计算机工程与应用 ›› 2024, Vol. 60 ›› Issue (11): 233-241.DOI: 10.3778/j.issn.1002-8331.2302-0054

• 图形图像处理 • 上一篇    下一篇

机载LiDAR点云数据的建筑屋顶面提取算法

李海旺,周恒可,赵兴,郭彩玲,李柏林   

  1. 1.西南交通大学 唐山研究院,河北 唐山 063000
    2.唐山学院 河北省智能装备数字化设计及过程仿真重点实验室,河北 唐山 063000
    3.西南交通大学 机械工程学院,成都 610000
  • 出版日期:2024-06-01 发布日期:2024-05-31

Algorithm for Extracting Building Roof Surfaces from Airborne LiDAR Point Cloud Data

LI Haiwang, ZHOU Hengke, ZHAO Xing, GUO Cailing, LI Bailin   

  1. 1.Graduate School of Tangshan, Southwest Jiaotong University, Tangshan, Hebei 063000, China
    2.Key Lab of Intelligent Equipment Digital Design and Process Simulation, Tangshan University, Tangshan, Hebei 063000, China
    3.School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610000, China
  • Online:2024-06-01 Published:2024-05-31

摘要: 针对机载LiDAR点云数据的屋顶面提取过程中因受植被影响导致提取精度低的问题,提出了一种基于区域生长的屋顶面点云提取算法。进行滤波处理得到非地面点云,利用屋顶面点云邻域特征信息提取屋顶面种子点,引入植被指数和RGB差值信息作为生长约束条件对屋顶面点云进行生长分割,利用屋顶面的高程与面积值对提取结果进行过滤优化,得到屋顶面点云。选取了农村、城市、工厂三组不同场景的测试数据进行实验,结果表明:Kappa系数分别达到了97.29%、97.82%、97.13%,算法可实现较好的建筑屋顶面提取效果,且针对不同建筑场景具有良好的适应性。

关键词: 机载LiDAR, 屋顶面提取, 邻域信息, 区域生长, 植被指数

Abstract: A region-growing-based point cloud extraction algorithm for roof surfaces is proposed to address the issue of low extraction accuracy caused by vegetation interference in airborne LiDAR point cloud data. Firstly, the filtering treatment is conducted to obtain the non-ground point cloud. Then, the neighborhood feature information of the roof point is used to extract roof surface seed points, in which the vegetation index and RGB difference information are introduced as growth constraints to segment the roof surface point cloud. Finally, the extracted results are filtered and optimized by using the elevation and area values of the roof surface to get the point cloud of the roof surface. The experiment is carried out by selecting three groups of test data from different scenarios, such as rural areas, urban and factory. In accordance with the results, it indicates that the Kappa coefficients reach 97.29%, 97.82% and 97.13% respectively, a relatively better building roof surface extraction effect can be realized, and a good adaptability for different building scenes is embodied.

Key words: aerial LiDAR, roof surface extraction, neighborhood information, region growing, vegetation index