Patch-Based Progressive 3D Point Set Upsampling

Wang Yifan, Shihao Wu, Hui Huang, Daniel Cohen-Or, Olga Sorkine-Hornung; The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 5958-5967

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.

Related Material


[pdf] [supp]
[bibtex]
@InProceedings{Yifan_2019_CVPR,
author = {Yifan, Wang and Wu, Shihao and Huang, Hui and Cohen-Or, Daniel and Sorkine-Hornung, Olga},
title = {Patch-Based Progressive 3D Point Set Upsampling},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}