华中科技大学学报（自然科学版）2025，Vol.53Issue(8)：24-29,6.DOI:10.13245/j.hust.250290
点云数据驱动的建筑群轮廓图快速生成方法
Automatic contour extraction of large-scale building group using point clouds
摘要
Abstract
An automated method was proposed for generating architectural group outlines from point cloud data.Initially,the original 3D point cloud model of the building group was captured by a laser scanner,followed by the point cloud preprocession,including downsampling and denoising.Subsequently,the preprocessed data were horizontally sliced and projected.The projected planar point clouds were transformed into an image realizing data dimensionality reduction.Finally,by integrating Gaussian blur,Canny edge detection,and Hough transform,wall lines were extracted from the binary image and the outlines of the building group are generated.The proposed algorithm was integrated and a Web application was developed to quickly generate the end to end architectural profile.The proposed method was validated through a case study.The results show that the automatically generated outline achieves an Intersection over Union(IOU)of 96%,with an average dimensional error of only 1.4%,and the entire framework operates in less than 1 min.This validates the efficiency and accuracy of the proposed method in processing large-scale,complex architectural group point clouds.关键词
点云/点云投影/直线拟合/建筑轮廓提取/自动重建Key words
point clouds/point cloud projection/line fitting/architectural contour extraction/automatic reconstruction分类
建筑与水利引用本文复制引用
张丰羽,孔庆钊,李培振..点云数据驱动的建筑群轮廓图快速生成方法[J].华中科技大学学报（自然科学版）,2025,53(8):24-29,6.基金项目
国家重点研发计划资助项目(2020YFC1512504). （2020YFC1512504）
