A 3D Reconstruction System Based On Dual Line-Scan Cameras

With the level of science and technology overcoming one after another difficulties,computer vision has developed rapidly,and three-dimensional(3D)reconstruction based on computer vision has gradually been favored by scholar.In recent years,3D reconstruction technology has attracted more and more scholars,realizing its important value in different fields and gradually aligning with life.In this paper,a 3D reconstruction system based on dual Line-Scan cameras is built,and the problems of calibration of two-line array camera and image stereo matching are deeply studied.The main research contents of this paper are listed below:According to the stereoscopic imaging structure of the dual Line-Scan cameras and its 3D imaging principle,the 3D reconstruction system is built by considering the hardware selection from various aspects.The main functions of hardware acquisition are realized.Through the projection mapping model of the area array camera,the mapping relationship between the Line-Scan camera and the spatial point is derived,and the distortion model of the line-Scan camera is obtained.In the calibration process of the dual Line-Scan cameras,the influence of the external parameters of the camera on the calibration equation is analyzed,and the calibration equation of the dual Line-Scan cameras is obtained,which lays a foundation for the 3D reconstruction.For the stereo matching of images,this paper studies corresponding algorithms based on region matching,matching images by using Normalized Cross Correlation(NCC)algorithm that is insensitive to light and dark variations.Then,according to the phenomenon that the NCC algorithm is disorganized in the complex background,the method of guiding the matching of low-authenticity regions from the matching region with high authenticity is proposed,which makes the chaotic parallax image smoother.Finally,bicubic interpolation and median filter are carried out on the smoothed parallax image to improve the matching effect.Finally,a 3D reconstruction system based on dual Line-Scan cameras is built in this paper,and the calibration of the system is completed by using a high-precision checkerboard calibration board.Moreover,the 3d surface reconstruction of the objectis realized through the proposed matching algorithm and post-processing on Matlab software,which verifies the feasibility of the method presented in this paper.