Research On 3D Reconstruction Image Processing Technology Based On Depth Information

With the rapid development of computer graphics and computer vision,as one of the research hot spots,3D reconstruction technology has been used in the reconstruction of diseased organs for medical treatment,reconstruction of surgical procedures,and the protection of cultural relics after the excavation widely.Because of the rational price,Kinect can acquire RGB images and depth images simultaneously,so research on multi-dimensional reconstruction using Kinect has become a trend at present.This paper takes the depth information obtained by Kinect as background,then study the image processing technology in the three-dimensional reconstruction process.Researches on camera calibration,registration algorithm optimization and other technologies in the three-dimensional reconstruction process.The main works of this article as follows.For the problem that the accuracy of the calibration of the camera is too large with the physical checkerboard,the influence of the distance between the camera and the calibration object on the calibration accuracy is analyzed,and a camera calibration method based on a virtual chessboard is proposed.Comparing with the classic Zhang's calibration method,the re-projection error is used as the evaluation basis.The experimental results show that the camera calibration error based on the virtual chessboard is reduced by 30%.Aiming at the problem of low efficiency and low precision in the traditional reconstruction method,a patch-based subarea registration optimization method was proposed.Compared with the traditional registration algorithm,the experimental results show that this method can effectively improve the accuracy and rapidity of point cloud registration.The registration time is reduced to around 30 seconds,and the root mean square error is used as the evaluation index.Registration error It has been reduced by 27.5%.