点云数据三维网格构造方法研究-系统分析与集成专业论文.docx

I I 摘 要 从曲面的三维采样点集恢复出曲面的几何模型称之为曲面重建。曲面重建是许 多研究领域如逆向工程，医学图像可视化中的重要问题，也是研究的热点。 本文主要研究无序点云的三维重建，在研究点云重建理论的基础上，对以下问 题进行了深入探讨和分析：首先为了保证重建的效果，本文对点云中的噪声点去除 问题进行了研究，先分析了现有点云去噪的算法，然后在讨论了如何对无序点云建 立拓扑关系的基础上，采用了基于 k-d tree 的去噪算法实现无序点云去噪。其次，本 文研究了有理论证明的基于 Voronoi 图的 Power Crust 算法。该算法不管是对均匀采 样，还是非均匀采样的无序点云，都有较好的重建结果。但是在重建大数量级的点 云时速度过慢。针对这一问题，本文提出了一种基于 Power Crust 算法的快速点云重 建算法。在曲面重建前，根据采样点的局部特征尺度对原始采样点进行不均匀降采 样，在保证采样点满足重建条件的前提下，减少参与运算的采样点，从而减小重建 算法的计算量，大大提高了重建的速度。为了让重建后的模型表面更加光滑，接近 原始模型，本文还介绍了拉普拉斯平滑算法。最后，结合 VC6.0 和 VTK，采用本文 提出的快速的 Power Crust 算法和 Hoppe 的零集法，开发了一个点云重建系统，同时 对网格进行了平滑处理，可以根据点云数量和采样特点选择不同的重建方法。 关键词：点云;三维重建;去噪;快速 Power Crust;VTK II II Abstract Reconstruction the geometry model from the 3D sampling point set is called surface reconstruction. Surface reconstruction is the key point in many fields such as Reverse Engineering, Visualization of medical image, and it is also the hotspot. This paper focuses on surface reconstruction for unorganized points. On the basis of studying the theory of the reconstruction of point cloud, this paper has carried out thorough discussion and analysis on the following parts: Firstly, for the effect of reconstruction, noise points in point cloud has been studied. we analyses the current algorithms of removing noise points ,then adopts an algorithm based on k-d tree to removing noise points based on discussing how to establish the topology of unorganized point cloud . Secondly, we introduce the principle of Power Crust algorithm. The algorithm carries out good results for either uniformly or non-uniformly sample, but it runs for a long time when a magnitude point cloud is reconstructed. In order to improve its speed, a fast surface reconstruction algorithm based on Power Crust is proposed, which has improved the reconstructing speed greatly. Before reconstruction ,non-uniformly samples the point cloud according to the local feature size .With the guarantee that the remaining point cloud are sufficient to reconstruction ,the amount of points