基于TIN法向量的边缘检测与建筑物提取方法研究

发布时间：2018-04-23 07:15

本文选题：LiDAR + 点云　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：机载激光雷达技术(Light Detection And Ranging,LiDAR)结合了航空影像技术与遥感技术的优点,这种创新性的思路为地形探测带来了一次重大的改革。机载LiDAR技术可以获取高精度海量的离散点云数据,并且可以根据需求调整点云的密度,使得测量技术更加智能化,实用化。近年来,随着城市建设的高速发展,如何精确快速获取城市地形、建筑物分布、以及各类地物的边缘就显得特别重要。目前,通过LiDAR数据进行城市三维重建越来越受到人们的重视,而建筑物的提取是其关键的一步。本文根据不同地形和地物的外部特征,对LiDAR数据的边缘检测以及建筑物提取进行了研究。主要内容包括以下两部分:1.针对传统的边缘检测方法梯度计算量大的问题,提出了一种基于不规则三角网(Triangulated Irregular Network,TIN)法向量的LiDAR点云数据的边缘检测新方法。首先构建不规则三角网,计算每个三角形的法向量以及每个法向量与水平面的夹角,通过跟夹角阈值比较,提取出边缘三角形,将边缘三角形的最高点作为边缘点。夹角阈值通过直方图统计得到。实验结果表明,该算法能够较好的提取LiDAR点云数据的边缘点。本文对河流区域出现的数据空白进行插值,得到了较好的河流边缘。2.针对传统建筑物检测数据点不完整的问题,提出了一种通过LiDAR数据轮廓线提取建筑物的新方法。首先对滤波后的点云数据构造不规则三角网,根据每个点临近的三角形法向量夹角判断三角面共面,检测出初始建筑物点。然后利用边长阈值把三角网分离成建筑物内部三角网和边界三角网,提取边界三角网中的较短边拼接得到建筑物轮廓。最后通过判断每个点是否在建筑物轮廓内部,提取出轮廓线内部的建筑物点。实验结果表明,该算法能够较好的提取Li DAR数据的建筑物点,并准确地得到了建筑物的三维轮廓线。最终的建筑物提取正确率可以达到95%以上,完整率可以达到90%以上,达到了建筑物提取的精度要求。
[Abstract]:Airborne lidar technology combines the advantages of aerial image technology and remote sensing technology, which brings an important innovation for terrain detection. Airborne LiDAR technology can obtain high accuracy and mass discrete point cloud data, and can adjust the density of point cloud according to the demand, making the measurement technology more intelligent and practical. In recent years, with the rapid development of urban construction, how to accurately and quickly obtain the urban topography, the distribution of buildings, and the edge of various features is particularly important. At present, people pay more and more attention to urban 3D reconstruction through LiDAR data, and the extraction of buildings is a key step. In this paper, the edge detection and building extraction of LiDAR data are studied according to the external features of different terrain and features. The main content includes the following two parts: 1. A new edge detection method for LiDAR point cloud data based on triangulated Irregular network normal vector is proposed to solve the problem of large gradient computation in traditional edge detection methods. Firstly, the irregular triangular network is constructed, the normal vectors of each triangle and the angle between each normal vector and the horizontal plane are calculated. By comparing with the angle threshold, the edge triangle is extracted and the highest point of the edge triangle is taken as the edge point. The angle threshold is obtained by histogram statistics. Experimental results show that the algorithm can extract the edge points of LiDAR point cloud data. In this paper, the data gaps in the river region are interpolated, and a better river edge. 2. 2. In order to solve the problem of incomplete data points in traditional building detection, a new method for extracting buildings by LiDAR data contours is proposed. Firstly, irregular triangulation is constructed for the filtered point cloud data, and the initial building point is detected according to the angle of the triangle normal vector near each point. Then the triangulation network is separated into the building inner triangular network and the boundary triangulation network by using the edge length threshold, and the building contour is obtained by extracting the shorter edges from the boundary triangulation network. Finally, by judging whether each point is inside the building contour, the building points inside the contour line are extracted. The experimental results show that the algorithm can extract the building points of Li DAR data and get the 3D contour of the building accurately. The final correct rate of building extraction can reach more than 95%, and the complete rate can reach more than 90%, which meets the precision requirement of building extraction.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU198;TN958.98

【参考文献】