PCPNET: Learning Local Shape Properties from Raw Point Clouds

Abstract

In this paper, we propose PCPNet, a deep-learning based approach for estimating local 3D shape properties in point clouds. In contrast to the majority of prior techniques that concentrate on global or mid-level attributes, e.g., for shape classification or semantic labeling, we suggest a patch-based learning method, in which a series of local patches at multiple scales around each point is encoded in a structured manner. Our approach is especially well-adapted for estimating local shape properties such as normals (both unoriented and oriented) and curvature from raw point clouds in the presence of strong noise and multi-scale features. Our main contributions include both a novel multi-scale variant of the recently proposed PointNet architecture with emphasis on local shape information, and a series of novel applications in which we demonstrate how learning from training data arising from well-structured triangle meshes, and applying the trained model to noisy point clouds can produce superior results compared to specialized state-of-the-art techniques. Finally, we demonstrate the utility of our approach in the context of shape reconstruction, by showing how it can be used to extract normal orientation information from point clouds.

presented at Eurographics 2018

References in corpus (2)

• PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
• Deep Convolutional Networks on Graph-Structured Data

Cited by in corpus (62)

• Fast Graph Representation Learning with PyTorch Geometric
• MeshCNN: A Network with an Edge
• Dynamic Graph CNN for Learning on Point Clouds
• Differentiable Surface Splatting for Point-based Geometry Processing
• Supervised Fitting of Geometric Primitives to 3D Point Clouds
• Points2Surf: Learning Implicit Surfaces from Point Cloud Patches
• Relation-Shape Convolutional Neural Network for Point Cloud Analysis
• Self-Contrastive Learning with Hard Negative Sampling for Self-supervised Point Cloud Learning
• Unpaired Point Cloud Completion on Real Scans using Adversarial Training
• PU-Net: Point Cloud Upsampling Network
• Orienting Point Clouds with Dipole Propagation
• PCEDNet : A Lightweight Neural Network for Fast and Interactive Edge Detection in 3D Point Clouds
• CNNs on Surfaces using Rotation-Equivariant Features
• PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling
• PIE-NET: Parametric Inference of Point Cloud Edges
• ASSANet: An Anisotropic Separable Set Abstraction for Efficient Point Cloud Representation Learning
• Patch-based Progressive 3D Point Set Upsampling
• Dynamic Point Cloud Denoising via Manifold-to-Manifold Distance
• PointCleanNet: Learning to Denoise and Remove Outliers from Dense Point Clouds
• H-CNN: Spatial Hashing Based CNN for 3D Shape Analysis
• DensePoint: Learning Densely Contextual Representation for Efficient Point Cloud Processing
• Fast and Globally Consistent Normal Orientation based on the Winding Number Normal Consistency
• Deep Learning for 3D Point Cloud Understanding: A Survey
• Total Denoising: Unsupervised Learning of 3D Point Cloud Cleaning
• Contrastive Learning for Joint Normal Estimation and Point Cloud Filtering
• Neural-IMLS: Self-supervised Implicit Moving Least-Squares Network for Surface Reconstruction
• Point normal orientation and surface reconstruction by incorporating isovalue constraints to Poisson equation
• Pointwise Rotation-Invariant Network with Adaptive Sampling and 3D Spherical Voxel Convolution
• MODNet: Multi-offset Point Cloud Denoising Network Customized for Multi-scale Patches
• PointWise: An Unsupervised Point-wise Feature Learning Network
• PPSURF: Combining Patches and Point Convolutions for Detailed Surface Reconstruction
• Learning Modified Indicator Functions for Surface Reconstruction
• SAWNet: A Spatially Aware Deep Neural Network for 3D Point Cloud Processing
• Improving Semantic Analysis on Point Clouds via Auxiliary Supervision of Local Geometric Priors
• Deep Feature-preserving Normal Estimation for Point Cloud Filtering
• AutoMate: A Dataset and Learning Approach for Automatic Mating of CAD Assemblies
• A Survey of Methods for Converting Unstructured Data to CSG Models
• GMCR: Graph-based Maximum Consensus Estimation for Point Cloud Registration
• Deformation Recovery: Localized Learning for Detail-Preserving Deformations
• MeshWalker: Deep Mesh Understanding by Random Walks
• Deep Mesh Prior: Unsupervised Mesh Restoration using Graph Convolutional Networks
• Multimodal Shape Completion via Conditional Generative Adversarial Networks
• Neighbourhood-Insensitive Point Cloud Normal Estimation Network
• Geometry Sharing Network for 3D Point Cloud Classification and Segmentation
• Shape As Points: A Differentiable Poisson Solver
• Normal Estimation for 3D Point Clouds via Local Plane Constraint and Multi-scale Selection
• Normal Transformer: Extracting Surface Geometry from LiDAR Points Enhanced by Visual Semantics
• Improvement of Normal Estimation for PointClouds via Simplifying Surface Fitting
• EC-Net: an Edge-aware Point set Consolidation Network
• Tranquil Clouds: Neural Networks for Learning Temporally Coherent Features in Point Clouds
• CPSeg: Cluster-free Panoptic Segmentation of 3D LiDAR Point Clouds
• Orderly Disorder in Point Cloud Domain
• Weakly Supervised Point Clouds Transformer for 3D Object Detection
• Meshlet Priors for 3D Mesh Reconstruction
• Fast and Accurate Normal Estimation for Point Cloud via Patch Stitching
• Going Deeper with Lean Point Networks
• Towards an Automatic System for Extracting Planar Orientations from Software Generated Point Clouds
• Generic Primitive Detection in Point Clouds Using Novel Minimal Quadric Fits
• Meshing Point Clouds with Predicted Intrinsic-Extrinsic Ratio Guidance
• Learning to Reconstruct and Segment 3D Objects
• Region segmentation via deep learning and convex optimization
• Deep Iterative Surface Normal Estimation