本文为美国乔治梅森大学（作者：Bahman Moghaddame-Jafari）的硕士论文，共73页。

目前，点云技术被用于桥梁和隧道的检测，在这些应用中需要远程测量结构的运动，如沉降和构件变形。目前测量结构变形的点云分析方法需要对点云数据进行网格化或直线/曲线拟合，这一步骤增加了整体测量的误差，属于不稳定计算。

本文提出了一种新的逐点取样法测量结构变形的方法，该方法不需要网格化或直线/曲线拟合，可以直接对点云进行分析。采用摄影测量和结构自运动（SfM）算法采集三维点云数据，利用云到云（C2C）距离测量和多尺度模型到模型的云比较（M3C2）算法计算了实际偏差。通过三点弯曲试验，对一系列铝试件进行了弯曲验证，采用了两种不同的统计方法来确定沿构件的挠度。结果表明，在垂直偏转测量中，最终结果可以达到亚毫米精度。此外，研究结果表明，该方法可作为局部更新结构有限元模型来考虑结构变形的工具。

Point cloud technology is now used ininspection of bridges and tunnels where it is desired to remotely measure thestructure’s movements such as settlements and member deformations. Currentmethods of point cloud analysis for measuring structural deflections requiremeshing or line/curve fitting to the point cloud data. This step adds an errorto the overall accuracy of this technology and is not computationally stable.This study presents a novel method in measuring structural deflections througha per-point sampling method in which the point cloud is directly analyzedwithout the need for meshing or line/curve fitting. The 3D point cloud data iscollected using Photogrammetry and the Structure-from-Motion (SfM) algorithm.The deflections are computed using Cloud-to-Cloud (C2C) distance measurement and Multiscale Model to ModelCloud Comparison (M3C2)algorithms. Through three-point flexural testing, series of aluminum specimenswere deflected for validation. Two different statistical methods were used to determinethe deflections along the member. The results indicate sub-millimeter accuracyxiii in measuring vertical deflection. Furthermore, results suggest that thistechnique can be a tool to update locally the structure’s Finite Element Modelto account for deformations in the structure.

1 引言

2 文献回顾

3 相关算法

4 实验分析

5 结论

附录 部分MATLAB代码

更多精彩文章请关注公众号：