P2P-Bridge

Diffusion Bridges for 3D Point Cloud Denoising

ECCV 2024

Mathias Vogel1, Keisuke Tateno2, Marc Pollefeys1,4, Federico Tombari2,3,
Marie-Julie Rakotosaona2,*, Francis Engelmann1,2,*
1ETH Zürich 2Google 3TUM 4Microsoft *Equal contribution
Paper Supplementary Code Poster


P2P-Bridge introduces a novel approach for point cloud denoising by adapting Diffusion Schrödinger bridges to learn an optimal transport plan between paired point sets. Further enhancements are possible by incorporating additional features such as RGB data and point-wise DINOV2 features.

Abstract

In this work, we tackle the task of point cloud denoising through a novel framework that adapts Diffusion Schrödinger bridges to points clouds. Unlike previous approaches that predict point-wise displacements from point features or learned noise distributions, our method learns an optimal transport plan between paired point clouds. Experiments on object datasets like PU-Net and real-world datasets such as ScanNet++ and ARKitScenes show that P2P-Bridge achieves significant improvements over existing methods. While our approach demonstrates strong results using only point coordinates, we also show that incorporating additional features, such as color information or point-wise DINOv2 features, further enhances the performance.

Methodology

Methodology of P2P-Bridge.

Methodology of P2P-Bridge, modeling point cloud denoising as a reverse data-to- data diffusion process. Our model can effectively transform noisy data into cleaner data by learning a bridge between clean and noisy data.

Results

Object-Level Denoising

teaser-fig.

Qualitative comparison of our P2P-Bridge with recent deep-learning-based point cloud denoising methods on the PU-Net dataset under 3% isotropic Gaussian noise.

Szene-Level Denoising

teaser-fig.

Qualitative comparison on the ScanNet++ dataset (top 3 rows) and the ARKitScenes dataset (bottom 2 rows) using noisy iPhone scans as input. P2P-Bridge effectively removes real-world noise

Before and After Denoising

Qualitative Results

Object-Level Denoising

teaser-fig.

Qualitative comparison of our P2P-Bridge with recent deep-learning-based point cloud denoising methods on the PU-Net dataset under 3% isotropic Gaussian noise.

Scene-Level Denoising

teaser-fig.

Qualitative comparison on the ScanNet++ dataset (top 3 rows) and the ARKitScenes dataset (bottom 2 rows) using noisy iPhone scans as input. P2P-Bridge effectively removes real-world noise

BibTeX

@inproceedings{vogel2024p2pbridge, 
  title={P2P-Bridge: Diffusion Bridges for 3D Point Cloud Denoising},
  author = {Mathias Vogel, Mathias and Tateno, Keisuke and Pollefeys, Marc and Tombari, Federico and Rakotosaona, Marie-Julie and Engelmann, Francis}, 
  booktitle = {European Conference on Computer Vision (ECCV)}, 
  year = {2024}
}