在一些模拟程序我们生成对象表面在点方面，每个点具有三维坐标，并且重新$ P $矢量psents的垂直于表面在该点处。可视化的目的，我们希望产生一个三角形组成的网格;每三贴近点形成一个三角形正常。然后我们就可以将这些信息发送给呈现象VMD（视觉分子动力学）表面一些标准的可视化程序。

In some simulation program we generate object surfaces in terms of points, each point has 3D coordinates and the vector that represents the normal to the surface at that point. For visualization purposes we would like to generate a mesh composed of triangles; each three close points form one triangle with its normal. Then we can send this information to some standard visualization programs that render the surface like VMD (Visual Molecular Dynamics).

我们不知道这是最快的/可算法用于执行此操作。

We wonder which is the fastest/available algorithm for doing this.

推荐答案

看看乔纳森Shewchuk 的工作，特别是对他（以及他的同事们）著名的论文和实现：

Take a look at Jonathan Shewchuk's work, especially on his (together with his colleagues) famous papers and implementations of:

Delaunay三角的流计算

一个二维质量网格生成和德劳 Triangulator

A Two-Dimensional Quality Mesh Generator and Delaunay Triangulator

也有快速实现的点云库（PCL）中实现无序点云。检查他们的presentation在的无序点云的快速三角。

There is also fast implementation of unsorted point clouds implemented in the Point Cloud Library (PCL). Check their presentation on Fast triangulation of unordered point clouds.