基于自适应球覆盖的散乱点采样与定向

Sampling and Orienting of Scattered Points Based on Adaptive Spherical Covering

对散乱点的正确曲面重构至今依然是一个难题,特别是对于一个带有噪声、孤立点、薄壳结构及分布不均匀数据点集.而正确的法向信息在曲面重构中起着至关重要的作用.在文中,作者提出了一个法向量方向一致化的方法,能处理上述特性的数据,实现对曲面重构的数据进行预处理.首先,使用基于自适应球覆盖技术生成原始数据的一个根据曲面几何特征进行分布的采样点集,以保证在重构曲面上保留重要几何特征.然后,借助于球体的相交关系构造出采样点的邻域,实现在非均匀分布的数据上进行法向量方向传播,获得具有正确法向信息的采样点集.文中例子能充分说明文中方法的稳定性和有效性.

The surface reconstruction from scattered points is still a difficult process, especially for the points with noise, outliers, non-uniformities, and thin-sharp features. Oriented normals at the points play a critical role in surface reconstruction. This paper presents a method to provide orientation consistent normals on points with the above defects as a preprocessing for surface reconstruction. Firstly, this method generates a particle set by an adaptive spherical cover. The particles are distributed on the reconstructed surface according to the surface shape for preserving important geometric features. Secondly, a priority-driven normal propagation procedure is performed on an uneven distributed data to assign consistent oriented normal vectors to the particles. During this procedure, the neighborhoods of the particles are constructed depending on the intersection relationships between spheres. The results in this paper demonstrate that the method is robust and efficient.