利用车载点云数据构建建筑物立面窗户模型  

Building facade window model construction of vehicle point cloud data

作　　者：范辉龙 李永强[1] 杨清科 李鹏鹏 刘亚坤 赵上斌 FAN Huilong;LI Yongqiang;YANG Qingke;LI Pengpeng;LIU Yakun;ZHAO Shangbin(School of Surveying and Land Information Engineering,Henan Polytechnic University,Jiaozuo 454000,China;Henan Water Conservancy Survey Design Research Co.,Ltd.,Zhengzhou 450000,China)

机构地区：[1]河南理工大学测绘与国土信息工程学院,河南焦作454000 [2]河南省水利勘测设计研究有限公司,河南郑州450000

出　　处：《测绘通报》2020年第5期95-100,共6页Bulletin of Surveying and Mapping

基　　金：测绘地理信息公益性行业科研专项(201412020);国家自然科学基金(41771491)。

摘　　要：采用车载LiDAR数据进行窗户模型构建是一项艰巨的工作,本文提出了一整套窗户模型构建方法。首先利用RANSAC算法对建筑物立面进行探测分离主墙面,基于空洞思想对主墙面窗户进行聚类,然后采用动态椭圆凸壳算法探测窗户边界轮廓点。对获取窗户边界点采用RANSAC算法进行分割,采用基于稳健整体最小二乘算法进行直线拟合和角点恢复,最终结合窗户的几何特征完成窗户模型构建。试验结果证明了该方法能够准确有效地构建建筑物立面中的窗户模型。Using vehicle LiDAR data to build window model is a hard work. First, the RANSAC algorithm is used to detect and separate the main wall of the building facade, and the Windows on the main wall are clustered based on empty ideas. Then, the dynamic elliptic convex hull algorithm is used to detect the boundary contour points of windows. The obtained window boundary points are segmented by RANSAC algorithm, and the linear fitting and corner point restoration are carried out by using the robust global least squares algorithm. Finally, the window model construction is completed by combining the geometric features of the Windows. The experimental results show that this method can accurately and effectively construct window models in building facades.

关 键 词：车载LiDAR 建筑物立面 RANSAC 聚类分割 模型构建 

分 类 号：P228[天文地球—大地测量学与测量工程]