Asymmetrical Siamese Network for Point Clouds Normal Estimation

TL;DR

The paper tackles robust normal estimation for 3D point clouds across different noise scales, addressing overfitting in datasets like PCPNet and poor cross-domain generalization. It proposes an Asymmetric Siamese Network with two branches processing clean and noisy point clouds, enforcing feature consistency through a constraint to improve normal estimation quality. A novel multi-view normal estimation dataset with diverse shapes and noise levels is introduced to better evaluate cross-domain performance. Experiments indicate existing methods struggle on the new dataset, while the proposed feature-constraint mechanism reduces overfitting and yields improved normal estimation results across varying noise conditions.

Abstract

In recent years, deep learning-based point cloud normal estimation has made great progress. However, existing methods mainly rely on the PCPNet dataset, leading to overfitting. In addition, the correlation between point clouds with different noise scales remains unexplored, resulting in poor performance in cross-domain scenarios. In this paper, we explore the consistency of intrinsic features learned from clean and noisy point clouds using an Asymmetric Siamese Network architecture. By applying reasonable constraints between features extracted from different branches, we enhance the quality of normal estimation. Moreover, we introduce a novel multi-view normal estimation dataset that includes a larger variety of shapes with different noise levels. Evaluation of existing methods on this new dataset reveals their inability to adapt to different types of shapes, indicating a degree of overfitting. Extensive experiments show that the proposed dataset poses significant challenges for point cloud normal estimation and that our feature constraint mechanism effectively improves upon existing methods and reduces overfitting in current architectures.