基于机载LiDAR点云数据的建筑物自动提取  

Automatic Building Extraction Based on Airborne LiDAR Point Cloud Data

作　　者：潘中华 金晶 陈胜林 苏韬 刘申 高能攀 PAN Zhonghua;JIN Jing;CHEN Shenglin;SU Tao;LIU Shen;GAO Nengpan(Wuhan Geomatics Institute,Wuhan 430022,China)

机构地区：[1]武汉市测绘研究院,湖北武汉430022

出　　处：《地理空间信息》2022年第5期57-59,101,共4页Geospatial Information

摘　　要：在总结前人研究成果的基础上,提出了一种将LiDAR点云数据转换为灰度图像提取建筑物的方法。首先,采用最近邻插值法将离散的LiDAR点云数据规则格网化,并标记每个规则格网点的平面坐标;再利用重采样的LiDAR点云数据生成DSM灰度图像;然后借助于成熟的图像处理技术(灰度直方图统计、连通区域面积统计、二值化、边缘检测等方法)对灰度图像进行预处理;最后利用预处理后的灰度图像标记出LiDAR点云数据建筑物点。利用VC++编写LiDAR点云数据处理程序,并对实验数据进行了验证。实验结果表明,在较平坦的区域,该方法能有效提取建筑物点,取得了良好的实验效果。With the development of LiDAR technology, more and more attention has been paid to rapid building extraction. Based on summarizing the previous research results, this paper proposed a method of extracting buildings by transforming LiDAR point cloud data into gray images.Firstly, the regular grid of discrete LiDAR point cloud data is generated by the nearest interpolation method, and the plane coordinates of each regular grid node are marked. Secondly, DSM gray images are generated from resampled LiDAR point cloud data. Then, gray images are preprocessed with the mature image processing technology(gray histogram statistics, connected area statistics, binarization, edge detection, etc.). Finally,LiDAR point cloud data building points are marked with pre-processed gray images. A“LiDAR point cloud data processing”program was written with VC++ to verify the experiment data. The experiment results show that the method proposed can effectively extract building points in a relatively flat area, and good experiment results were achieved.

关 键 词：LIDAR点云 规则格网化 灰度图像 建筑物提取 

分 类 号：P231[天文地球—摄影测量与遥感]