DUDE: Deep Unsigned Distance Embeddings for Hi-Fidelity Representation of Complex 3D Surfaces - Jianfei Guo

目录

目录

DUDE: Deep Unsigned Distance Embeddings for Hi-Fidelity Representation of Complex 3D Surfaces

Jianfei Guo 出版于 paper_reading

2020-12-03 2020-12-03 约 227 字预计阅读 1 分钟

DUDE: Deep unsigned distance embeddings for hi-fidelity representation of complex 3D surfaces

arxiv2020 arXiv preprint arXiv:2011.02570

Rahul Venkatesh, Sarthak Sharma, Aurobrata Ghosh, Laszlo Jeni, Maneesh Singh

CMU Verisk Analytics

unsigned distance field, normal vector field, open topogoly surfaces

目录

Motivation

现有的隐式表面deep networks方法只能表征拓扑上闭合的形状；
并且结果是，训练时候经常需要clean watertight meshes
本篇提出无符号的距离嵌入减轻了上述问题
- 利用unsigned distance field (uDF)无符号距离场来表达对表面的接近程度
- 利用normal vector field (nVF)法向量场来表达表面朝向
- uDF + nVF 可以表达任意开/闭拓扑的high fidelity形状
- 可以从带噪声的triangle soups学习，不需要watertight mehses
- 并且额外提供了学到的表征提取、渲染等值面的新方法

overview

uDF+nVF