Score-Based Point Cloud Denoising

Abstract

Point clouds acquired from scanning devices are often perturbed by noise, which affects downstream tasks such as surface reconstruction and analysis. The distribution of a noisy point cloud can be viewed as the distribution of a set of noise-free samples $p(x)$ convolved with some noise model $n$, leading to $(p\ast n)(x)$ whose mode is the underlying clean surface. To denoise a noisy point cloud, we propose to increase the log-likelihood of each point from $p\ast n$ via gradient ascent -- iteratively updating each point's position. Since $p\ast n$ is unknown at test-time, and we only need the score (i.e., the gradient of the log-probability function) to perform gradient ascent, we propose a neural network architecture to estimate the score of $p\ast n$ given only noisy point clouds as input. We derive objective functions for training the network and develop a denoising algorithm leveraging on the estimated scores. Experiments demonstrate that the proposed model outperforms state-of-the-art methods under a variety of noise models, and shows the potential to be applied in other tasks such as point cloud upsampling. The code is available at \url{https://github.com/luost26/score-denoise}.

arXiv commentICCV 2021

References in corpus (7)

• Generative Modeling by Estimating Gradients of the Data Distribution
• Accelerating 3D Deep Learning with PyTorch3D
• Differentiable Manifold Reconstruction for Point Cloud Denoising
• Feature Graph Learning for 3D Point Cloud Denoising
• Graph-based denoising for time-varying point clouds
• Graph Signal Processing for Geometric Data and Beyond: Theory and Applications
• 3D Dynamic Point Cloud Denoising via Spatial-Temporal Graph Learning

Cited by in corpus (3)

• Contrastive Learning for Joint Normal Estimation and Point Cloud Filtering
• Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation
• Incomplete Gamma Kernels: Generalizing Locally Optimal Projection Operators