基于点云数据的道路面建模优化算法研究  

Study on algorithm of road surface model optimization overpoint cloud data

作　　者：赵婧文 Zhao Jingwen(Shanghai Surveying and Mapping Institute,Shanghai 200063,China;Key Laboratory of Spatial-Temporal Big Data Analysis and Application of Natural Resources in Megacities,MNR,Shanghai 200063,China)

机构地区：[1]上海市测绘院,上海200063 [2]自然资源部超大城市自然资源时空大数据分析应用重点实验室,上海200063

出　　处：《工程勘察》2025年第1期54-58,共5页Geotechnical Investigation & Surveying

基　　金：上海市2021年度“科技创新行动计划”社会发展科技攻关项目(21DZ1204100).

摘　　要：针对难以从车载激光雷达数据中提取高精度道路路面不规则三角网的问题,本文提出一种在道路坐标系通过卷积滤波进行不规则三角网优化和建模的方法。该方法在道路坐标系进行卷积运算,可避免传统滤波方法无法解决的城市道路点云稀疏分布和道路路面特征不一致性等问题,结合占据格三维卷积方法对三角网种子点进行可靠性筛选和高程优化,最终实现道路路面不规则三角网构建。通过实验对比可知,本文方法在各种形态道路上构建的不规则三角网的投影误差稳定,且均优于其他方法,可为城市实景三维建模提供高精度的道路面DTM数据,也能够为激光雷达点云配准和更新提供路面法向量基准。In the light of difficulties to generate highly accurate TIN from vehicle borne LiDAR data,this paper proposed a method to optimize the irregular triangulation and build model in the road coordinate system through convolution filtering.This method uses the convolution operation in the road coordinate system to avoid the sparse distribution of urban road point clouds and the inconsistency of road surface characteristics that cannot be solved by traditional filtering methods.Combining with the occupying grid 3D convolution method,the reliability screening and elevation optimization of the triangulation seed points are carried out,and finally the construction of road surface Triangulated Irregular Network(TIN)is realized.The comparison experiment proved that the projection errors of the TINs constructed by this method on various roads are stable and better than other methods.This method can provide high-precision road surface DTM(Digital Terrain Model)for urban 3D real scene modeling,and road surface normal vector reference for the registration and update of LiDAR point clouds.

关 键 词：激光雷达点云 道路面建模 三维滤波器 高度估计 不规则三角网 

分 类 号：P237[天文地球—摄影测量与遥感]