A Surface Reconstruction Of Point Clouds Based On Spacial Status Octree

In the field of computer graphics, a lot of research has been done in striving to physically realistic simulations. Reconstructing the point cloud data which is from 3d scanner to mesh model is used widely in virtual museum, virtual battle, virtual surgery, games, movies and so on. Because 3d scanner get point cloud data with information loss, holes, noise, etc. the surface reconstruction, which is efficient and quality, brings some difficulties. Therefore, how to deal with the point cloud with noise and information loss to mesh is currently very hot in graphics and virtual reality.The normal is very important in surface construction process and there is always a point cloud without normal information. Some surface reconstruction algorithms such as Wavelet, Poisson, are highly resilient to data noise. But they require the input data with normal information. Others, such as Cocone or Layer Peeling, can process the data without normal information. But they are difficult to deal with point cloud data with holes. Layer Peeling is very complex and hard to realize.In this thesis, we propose a novel surface construction algorithm based on octree, which can deal with the point cloud with holes or without normal information. We can get the status in corner of cell and calculate the every point normal.The innovation of our algorithm are as follows:1. This algorithm can handle the point cloud without the normal information, and the point cloud even has holes. Experiments prove the reconstruction result to be okay.2. Our algorithm can also handle the non-uniform sampling point cloud. It can effectively avoid the errors in the reconstruction result for the non-uniform sampling.