Abstract:Signed Distance Functions (SDFs) are vital implicit representations to represent high fidelity 3D surfaces. Current methods mainly leverage a neural network to learn an SDF from various supervisions including signed distances, 3D point clouds, or multi-view images. However, due to various reasons including the bias of neural network on low frequency content, 3D unaware sampling, sparsity in point clouds, or low resolutions of images, neural implicit representations still struggle to represent geometries with high frequency components like sharp structures, especially for the ones learned from images or point clouds. To overcome this challenge, we introduce a method to sharpen a low frequency SDF observation by recovering its high frequency components, pursuing a sharper and more complete surface. Our key idea is to learn a mapping from a low frequency observation to a full frequency coverage in a data-driven manner, leading to a prior knowledge of shape consolidation in the frequency domain, dubbed frequency consolidation priors. To better generalize a learned prior to unseen shapes, we introduce to represent frequency components as embeddings and disentangle the embedding of the low frequency component from the embedding of the full frequency component. This disentanglement allows the prior to generalize on an unseen low frequency observation by simply recovering its full frequency embedding through a test-time self-reconstruction. Our evaluations under widely used benchmarks or real scenes show that our method can recover high frequency component and produce more accurate surfaces than the latest methods. The code, data, and pre-trained models are available at \url{<a class="link-external link-https" href="https://github.com/chenchao15/FCP" rel="external noopener nofollow">this https URL</a>}.

Sharpness fields in point clouds using deep learning

Deep Shape Representation with Sharp Feature Preservation

DeepSharpen: An Unsupervised Sharp Feature Recovery Method Based on Quadric Projection

Robust Sharp Features Infer From Point Clouds

MSL-Net: Sharp Feature Detection Network for 3D Point Clouds

PoU Based Sharp Features Extraction from Point Cloud

Automatic Sharp Feature Based Segmentation of Point Clouds

Sharp Feature-Preserving 3D Mesh Reconstruction from Point Clouds Based on Primitive Detection

Neural Octahedral Field: Octahedral prior for simultaneous smoothing and sharp edge regularization

Extracting Sharp Features from RGB‐D Images

Guided Point Cloud Denoising Via Sharp Feature Skeletons

Robust Normal Estimation for Point Clouds with Sharp Features

CS-Net: Deep Multibranch Network Considering Scene Features for Sharpness Assessment of Remote Sensing Images

SmoothNet: Smooth Point Cloud Up-sampling

PointCleanNet: Learning to Denoise and Remove Outliers from Dense Point Clouds

Deep Point Cloud Reconstruction

Sharpening Neural Implicit Functions with Frequency Consolidation Priors

Deep Neural Network for Point Sets Based on Local Feature Integration

Algorithm for Extracting Sharp Features from Point Cloud Models

FEATURE RELEVANCE ANALYSIS FOR 3D POINT CLOUD CLASSIFICATION USING DEEP LEARNING