Study On Measuring Technology For Surface Deformation Of Geotechnical System Based On Image Feature Matching

The safety of civil engineering is inseparable from the safety monitoring throughout the project.It is of great significance to study the progress of efficient and reliable engineering monitoring methods for engineering construction theory and national economic security.Computer vision measurement technology based on digital image technology and optical imaging principle has the advantages of high efficiency,real-time,accuracy and no damage.It is of great practical significance to apply visual measurement technology to civil engineering,especially geotechnical engineering.This paper summarizes the implementation process and algorithm classification of visual measurement methods,and analyzes the advantages and disadvantages of different measurement methods at present.Design an efficient image feature detection and matching combination algorithm,combine the combination algorithm with the finite element triangle element displacement and strain calculation method,and propose a displacement and strain solution method suitable for two-dimensional surface deformation;according to the principle of optical imaging,the combination algorithm is further It is applied to three-dimensional space displacement measurement,and a displacement measurement method suitable for three-dimensional space deformation is obtained,and multi-camera three-dimensional monitoring is realized.According to the above algorithm,Geo Surf Deform,a measurement program with an interactive interface was written in PYTHON.The main research work includes the following aspects:(1)Based on the feature detection algorithm(surf),matching algorithm(FLANN)and error elimination algorithm(PROSAC),a more efficient image feature matching combination algorithm is implemented which realizes the displacement monitoring of the region of interest in the engineering deformation image sequence.The accuracy and efficiency of the combined algorithm are verified by the displacement test of engineering materials.(2)Aiming at the problem of two-dimensional deformation,combined with Kriging interpolation method and Delaunay triangulation principle,the discrete image feature points are divided into finite element triangular mesh,so the finite element triangular element displacement and strain calculation method is applied to image two-dimensional surface displacement monitoring,and a two-dimensional displacement and strain calculation method based on image feature is implemented.The accuracy of this method is verified by measuring the same tensile test image with Ncorr software based on region matching.(3)Aiming at the problem of three-dimensional displacement,according to the different cameras,the monocular imaging principle and binocular imaging principle are combined with the image feature matching algorithm,and the monocular measurement method and binocular measurement method suitable for three-dimensional displacement measurement are proposed.The indoor retaining wall deformation simulation experiments are carried out to verify the reliability of the method.(4)The problem of visual occlusion in engineering is studied.The GICP algorithm is used to solve the conversion matrix of camera coordinate system,realize the threedimensional monitoring of the whole scene with multiple cameras,and reconstruct the engineering practice with Poisson surface reconstruction.