technology. Base on summarizing domestic and overseas status of technology of obtaining point cloud data, registering point cloud data and 3D reconstruction, it analyses the registration method base on feature marked point in theory, and designs a plan for feature marked point image processing; And write a program to extract the two-dimensional coordinates of landmarks; and then compare the two registration algorithm. Eventually, SVD algorithm has the highest accuracy. This paper studies the specific contents include the following three aspects: 1.First part is the signs point selection and matching. This article describes the selection of signs point, including ashing, smooth, threshold segmentation, edge detection and tracking, ponsitioning the center and other methods. And program to determine two-dimensional coordinates of landmarks. 2.Second part describes the principle and two rapid registration algorithm of the ba