基于图像特征识别的隧道三维重构方法研究及应用  

Research and application of 3D reconstruction method for tunnels based on image feature recognition

作　　者：柳厚祥 刘诗雨 LIU Houxiang;LIU Shiyu(College of Civil Engineering,Changsha University of Sciences&Technology,Changsha 410114,China)

机构地区：[1]长沙理工大学土木工程学院,湖南长沙410114

出　　处：《湖南文理学院学报(自然科学版)》2025年第1期79-84,共6页Journal of Hunan University of Arts and Science(Science and Technology)

基　　金：湖南省水利厅科技项目(XSKJ2019081-39);湖南省教育厅科学研究重点项目(19A025)。

摘　　要：隧道三维模型能直观的显示隧道围岩的裂隙节理信息,准确实现隧道的三维重构有助于获取隧道地质情况。本文拟采用Python语言,借助于开源数据库提供的相关函数,首先对拍摄的多幅隧道图像进行特征点识别,采用随机抽样一致性算法实现了相邻图像的正确匹配。其次采用张氏标定法求解出相机内参数,并通过单目视觉计算求解出外参矩阵,计算出由图像像素到三维点的变换,提取全部像素点得到三维稠密点云模型后,用德劳内三角划分对导出点云做网格生成处理形成隧道表面,再对三角网格面片做颜色平均值插入处理重现彩色隧道模型,并集成编译了三维重构相关算法。最后以湖南邵阳某隧道为例,完整地实现了隧道的重构,验证了基于图像特征识别实现三维重构的可行性。The three-dimensional tunnel model can visually display the fracture joint information of the surrounding rock of the tunnel,and the accurate realization of the three-dimensional reconstruction of the tunnel is helpful for obtaining the geological conditions of the tunnel.This paper intends to use Python language,with the help of the correlation function provided by the open-source database,to identify the feature points of multiple tunnel images,and realize the correct matching of adjacent images by random sampling consistency algorithm.Then,Zhang's calibration method is used to solve the internal parameters of the camera,and the external parameter matrix is solved by monocular vision calculation.The transformation from image pixels to three-dimensional points is calculated,and all pixels are extracted to obtain a three-dimensional dense point cloud model.After that,Delauney triangulation is used to mesh the derived point cloud to form a tunnel surface,and then the color average of the triangular mesh patch is inserted to reproduce the color tunnel model,and the three-dimensional reconstruction related algorithm is integrated and compiled.Finally,taking a tunnel in Shaoyang,Hunan Province as an example,the reconstruction of the tunnel is completely realized,and the feasibility of realizing three-dimensional reconstruction based on image feature recognition is verified.

关 键 词：图像特征点 单目视觉 三维点云 三维重构 

分 类 号：TU82[建筑科学]