The datasets were generated in order to test our dense light field reconstruction method based on Epipolar Focus Spectrum (EFS). A portion of our work has been published in IEEE TCI
We have three dense light field datests:
1. The synthetic data Syn_POV_LF generated with open source software POV, containing 12 Light fileds. Fig.1 illustrates the central view of these light fields.
2. The synthetic data Syn_Ble_LF generated with open source software Blender, including 4 light fields of indoor scenes. Fig.2 shows the central view of these light fields.
3. Real scene light field datasets Real_Lytro_LF captured by Lytro Illum and used in Guo et al. Fig.3 shows the central view of these light fields.
Fig.1: Synthetic light fields Syn_POV_LF generated with open source software POV. Fig.2: Synthetic light fields Syn_Ble_LF generated with open source software Blender Fig.3: Real scene light fields Real_Lytro_LF captured by Lytro Illum
@ARTICLE{Li_FSS_TCI2021,
author={ Li, Yaning and Wang, Xue and Zhu, Hao and Zhou, Guoqing and Wang, Qing},
journal={IEEE Transactions on Computational Imaging},
title={Deep Anti-Aliasing of Whole Focal Stack Using Slice Spectrum},
year={2021},
volume={7},
pages={1328-1340},
doi={10.1109/TCI.2021.3132194}}
@inproceedings{guo2018dense,
title={Dense Light Field Reconstruction From Sparse Sampling Using Residual Network},
author={Guo, Mantang and Zhu, Hao and Zhou, Guoqing and Wang, Qing},
booktitle={Springer ACCV},
pages={50--65},
year={2018}}
Acknowledgement
The work was supported by NSFC under Grant 62031023, Grant 61801396 and Grant 62101242. The website template was borrowed from Jinlei Shi.
