Light Field Synthesis from a Single Image using Improved Wasserstein Generative Adversarial Network

Ruan, Lingyan; Chen, Bin; Lam, Miu Ling

Light Field Synthesis from a Single Image using Improved Wasserstein Generative Adversarial Network

Files

019-020.pdf (601.21 KB)

eg_poster_final_a4.pdf (7.19 MB)

Date

2018

Authors

Ruan, Lingyan
Chen, Bin
Lam, Miu Ling

Publisher

The Eurographics Association

Abstract

We present a deep learning-based method to synthesize a 4D light field from a single 2D RGB image. We consider the light field synthesis problem equivalent to image super-resolution, and solve it by using the improved Wasserstein Generative Adversarial Network with gradient penalty (WGAN-GP). Experimental results demonstrate that our algorithm can predict complex occlusions and relative depths in challenging scenes. The light fields synthesized by our method has much higher signal-to-noise ratio and structural similarity than the state-of-the-art approach.

        @inproceedings{10.2312:egp.20181017
,
booktitle = {EG 2018 - Posters
},
editor = {Jain, Eakta and Kosinka, Jirí
},
title = {{Light Field Synthesis from a Single Image using Improved Wasserstein Generative Adversarial Network
}},
author = {Ruan, Lingyan and 
Chen, Bin and 
Lam, Miu Ling
},
year = {2018
},
publisher = {The Eurographics Association
},
ISSN = {1017-4656
},
ISBN = {},
DOI = {10.2312/egp.20181017
}
}