Patch-based Progressive 3D Point Set Upsampling
Authors:
Preprint
Abstract
We present a detail-driven deep neural network for point set upsampling. A high-resolution point set is essential for point-based rendering and surface reconstruction. Inspired by the recent success of neural image super-resolution techniques, we progressively train a cascade of patch-based upsampling networks on different levels of detail end-to-end. We propose a series of architectural design contributions that lead to a substantial performance boost. The effect of each technical contribution is demonstrated in an ablation study. Qualitative and quantitative experiments show that our method significantly outperforms the state-of-the-art learning-based and optimazation-based approaches, both in terms of handling low-resolution inputs and revealing high-fidelity details.Keywords
All Related Versions
This publication has 20 references indexed in Scilit:
- Data-driven Upsampling of Point CloudsComputer-Aided Design, 2019
- Know What Your Neighbors Do: 3D Semantic Segmentation of Point Clouds2019
- Multiresolution Tree Networks for 3D Point Cloud Processing2018
- PPF-FoldNet: Unsupervised Learning of Rotation Invariant 3D Local Descriptors2018
- Fully-Convolutional Point Networks for Large-Scale Point Clouds2018
- GUN: Gradual Upsampling Network for Single Image Super-ResolutionIEEE Access, 2018
- Frustum PointNets for 3D Object Detection from RGB-D Data2018
- Image Super-Resolution Using Deep Convolutional NetworksIEEE Transactions on Pattern Analysis and Machine Intelligence, 2015
- Computing and rendering point set surfacesIEEE Transactions on Visualization and Computer Graphics, 2003
- Triangulations in CGALComputational Geometry, 2002