Design And Implementation Of A Multi View 3D Reconstruction System

3D reconstruction is a technology of reverse engineering that integrates computer graphics,image processing,computer vision and pattern recognition.That technology has been widely used in social life,including robot vision,medical imaging,virtual reality and so on.The three-dimensional reconstruction technique is aimed to recover the 3D structure of real scene from the two from a two dimensional image series obtained from a conventional camera.In this process,at first,we need to compute the interior and external camera parameters by extracting and matching of the images.Next,generate the dense point cloud of the target scene through dense matching based on epipolar geometry.Then,reconstruct the three-dimensional polygon models by triangulating the dense point cloud.Finally,the 2D images are back projected on the polygon models to obtain high realistic 3D models.To realize the 3D reconstruction image-based system,it is necessary to master the basic theory of 3D reconstruction of single hole camera model,camera calibration and epipolar geometry.In the feature extraction and matching process,this paper uses the SIFT(scale-invariant feature transform)feature points and the normalized cross correlation algorithm,also the RANSAC algorithm is used to optimize the feature matching and improve the accuracy.After the camera parameters are obtained,we use the PMVS(patch-based multi-view stereo)algorithm to generate the dense point cloud,which is very good to deal with poorly-textured scenes.The Power Crust triangulation algorithm is used for the dense point cloud triangulation.In the final texture mapping process,MeshLab is used,which is an excellent open source interactive mesh processing tool.After the 3D point cloud model is obtained,the results obtained from the image based 3D reconstruction system are evaluated by comparing the results with the scanner scans,Finally,the results of the three dimensional reconstruction system are quantitatively analyzed by comparing with the results obtained by the laser scanner.The results show that the accuracy of the 3D reconstruction system has reached the application level.