基于机载LiDAR点云数据的建筑物提取与建模研究

发布时间：2019-07-05 14:44

【摘要】：建筑物是城市中最主要的元素，对城市地区建筑物进行建模对于数字城市的建立具有较为重要的意义。机载LiDAR测量技术的发展为获取城市空间信息提供了一种全新的技术手段，其中机载LiDAR所获取的点云数据中包含有大量的建筑物空间信息，特别是由于机载LiDAR的工作原理，建筑物信息拥有大量的丰富的建筑物屋顶信息。因此机载激光雷达的发展为数字城市中建筑物建模需求提供了有力的支持。本文主要对机载激光雷达的技术发展、数据的特点、国内外发展状况及未来的发展趋势做出相关的描述，并提出了关于建筑物点云数据的建筑物提取及重建的理论方法。本文的具体内容如下：一、对原始数据的预处理工作，包括误差点的剔除，滤波和提取建筑物点三个过程，经过这几步为后面的建筑物点预处理做好提前准备工作。二、对建筑物群点云采用区域增长的聚类方法将建筑物与建筑物之间分割开，由于剔除非建筑物数据点过程存在误差等原因，并不能将每个建筑物都分割成单独建筑物，作者对现有射线法进行改进采用人工对错误分割的建筑物再次目视判断分割。三、对现有的凸壳算法进行了改进，利用改进后算法获取建筑物点云数据的包围壳点，并将这些点连接并规则化后生成建筑物的规则边界轮廓线。四、除平顶房屋外，建筑物顶部通常是由多个面构成的，构建建筑物的简单模型就需要判定建筑物屋顶点的面片归属类别，文中对现有的三角网法向量法进行了改进，对点的面片类别进行分类取得良好的效果。五、采用最小二乘法，对各建筑物中同属于一个面片的点云拟合最佳面片方程，求取相交平面的直线方程，结合规则轮廓线以及投影理论求建筑物的特征点。六、本文根据以上步骤，，对平顶建筑物以及人字顶、四坡顶、L形人字顶等简单建筑物构建模型，并取得较好效果。
文内图片：

图片说明：1规则房屋屋顶类型Fig.1.3.1Thetypeofregularbuildingroofs
[Abstract]:Buildings are the most important elements in the city, so it is of great significance to model the buildings in urban areas for the establishment of digital cities. The development of airborne LiDAR measurement technology provides a new technical means for obtaining urban spatial information. The point cloud data obtained by airborne LiDAR contains a large number of building spatial information, especially because of the working principle of airborne LiDAR, building information has a large number of building roof information. Therefore, the development of airborne lidar provides strong support for building modeling requirements in digital cities. In this paper, the technical development of airborne lidar, the characteristics of data, the development status at home and abroad and the development trend in the future are described, and the theoretical methods of building extraction and reconstruction of building point cloud data are put forward. The specific contents of this paper are as follows: first, the preprocessing of the original data, including the elimination of error, filtering and extraction of building points, after these steps for the post-building point preprocessing to do the preparatory work in advance. Secondly, the clustering method of regional growth is used to divide the building from the building. Because of the error in the process of eliminating the non-building data points, each building can not be divided into separate buildings. The author improves the existing beam method and adopts the artificial visual judgment segmentation of the incorrectly segmented buildings. Thirdly, the existing convex hull algorithm is improved, and the surrounded shell points of building point cloud data are obtained by using the improved algorithm, and these points are connected and standardized to generate the regular boundary outline of the building. Fourth, in addition to the flat roof house, the top of the building is usually composed of multiple faces. It is necessary to determine the classification of the facet at the top of the building. In this paper, the existing triangular net normal vector method is improved, and good results are obtained in the classification of the facet categories of the points. Fifth, the least square method is used to fit the best patch equation for the point cloud which belongs to the same facet in each building, and the straight line equation of the intersecting plane is obtained, and the characteristic points of the building are obtained by combining the regular outline and the projection theory. 6. According to the above steps, this paper constructs models for flat-topped buildings and simple buildings such as herringbone roof, four-slope roof, L-shaped herringbone top and so on, and obtains good results.
【学位授予单位】：辽宁工程技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P225.1

【参考文献】