.png){kind=icon}
返回
A Noising-Denoising Framework for Point Cloud Upsampling via Normalizing Flows
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2023.109569.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Point cloud upsampling aims to generate dense and uniform point cloud from the sparse input point cloud. One challenge is how to flexibly upsample the sparse point cloud in arbitrary ratios, even without the given supervised high resolution point cloud. To address this challenge, we propose a noisingdenoising framework, dubbed ND-PUFlow, for arbitrary 3D point cloud upsampling (3DPU) in supervised and self-supervised settings. It consists of two stages, i.e., dense noisy points generation and noisy points denoising via continuous normalizing flows (CNFs). In the first stage, noisy points are generated by adding noise to the input points. In the second stage, CNFs move each noisy point to the underlying surface, forming a dense and clean point cloud. Extensive experiments show that our method is competitive in both supervised and self-supervised settings, and in most cases, it achieves the best performance on benchmark datasets for 3DPU.
Keyword:
Point cloud
Arbitrary ratio upsampling
Normalizing flows
期刊
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
引用论文
Spatial information enhancement network for 3D object detection from point cloud基于空间信息增强网络的点云三维目标检测.png)
PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Continuous conditional random field convolution for point cloud segmentationPATTERN RECOGNITION
OAAI
PATTERN RECOGNITIONIF7.6
OAAI分享收藏3D-CenterNet: 3D object detection network for point clouds with center estimation priority *3D-CenterNet: 具有中心估计优先级的点云3D物体检测网络 *PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏SPU-Net: Self-Supervised Point Cloud Upsampling by Coarse-to-Fine Reconstruction With Self-Projection OptimizationSpu-net: 自投影优化从粗到精重建的自监督点云上采样
OAAI
OAAI分享收藏
查看更多内容