Research And Application Of 3d Reconstruction System Based On Multiline Structured Light

With the continuous and rapid development of science and technology,non-contact threedimensional measurement technology is increasingly favored by everyone.Among them,3D reconstruction based on multiline structured light is an indispensable reconstruction method.This paper takes the 3D reconstruction technology of multiline structured light as the research object,and focuses on the key technologies such as system calibration,3D reconstruction of multiline structured light,and the final point cloud stitching,which are verified by experiments.The main research content of this article is as follows:Firstly,a new optical plane calibration method based on multi line laser contour classification is proposed to address the issues of difficult calibration of multiple lines and low accuracy in extracting feature points in the optical plane calibration process of existing 3D measurement systems.Firstly,multi line structured light is classified based on contour judgment,which can be converted into a single line calibration problem;Based on the camera imaging model,the corresponding relationship matrix between the camera and the calibration board can be obtained,and then the three-dimensional coordinates in the camera coordinate system can be obtained.This process does not need to project the line structured light onto the calibration plate,which can solve the problem that the feature points cannot be accurately extracted due to the interference of laser lines,improve the accuracy and stability of calibration,and reduce the complexity of light plane calibration.Secondly,in the phase of center extraction of online structured light,a laser centerline extraction algorithm based on normal guidance is proposed.Divide the angle eight neighborhoods at the laser center point,and search for an effective point set through normal angle guidance.In the phase of binocular multi line structured light matching,aiming at the situation that one to many matching of binocular multi line structured light causes errors,this paper combines the spatial position and attitude relationship of binocular cameras,uses the consistency of light path,and uses re projection and reverse re projection to reduce the error matching;By utilizing the method of laser segment matching again,a very small number of matching incorrect laser point pairs are eliminated,ensuring the reliability of the overall laser segment matching and improving the matching accuracy and reconstruction effect.Then,during point cloud stitching,in order to improve the accuracy of double target point matching,the circumference and circularity characteristics of manually marked points can be used to accurately extract the center coordinates,and then the angle difference method is used to ensure unique matching.By combining the invariance of spatial Euclidean distance under rigid body transformation,common marker point matching pairs can be obtained under different fields of view.This method is validated using the congruent triangle method,which can quickly align point clouds and obtain a complete point cloud model.Finally,a binocular multi line structured light 3D scanning system is built,and the operation interface of data acquisition and reconstruction is designed and compiled.To verify the accuracy of the 3D scanning system,high-precision standard blocks with known sizes were scanned and reconstructed,and the reconstruction accuracy of this system was tested.The experiment showed that the average error of this system was 0.03 mm.Other objects were also reconstructed to verify the usability of the scanning system.