改进的包裹圆算法提取建筑立面轮廓  

Improved wrapped circle algorithm for extracting building facade edge points

作　　者：张睿 余代俊[1] ZHANG Rui;YU Daijun(College of Earth Sciences,Chengdu University of Technology,Chengdu 610059,China;GuangDong Architectural Design&Research Institute Co.,Ltd.,Guangzhou 510010,China)

机构地区：[1]成都理工大学地球科学学院,成都610059 [2]广东省建筑设计研究院有限公司,广州510010

出　　处：《时空信息学报》2023年第3期384-389,共6页JOURNAL OF SPATIO-TEMPORAL INFORMATION

摘　　要：针对包裹圆算法仅对密度均匀的点云有较好的适应性,当点的密度分布不均匀或点的密度未知时,包裹圆算法难以确定包裹圆半径及其内部点个数的阈值等问题,本文提出了改进的包裹圆算法,以方位角为基准将包裹圆划分为若干区域,以每个区域内是否都存在点为条件,判断各个点是否为边界点。经建筑物立面点云边缘点提取实验研究表明:改进算法提取边缘点效果好于包裹圆算法;当点云规模较大时,处理速度优于包裹圆算法,且提取三维点云时能更好地保留其三维特征。The existing bounding circle algorithm has proven effective when dealing with point clouds of uniform density.However,it faces certain limitations when applied to point clouds characterized by non-uniform or unknown density distributions.These limitations primarily stems from challenges in accurately determining the radius of the bounding circle and the threshold for the number of points within it.To overcome these challenges,this paper introduces an enhanced bounding circle algorithm designed to address these issues.The novel algorithm divides the bounding circle into multiple regions based on azimuth angles.It employs this division to assess whether a point qualifies as a boundary point by examining the presence of neighboring points in each region.This approach not only demonstrates superior efficiency compared to traditional algorithms but also extends its applicability to three-dimensional point clouds.In the context of experimental validation,where the objective is to extract edge points from point clouds representing building facades,the improved algorithm exhibits remarkable performance advantages over the conventional bounding circle algorithm.It becomes evident that the enhanced algorithm exces when dealing with large-scale point cloud datasets,primarily in terms of processing speed.It is important to note that this enhanced performance is achieved without compromising the representation of the point cloud’s three-dimensional characteristics.This enhancement in performance is of particular significant when working with point clouds that model complex structures or environments.In such scenarios,variations in point density are common,and the improved algorithm demonstrates its adaptability by accurately extracting edge points.These capabilities make it an invaluable tool in a wide array of fields,including architecture,robotics,and computer vision.In summary,the proposed improvements to the bounding circle algorithm represent a substantial advancement in the processing and analysis of point cloud data.

关 键 词：包裹圆算法 点云 建筑立面 边缘点提取 三维特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]