Research On 3D Surface Reconstruction Based On Surface Structured Light

With the continuous development of computer science,communication technology and many other disciplines and three-dimensional scanning technology,people began to use three-dimensional reconstruction technology to meet their own needs for the modeling of complex surface objects.Compared with traditional reconstruction techniques,point cloud three-dimensional reconstruction has the advantages of high accuracy,low cost,and short development cycle.Based on this,three-dimensional surface reconstruction is widely used in the fields of geology,robotics,CAD/CAM,medical care,urban management,and cultural relics protection.The three-dimensional reconstruction process mainly includes point cloud data acquisition,point cloud filtering,point cloud topology relations,point cloud triangulation grid and de-surface construction.For the scheme of point cloud acquisition,this paper chooses surface structured light to acquire the point cloud for accuracy.In this paper,the principle of dissolving the phase of the surface structure is studied in depth,and the phase-out effect of the four-step phase-shifting method is explored.For the four-phase phase-shifting solution,the phenomenon of the pull line appears,a multi-frequency heterodyne method is used to obtain a relatively good solution phase effect.After completion of the dephasing,phase-to-depth conversion is achieved by camera-projector calibration.In order to obtain more accurate point cloud data,the hardware platform of the small black studio and ace camera + acer blue projector is built in this paper.After obtaining the point cloud,the point cloud filter can be used to remove the pointcloud background.In the establishment of topological relations of point clouds,this paper introduces the construction methods of kd-tree,octree,and space bounding boxes.The kd-tree is used in the three-dimensional reconstruction based on the mapping method and the octree is used in the Poisson reconstruction.In the reconstruction of three-dimensional surfaces,this paper implements Poisson reconstruction based on implicit functions and fast Delaunay triangular surface construction based on mapping methods.The key to the realization of Poisson reconstruction lies inextracting the indicator function that describes the object model.The process mainly consists of constructing an octree to store data,setting up a spatial function,solving a vector field,solving a Poisson square to obtain a pointing function,and extracting an isosurface.Surface construction based on the mapping method mainly includes local tangent plane projection based on k-adjacency,local Delaunay triangulation,triangle grid splicing and surface fitting.Delaunay triangulation is an extremely important part of this method.This article focuses it on the analysis in chapter 4,which includes the analysis of point-by-point interpolation,segmentation and merger methods,and recursive growth method.The point-by-point insertion optimization algorithm Bowyer-Watson algorithm triangulates.Among them,the surface reconstruction based on the mapping method has better results.In order to avoid hole phenomenon for the fast Delaunay triangulation based on the mapping method,this paper uses the hole filling algorithm and has a good filling effect.In the aspect of personalized surface exploration of model surface,this article uses the texture mapping method to make the texture pattern be attached to the surface,which make the model surface have a very vivid surface effect.At the end,this article reviews the research and work of the project,having madethe conclusion and outlook.