Research On The Technology Of Surface Reconstruction Baced On The Three Dimensional Point Cloud Data

With the rapid development of computer technology and increasing requirement of virtual reality application, more and more researchers have devoted to the study of three-dimensional reconstruction. The point cloud scanned by a 3-D Laser Scanner is Possessed of good properties for its easily obtain and simply storing features so that their inherent advantages in the expression of the three-dimensional geometry is obvious. Reconstruction processing technology based on three-dimensional point cloud is growing fast so new issues raises. For demand of higher accuracy and realism of the model reconstruction, the improvement of traditional reconstruction methods has great practical value.The paper introduces relative concepts about the scattered point cloud reconstruction which includes technology of obtaining and simplifying point cloud, and summarizes the common three-dimensional reconstruction algorithms. Mostly, improved algorithm for the reconstruction of traditional triangulation algorithm is proposed. The main work completed are :(1) To reduce the number of points, we simplify the point cloud from data file. The traditional k-nearest neighbor searching algorithm needs large amount of computation and is not suited in terms of massive spatial point cloud data, so using the grid-based k-nearest neighbor can speed up searching algorithm through grid divides the space of point cloud into separate small bounding box. Then, search each point’s k-nearest closed points by the bounding box owing to their neighboring connection in space positions. The method is more efficient compared to the traditional searching in whole space. In the next, simplify each grid’s points list by bounding box method.(2) To improve mesh frontier generation algorithm, we judge the minimal interior angle in the first triangle and add the criterion of dihedral angle based on projection in searching candidate point to triangular the simplified points. Compared to the traditional mesh frontier generation algorithm, the algorithm choose the first triangle which satisfy the condition of more than the minimal interior angle, then search candidate point along the boundary edge and projects communal adjacent points of extended edge to the tangent plane. Eligible points whose new triangle and extended dihedral having the common edge match a consistent standard in extending direction shall be maintained to ensure the smoothness of the surface being built. Thus, the reconstruction result avoids cross-cutting issue of finding candidate points in the three- dimensional space, so as to achieve the surface meshes of objects quickly and efficiently.(3) A surface reconstruction system based on three-dimensional point cloud is implemented. The system gets the data from point cloud files and divide the space by grids to establish each point’s neighborhood information, then simplify point cloud data by the bounding box algorithm to reduce the amount of data. In the next, we use improved method of searching candidate point based on projection in frontier generation algorithm to triangulate the points of being simplified and achieve satisfactory reconstruction results.Finally, we design and realize a three-dimensional point cloud reconstruction system basing on the algorithms which can turn point cloud into computer models with triangle meshes quickly and accurately. The system also demonstrate the feasibility and validity of the proposed algorithms.