基于三维点云的大坝缺陷体积测量方法  

Volume Measurement Method for Dam Defects Based on 3D Point Cloud

作　　者：张昱恒 徐晓龙[1] 江雅馨 魏瑞凯 ZHANG Yuheng;XU Xiaolong;JIANG Yaxin;WEI Ruikai(School of Information Science and Engineering,Hohai University,Changzhou 213200,China)

机构地区：[1]河海大学信息科学与工程学院,江苏常州213200

出　　处：《计算机测量与控制》2025年第1期78-84,共7页Computer Measurement &Control

基　　金：国家重点研发计划(2022YFB4703404);中国建筑股份有限公司2023年度科技研发课题(CSCEC-2023-Z-10)。

摘　　要：大坝在运行过程中不可避免地会出现混凝土呈块状剥落的情况;因此,精确地剥落缺陷体积测量对于其结构修缮具有重要意义,而缺陷的形状往往是不规则的,难以通过简单的几何运算求出;为实现非接触精准测量,提出一种将点云平面拟合、滤波和三角化相结合的缺陷体积测量方法;利用单目相机拍摄待测结构物,对其进行多视图三维重建,获取点云数据,分割出缺陷点云并对其进行体素降采样,采用Delaunay三角剖分法计算缺陷体积;为提高测量精度,改良传统RANSAC平面拟合算法,在其中融入统计离群值移除,实现缺陷区域的精准分离;经过多次实验证明,该方法在测量缺陷时,无论形状是否规则,都能较为精准地测量出其真实体积;将测量结果与仅用RANSAC算法分割相比,测量精度提高70.32%;该方法大幅度提升了缺陷的测量精度。During the operation of dams,it is inevitable that concrete spalling will occur.Therefore,it is of great significance to accurately measure the volume of spalling defects for effective structural rehabilitation.Defects are often irregular and difficult to calculate through simple geometric means.To achieve non-contact and precise measurement,this paper proposes a defect volume measurement method for integrating with the point cloud plane fitting,filtering,and triangulation.A monocular camera is used to capture the images of the structure to be tested,and obtain point cloud data through multi-view 3D reconstruction.The defect point cloud is segmented and downsampled using the voxel-based method,and the Delaunay triangulation algorithm is applied to calculate the defect volume.To enhance measurement accuracy,the traditional random sample consensus(RANSAC)plane fitting algorithm is improved by incorporating the statistical outlier removal(SOR),achieving the precise separation of the defect region.Multiple experiments have proven that regardless of whether the shape is regular or not,,this method can accurately measures the true volume of defects,improves the measurement accuracy by 70.32%than the RANSAC algorithm,and significantly enhances the precision of defect volume measurement.

关 键 词：多视图三维重建 RANSAC算法 统计离群值移除 体素降采样 DELAUNAY三角剖分 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]