利用航空影像和LiDAR点云进行建筑物重建的方法研究

发布时间：2018-03-08 11:41

本文选题：航空影像　切入点：LiDAR点云　出处：《西南交通大学》2013年硕士论文　论文类型：学位论文

【摘要】：建筑物信息是数字城市的重要组成部分,也是智慧城市中的感知数据,而且还是智慧城市进一步对感知数据进行融合、分析和处理,与业务流程智能化集成,继而主动做出响应,促进城市各个关键系统和谐高效地运行的基础。然而建筑物不断发生变化,由此如何快速高效地获取建筑物的三维信息引起广大专家、学者的关注。针对数字城市中建筑物三维建模所需大量人工干预、耗费大量时间等问题,通过对航空影像和机载LiDAR点云数据进行建筑物特征提取研究,利用航空影像和机载LiDAR点云数据,研究采用区域增长、三维Hough变换以及RANSAC算法提取建筑物顶面面片,实现简单建筑物白模构建。本文开展了如下工作： (1)归纳总结了影像、LiDAR点云数据预处理过程。包括：LiDAR点云数据的粗差剔除、滤波、分类,影像正射纠正,以及影像与LiDAR点云数据配准方法。 (2)研究了现有建筑物顶面面片提取方法。分析总结现有提取建筑物顶面面片方法,主要包括区域增长、三维Hough变换以及RANSAC算法。 (3)归纳总结了基于影像、LiDAR点云数据建筑物三维重建的方法。从全自动和半自动角度进行三维建筑物重建方法的归纳分析。总结了三维建筑物重建过程中需要注意的地方。 (4)对实验区域的建筑物采用了区域增长、三维Hough变换、RANSAC算法进行建筑物顶面面片提取实验。结合三种方法的特征,分析了三种方法提取面片的效率等问题。结合实验结果,对比分析了三种方法的优缺点。实验表明：采用DBSCAN聚类分析,能够优化建筑物顶面面片提取流程,能够有效避免顶面面片错分情况；采用RANSAC算法提取建筑物顶面面片较区域增长结果可靠,较三维Hough变换效率高；采用三维Hough变换提取建筑物顶面面片可以获取较可靠的面片法向量信息,便于进行建筑物三维重建；采用影像和LiDAR点云数据可以提取建筑物地面轮廓线以及顶面特征线,从而实现建筑物三维重建,但仍需一定的人工干预。
[Abstract]:Building information is an important part of the digital city, and it is also the perceptual data in the intelligent city, and it is also the intelligent city to further fuse, analyze and process the perceptual data, and integrate with the business process intelligently. Then the foundation of promoting the harmonious and efficient operation of every key system in the city is promoted. However, the buildings are constantly changing, so how to obtain the three-dimensional information of the buildings quickly and efficiently has aroused the majority of experts. Aiming at the problems of large amount of manual intervention and time consuming in 3D modeling of buildings in digital cities, the feature extraction of buildings based on aerial images and airborne LiDAR point cloud data is studied. Using aerial image and airborne LiDAR point cloud data, this paper studies the use of region growth, 3D Hough transform and RANSAC algorithm to extract the top surface of building to realize the construction of simple building white model. The work of this paper is as follows:. 1) the preprocessing process of LiDAR point cloud data is summarized, including gross error elimination, filtering, classification, orthophoto correction, and the registration method between image and LiDAR point cloud data. 2) the existing methods of extracting building top surface are studied, and the existing methods of extracting building top surface are analyzed and summarized, including region growth, 3D Hough transform and RANSAC algorithm. The methods of 3D building reconstruction based on LiDAR point cloud data are summarized. The methods of 3D building reconstruction from automatic and semi-automatic angles are summarized. The points needing attention in the process of 3D building reconstruction are summarized. In this paper, three dimensional Hough transform algorithm is used to extract the top surface of the building. The efficiency of the three methods is analyzed according to the characteristics of the three methods, and the experimental results are combined with the results of the experiment. The advantages and disadvantages of the three methods are compared and analyzed. The experimental results show that the DBSCAN clustering analysis can optimize the extraction process of the top surface of building, and can effectively avoid the fault of the top surface, and the RANSAC algorithm is used to extract the top surface of the building which is more reliable than the regional growth results. The efficiency of 3D Hough transform is higher than that of 3D Hough transform, and the information of normal vector of facet can be obtained by using 3D Hough transform, which is convenient for building 3D reconstruction. Image and LiDAR point cloud data can be used to extract the contour of the ground and the feature line of the top surface of the building, so as to realize the 3D reconstruction of the building, but some artificial intervention is still needed.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P231;P225

【参考文献】