海量采样点集法向聚类并行估计及增量统一算法

Normal Clustering Parallel Estimation and Incremental Orientation for Massive Point Clouds

为提高对海量采样点集法向估计的有效性,提出一种法向并行估计算法。首先利用聚类算法将采样点集分解成多个子集,并通过阈值过滤获取样点子集分界带;然后将各样点子集的法向计算过程并行化处理;最后,采用增量算法实现样点子集的法向统一。实验证明,对海量点云模型法向估计,在确保法向估计的准确性不低于现有算法的前提下,计算效率与内存利用率得到显著提高。

In this work,we describe a parallel implementation of the normal estimate for massive point cloud model.First,the point set is divided into several sub sets by using the clustering algorithm,and get subsets demarcation band through threshold filtering;And then the subset of the various points of the normal computing process parallelization;Finally,adopting incremental algorithm to realize normal uniformity of sample subset.Experimental results show that the computational efficiency and memory utilization are improved significantly while the accuracy of normal estimation is not less than the existing algorithms for the normal estimation of the massive point cloud model.