Research On Multi-view 3D Reconstruction And Stitching Method Based On Line Structured Light

With the continuous development of industrial automation,the requirements for the quality of industrial products are becoming more and more stringent,and the requirements for the accuracy of measuring equipment are also getting higher and higher.Among them,the three-dimensional measurement system based on line structured light has a wide range of applications in product inspection and reverse engineering due to its concise structure,non-contact,and high-precision characteristics.However,traditional monocular or binocular vision has certain limitations in specific scenes.Therefore,a multi-view 3D measurement system composed of multiple cameras can be used to solve problems such as insufficient field of view and missing point clouds caused by foreign objects.In this paper,a multi-view 3D reconstruction system is designed and implemented based on the principle of line structured light measurement,and the problems existing in the process of line structured light measurement are studied in detail.First,the Steger algorithm is selected as the centerline extraction algorithm of the measurement system built in this paper.Aiming at the low efficiency of centerline extraction caused by the multiple convolution operations of the Steger algorithm,combined with the characteristics of the slender laser stripes,a method based on connected regions is proposed.The method of dividing the ROI area of the line structured light stripe with the circumscribed rectangular aspect ratio,and then running the Steger algorithm on the segmented line structured light area.Secondly,in view of the low calibration accuracy of the monocular line structured light measurement system,the point cloud of the laser fringe is reconstructed by the method of binocular epipolar line constraint,which improves the accuracy of the optical plane calibration.Aiming at the lack of point cloud caused by the obstruction of the linear structured light strips during the visual measurement process,a dual-camera measurement system is proposed for measurement,and the accuracy and effectiveness of the dual-camera measurement system are evaluated.Finally,for multi-view 3D reconstruction,this article first uses a turntable to calibrate the relationship between the four sets of dual-camera measurement sensor coordinate systems,and for the four-view point cloud stitching and layering phenomenon,the system is secondarily calibrated with the help of cylindrical parts.By fitting the cylindrical point cloud of each viewing angle,the cylindrical axis parameters of each viewing angle are obtained,and the translation relationship between the base point of the other three sets of sensor fitting results and the base point of the main sensor1 is calculated to complete the secondary calibration of the system.Perform measurement to complete the accuracy evaluation of the multi-view measurement system.The experimental results show that compared with the extraction process of the Steger algorithm before the improvement,the average extraction speed of the improved algorithm is increased by 4.5 times.Compared with the traditional monocular measurement method and binocular matching method,the dual-camera measurement sensor built in this article can effectively solve the field of view occlusion problem in the visual measurement process.Among them,the binocular stereo calibration error is0.027 mm,and the dual-camera measurement sensor The average measurement accuracy can re ach 0.076 mm.The multi-view 3D reconstruction system built in this paper is within the effective measurement range of 0.9m?1.2m,and the overall stitching accuracy can reach 0.552 mm.The method studied in this paper can be applied to human3 D reconstruction and industrial online 3D detection.