Research On Binocular Camera Calibration Technology Based On 3D Point Cloud Reconstruction

In recent years,3D optical measurement technology and 3D reconstruction technology have shown great application potential in improving industrial production efficiency and improving people’s living standards.These two technologies have applications in workpiece detection,smart home,new energy vehicles,military radar,cultural relics digitization,industrial assembly line,medical beauty,and other aspects.The three-dimensional reconstruction technology based on binocular digital stripe structured light is an important implementation of three-dimensional vision technology,which has the advantages of high reconstruction accuracy and non-contact work.The general steps of 3D reconstruction are to calibrate the binocular camera first,project digital stripe structured light to the object under test,mark the feature information of the object,and finally realize the 3D reconstruction of the object under test.In the experimental environment,due to the diversity of lighting and the complexity of the field environment,the effect and stability of the 3D reconstruction may be greatly affected.It may also be due to the material problem of the measured object,which is overexposed when the camera collects data information,destroying the stripe-structured light information in the image,and resulting in an unsatisfactory effect of three-dimensional reconstruction.Firstly,a camera calibration algorithm based on multiple constraints is proposed to solve the ghosting problem of the 3D reconstruction in binocular systems,so as to improve the accuracy of the 3D reconstruction point cloud.During camera calibration,the internal and external parameters of the camera are obtained by calculating and optimizing the reprojection error under the two-dimensional coordinate system.However,in the 3D optical measurement system,the reprojection error is largly due to inaccurate camera calibration,and the 3D reconstruction accuracy of the object is also affected.In order to improve the accuracy of the 3D reconstruction,a camera calibration method based on multiple constraints is proposed,which can calculate more accurate reprojection error under the 3D coordinate system,so as to improve the accuracy of 3D reconstruction.In the3 D reconstruction experiment,the camera calibration method based on multiple constraints is easy to operate,the calibration results are accurate,and the accuracy of 3D reconstruction is significantly improved.And during the calibration process,there are no strict requirements for the position of the calibration plate.Secondly,a method of digital stripe structured light combined with gradient phase based on Gray code assisted phase shift is proposed to match the points with the same name,so as to improve the stability of three-dimensional reconstruction in binocular vision and solve the problem that due to phase jump or structured light jitter,the left and right cameras unfold phase is not unique,resulting in unstable three-dimensional reconstruction effect.The algorithm greatly improves the accuracy of matching points with the same name,ensures the uniqueness of the phase,makes the three-dimensional imaging effect not disordered,and ensures the stability of the three-dimensional reconstruction.Finally,the technology of single and binocular point cloud fusion is proposed,which solves the problem that the left and right cameras in the binocular system cannot find the phase information and three-dimensional imaging effect of the same name point due to reflection.In the monocular point cloud reconstruction,the line-structured light and surface-structured light model are combined,and the point cloud is reconstructed by using the method of light plane interpolation.The algorithm is simple,occupies less memory,runs fast,and has high feasibility and practicality.The point cloud integration of monocular reconstruction technology and binocular reconstruction technology realizes the compensation of point cloud,which improves the accuracy,stability,and integrity of three-dimensional reconstruction of objects in different scenarios and different materials,which can be widely used in workpiece detection,cultural relics digitization,intelligent cars and other aspects.