基于尺度不变特征变换算法(SIFT)立体匹配的结构三维振动位移监测  

3D Vibration Displacement Monitoring of Structures Base on SIFT Stereo Matching

作　　者：杜永峰[1,2] 马天军 韩博 李虎 DU Yongfeng;MA Tianjun;HAN Bo;LI Hu(Institute of Earthquake Protection and Disaster Mitigation,Lanzhou University of Technology,Lanzhou 730050,China;International Research Base of Seismic Mitigation and Isolation of Gansu Province,Lanzhou University of Technology,Lanzhou 730050,China)

机构地区：[1]兰州理工大学防震减灾研究所,兰州730050 [2]兰州理工大学土木工程减震隔震技术研发甘肃省国际科技合作基地,兰州730050

出　　处：《工业建筑》2023年第12期105-112,134,共9页Industrial Construction

基　　金：国家自然科学基金项目(52178291)。

摘　　要：为解决传统单目视觉位移测量方法无法获得结构三维运动信息的不足,将双目立体视觉应用于结构的三维位移监测中。使用Python语言,在Pycharm平台上进行算法编译,采用Shi-Tomasi角点检测算法与尺度不变特征变换算法相结合,实现左右两幅图像特征点的立体匹配。通过图像预处理,提取匹配的目标特征点坐标,进而实现结构三维位移监测,并将该方法应用在五层框架模型的振动台试验中,获得地震动作用下结构的三维位移时程。试验结果表明:该方法测得的位移时程曲线与拉线式位移计数据吻合较好,Z向最大峰值位移相差1.29 mm,误差绝对值在8%以内,在频域也有较好的表现。并通过改变基线距离和相机的偏转角度来验证该方法的鲁棒性,表明了该方法用作结构三维位移全过程监测的可行性。In order to solve the shortage of traditional monocular vision displacement measurement methods that cannot obtain 3D motion information of structures,binocular stereo vision was applied to the 3D displacement monitoring of structures.Using Python language,the algorithm was compiled on Pycharm platform,and Shi-Tomasi corner point detection algorithm was combined with scale invariant feature transformation algorithm to achieve stereo matching of feature points of left and right images.Through image pre-processing,the coordinates of the matched target feature points were extracted,and then the three-dimensional displacement monitoring of the structure was realized,and the method was applied to the shaking table test of a five-story frame model to obtain the three-dimensional displacement time history of the structure under the action of ground shaking.The test results showed that the displacement time history curve measured by the method matched well with the data of the pull-wire displacement meter,and the maximum peak displacement in Z-direction differed by 1.29 mm with the absolute value of error within 8%,and also had a good performance in the frequency domain.Besides,the robustness of the method was verified by changing the baseline distance and the deflection angle of the camera,which showed the feasibility of the method used as the whole process monitoring of the 3D displacement of the structure.

关 键 词：尺度不变特征变换 双目视觉 特征点匹配 三维位移监测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]