Research Of 3D Reconstruction System Based On Coded-structured Light And Binocular Vision

In recent years,with the continuous development of science and technology and automation level unceasing enhancement,the industrial testing,automobile manufacturing,biomedical,reverse engineering and other fields of an object threedimensional topography measurement put forward higher request,rapidly and exactly how to get the object on the surface of the 3D model has become an urgent problem in the field of computer vision.This paper studies the structure light and combination of binocular vision,3D reconstruction system can be fast accurate three-dimensional point cloud data of the workpiece surface finish 3D measurement of workpiece.First of all,this paper set up the structure light binocular vision 3D reconstruction system hardware platform,and develops the corresponding software and hardware system module,in order to achieve the effect of human-computer interaction,design the visual interface for the software.On the basis of insight into the camera imaging principle,the method based on moving plane template to calibrate camera,obtained the related parameters in the visual system.Through calibration results,distortion correction and extreme line correction of the image obtained are obtained.After an in-depth analysis of the existing structure optical encoding scheme,the method of combining the gray code pattern and the linear moving pattern is used to encode the light.In order to accurately extract the edge information of black and white stripes on the image,this paper takes the method of projecting a positive inverse pattern.In image processing,by comparing the code design and radix-minus-one complement the size of the pattern in the same pixel gray value,to reset the point grey value,then the points displayed in the form of design,get the binary image.Compared with the common twovalued method,the disturbance introduced in the outside environment is eliminated,and the precision of the edge extraction is greatly improved.Aimed at the problem of low efficiency of matching,this paper collected a pure background image and a take a test the patterns of the target,through the two patterns are poor effective measurement area,makes the search area of the matching point greatly small,matching efficiency increased by 30%.In order to improve the matching accuracy,this paper proposes a optimization algorithm based on gray level similarity matching,to search on the transverse and longitudinal constraints to the optimized matching point.This method in according to the constraint conditions within the neighborhood search to match point to search again,according to the gray similarity principle to select the optimal matching point.In after the completion of the match,to measure the dead zone and on the surface of the object matching parallax fill point of failure,and then calculated according to the principle of triangulation point of three-dimensional coordinates,generate 3D point clouds,completing the process of 3D reconstruction.The accuracy of 3D reconstruction of the system is analyzed,on the distance of 0.5 meter the relative error of physical dimension is 1.75%,and the average deviation error is 1.86 mm,the max deviation error is 3.84 mm...