基于单目视觉下的裂缝三向变化检测  

Three-dimensional Crack-Change Detection Based on Monocular Vision

作　　者：刘磊[1] 丁勇[1] 李登华[2,3] Liu Lei;Ding Yong;Li Denghua(School of Science,Nanjing University of Science and Technology,Nanjing 210094,Jiangsu,China;Nanjing Hydraulic Research Institute,Nanjing 210024,Jiangsu,China;Key Laboratory of Reservoir Dam Safety,Ministry of Water Resources,Nanjing 210024,Jiangsu,China)

机构地区：[1]南京理工大学物理学院,江苏南京210094 [2]南京水利科学研究院,江苏南京210024 [3]水利部水库大坝安全重点实验室,江苏南京210024

出　　处：《激光与光电子学进展》2024年第18期167-176,共10页Laser & Optoelectronics Progress

基　　金：国家重点研发计划(2022YFC3005502);国家自然科学基金(51979174);国家自然科学基金联合基金项目(U2040221);中央级公益性科研院所基本科研业务费专项资金(Y322008)。

摘　　要：单目视觉测量技术下存在二维图片深度信息缺失,导致无法快速直接测量出二维图像中三维坐标的问题。为在单目视觉下快速利用二维图像直接获取其三维坐标,提出了单目视觉的裂缝三维变化检测方法。根据裂缝变化等效模型设计了特制靶标,利用特征点进行Epnp(efficient perspective-n-point)求解,得到了多次拍摄图片时相机的相对位姿,利用最小二乘法还原其深度信息,通过迭代最近点算法进行坐标系转换,得出了实际的三维位移变化,绝对误差精度在0.5 mm以内,满足工程中对裂缝的检测要求。The depth information of two-dimensional(2D)images cannot be captured via monocular-vision measurement technology,consequently,the three-dimensional(3D)coordinates in the 2D image cannot be measured rapidly and directly.Hence,this paper proposes a 3D crack-change detection method based on monocular vision.Using an equivalent crack-change model,a special target was designed,feature points were used to solve the Epnp(efficient perspective-npoint),the relative pose of the camera during repeated photograph capturing was obtained,the depth information was restored using the least-squares method,and the actual 3D displacement change was obtained using the iterative closest point algorithm for coordinate-system conversion.The absolute-error accuracy was within 0.5 mm,which satisfies the requirements of cracks specified in engineering.

关 键 词：单目视觉 迭代最近点算法 三维重建 裂缝测量 

分 类 号：TV39[水利工程—水工结构工程]