Research On Monocular Structure Light 3D Reconstruction Method Based On Gray Code Coding

With the in-depth study of computer image vision and multi-view geometry,3D measurement technology has been widely used in the fields of automatic driving technology,drone navigation,logistics sorting,industrial automation,robot scene map construction and so on.The monocular structured light system differs from the traditional binocular vision system in that it replaces one of the cameras with a projector to avoid the most problematic stereo matching.As long as the projector is accurately calibrated,a larger viewing angle and efficient 3D measurement reconstruction can be achieved while reducing its cost.System calibration.The pinhole imaging model of the camera,the transformation between different defined coordinate systems and the calibration method of the camera are introduced.Aiming at several key problems in monocular structured light measurement system based on Gray code coding: projector calibration,Gray code coding and decoding,and point cloud automatic splicing,this paper has carried out in-depth research and proposed corresponding solutions.Projector calibration.By using the same imaging principle of the projector and the camera,a projector calibration plate with a checkerboard is designed,and the single-shot camera can simultaneously acquire the projection pattern and the checkerboard image on the calibration plate.The checkerboard image determines the plane equation of the calibration plate.The projection pattern determines the intersection of the camera light and the plane of the calibration plate,thereby calculating the spatial position of the projection point and realizing the calibration of the projector.Gray code encoding and decoding.The Gray code coding in the time coding technique has several advantages over the binary coding,which only one of the two adjacent codes has a different binary number,and the maximum value and the minimum value are only one bit different,and the resolution of the coding pattern is improved.The Gray code coding technique can be used to accurately restore the projection point on the corresponding projectoraccording to the point of the camera imaging plane.In this paper,there are two main innovations in designing the coding pattern.First,firstly,a black-white solid color pattern is projected onto the surface of the object to be tested,and the shadow portion is used to remove the pixel value in the two projection patterns.The shaded area is used to reduce the calculation amount and the correlation error.Secondly,in order to get the correct image gray threshold,a corresponding image with the same number of stripes but the black and white stripe position is reversed for each level of coding pattern.With such an image,gray scale determination can be efficiently performed with the original image.Point cloud automatic registration.According to the degree of change of the point cloud normal vector,the feature points of the multi-view cloud are selected,and then the histogram of the FPFH feature descriptor is established for each feature point,and the appropriate matching point pairs are initially selected.In order to verify the correctness of the matching point pairs,the Euclidean space distance invariance and the RANSAC algorithm are used for further screening.Finally,the four-element method and SVD are used to solve the initial rigid body transformation matrix to complete the coarse registration of the point cloud.Finally,the relevant experiments of the measurement system are carried out.The acquired point cloud data is evaluated by three indicators of flatness,orthogonality and accuracy.The results show that the measurement system has high robustness and small distortion.