Score-Based Point Cloud Denoising

@article{Luo2021ScoreBasedPC,
  title={Score-Based Point Cloud Denoising},
  author={Shitong Luo and Wei Hu},
  journal={2021 IEEE/CVF International Conference on Computer Vision (ICCV)},
  year={2021},
  pages={4563-4572}
}
  • Shitong LuoWei Hu
  • Published 23 July 2021
  • Computer Science
  • 2021 IEEE/CVF International Conference on Computer Vision (ICCV)
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 * 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 * n via gradient ascent—iteratively… 

Figures and Tables from this paper

Dynamic Point Cloud Denoising via Gradient Fields

  • Q. HuWei Hu
  • Computer Science, Environmental Science
    ArXiv
  • 2022
This paper proposes a novel gradient-field-based dynamic point cloud denoising method, exploiting the temporal correspondence via the estimation of gradient—a fundamental problem in dynamic point clouds processing and analysis.

Deep Point Set Resampling via Gradient Fields

A novel paradigm of point set resampling for restoration, which learns continuous gradient fields of point clouds that converge points towards the underlying surface that guarantees the continuity of the model for solvable optimization is proposed.

PCDNF: Revisiting Learning-based Point Cloud Denoising via Joint Normal Filtering

This work revisits point cloud denoising from a multitask perspective, and proposes an end-to-end network, named PCDNF, to denoise point clouds via joint normal filtering, which outperforms state-of-the-arts for both point cloudDenoising and normalflltering.

Point Cloud Denoising via Momentum Ascent in Gradient Fields

This work develops the momentum gradient ascent method that leverages the information of previous iterations in determining the trajectories of the points, thus improving the stability of the solution and reducing the inference time.

PD-Flow: A Point Cloud Denoising Framework with Normalizing Flows

This work presents a novel deep learning-based denoising model, that incorporates normalizing flows and noise disentanglement techniques to achieve high denoised accuracy, and demonstrates that this method outperforms previous state-of-the-art deeplearning-based approaches.

Deep Point Cloud Reconstruction

A deep point cloud reconstruction network consisting of a 3D sparse stacked-hourglass network as for the initial densification and denoising, and a refinement via transformers converting the discrete voxels into 3D points called amplified positional encoding is proposed.

Contrastive Learning for Joint Normal Estimation and Point Cloud Filtering

Experimental results show that the proposed novel deep learning method well supports the two tasks simultaneously and preserves sharp features and fine details, and generally outperforms state-of-the-art techniques on both tasks.

Gradient-based Point Cloud Denoising with Uniformity

This work proposes GPCD++, a gradient-based denoiser with an ultra-lightweight network named UniNet to address uniformity, and shows that this approach not only generates competitive or even better denoising results, but also improves uniformity which largely benefits applications such as surface reconstruction.

CMD-Net: Self-Supervised Category-Level 3D Shape Denoising through Canonicalization

This paper presents a self-supervised learning-based method, Canonical Mapping and Denoising Network (CMD-Net), and address category-level 3D shape denoising through canonicalization, capable of canonicalizing noise-corrupted clouds under arbitrary rotations.

SLiDE: Self-supervised LiDAR De-snowing through Reconstruction Difficulty

This work proposes a novel self-supervised learning framework for snow points removal in LiDAR point clouds that achieves the state-of-the-art performance among label-free approaches and is comparable to the fully- supervised method.