TL;DR本文提出了一种基于自编码器的 3D 点云去噪方法，该方法通过学习嵌入式邻域特征的稀疏点的局部和非局部特征表示，对 3D 点云中的噪声进行去除，并通过重建的流形重新采样得到去噪后的点云，该方法在合成噪声和真实世界噪声下均表现出显著的性能优于现有的去噪方法。
Abstract
3d point clouds are often perturbed by noise due to the inherent limitation
of acquisition equipments, which obstructs downstream tasks such as surface
reconstruction, rendering and so on. Previous works mostly infer the
displacement of noisy points from the underlying surface, which h