GeoGCN: Geometric Dual-domain Graph Convolution Network for Point Cloud Denoising

TL;DR

GeoGCN introduces a dual-domain graph convolution network that leverages real and virtual normals to improve point cloud denoising, achieving superior noise robustness and feature preservation over existing methods.

Contribution

It presents a novel dual-normal approach combining real and virtual normals within a graph convolution framework for enhanced point cloud denoising.

Findings

Outperforms state-of-the-art methods in noise robustness

Preserves local details and global shape during denoising

Effectively exploits geometric information and training data

Abstract

We propose GeoGCN, a novel geometric dual-domain graph convolution network for point cloud denoising (PCD). Beyond the traditional wisdom of PCD, to fully exploit the geometric information of point clouds, we define two kinds of surface normals, one is called Real Normal (RN), and the other is Virtual Normal (VN). RN preserves the local details of noisy point clouds while VN avoids the global shape shrinkage during denoising. GeoGCN is a new PCD paradigm that, 1) first regresses point positions by spatialbased GCN with the help of VNs, 2) then estimates initial RNs by performing Principal Component Analysis on the regressed points, and 3) finally regresses fine RNs by normalbased GCN. Unlike existing PCD methods, GeoGCN not only exploits two kinds of geometry expertise (i.e., RN and VN) but also benefits from training data. Experiments validate that GeoGCN outperforms SOTAs in terms of both noise-robustness and local-and-global feature preservation.