现代建筑物点云平面特征识别方法  

The Plane Feature Recognition Method for Modern Building Point Cloud

作　　者：王新静[1] 段晨鑫 姚怡烨 WANG Xinjing;DUAN Chenxin;YAO Yiye(North China University of Water Resources and Electric Power,Zhengzhou 450046,China)

机构地区：[1]华北水利水电大学,河南郑州450046

出　　处：《河南科技》2024年第4期4-8,共5页Henan Science and Technology

摘　　要：【目的】基于现代建筑物点云数据面片特征,提出一种基于随机抽样一致算法的平面分割识别方法。【方法】该方法先利用三维格网划分来建立空间格网单元,再根据随机采样点来确定局部格网单元,通过随机机制来拟合平面模型,经过局部打分来确定候选模型集,利用法向约束和共面分割来解决过分割和欠分割的问题。【结果】采用该方法可获取当前最优模型和一致集,并完成点云分割。【结论】试验结果表明,该方法能对富有平面特征的建筑物进行有效分割。[Purposes]According to the patch characteristics of modern building point cloud data,a plane segmentation recognition method based on random sampling consistent algorithm is proposed in this paper.[Methods]In this method,the spatial grid element is established by three-dimensional grid division,and then the local grid element is determined according to the random sampling points,the plane model is fitted by random mechanism,and the candidate model set is determined by local scoring.Normal constraints and coplanar segmentation are used to solve the problems of over-segmentation and insufficient segmentation.[Findings]Finally,the current optimal model and consistent set are obtained,and the point cloud segmentation can be completed.[Conclusions]The experimental results show that this method can segment the buildings with plane features effectively.

关 键 词：点云 分割 局部采样 一致集 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]