Asymmetrical Siamese Network for Point Clouds Normal Estimation

TL;DR

This paper introduces an asymmetric Siamese network architecture and a new multi-view dataset to improve point cloud normal estimation, addressing overfitting and cross-domain performance issues.

Contribution

It proposes a novel asymmetric Siamese network with feature constraints and a diverse dataset to enhance normal estimation accuracy and robustness.

Findings

The new dataset reveals existing methods overfit to specific shapes.

The proposed method improves normal estimation accuracy across different noise levels.

Feature constraints reduce overfitting and enhance cross-domain performance.

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.