Single Image Defogging Method based on Deep Learning

DOI

10.2991/mecae-17.2017.23How to use a DOI?

Keywords

Image Defogging; Deep Learning; Convolutional Neural Network; Transmission Map; Atmospheric Scattering Model.

Abstract

Single image defogging is a challenging ill-posed problem. Current image defogging methods usually get defogging solutions based on various priors or assumption, which is hardly satisfied in practice. In this paper, a single image defogging method based on deep learning is proposed, in which the priors and assumption do not hold. Firstly, the prediction of transmission map is progressively refined by using three scales convolutional neural networks. Secondly, the fog-free image can be recovered by the atmospheric scattering model after transmission map is got. Experiment results show that the proposed deep learning based defogging algorithm achieves superior performance over state-of-art algorithms on both natural foggy images and synthetic images.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Baoping Yuan
AU  - Yong Yang
AU  - Baofu Zhang
PY  - 2017/03
DA  - 2017/03
TI  - Single Image Defogging Method based on Deep Learning
BT  - Proceedings of the 2017 International Conference on Mechanical, Electronic, Control and Automation Engineering (MECAE 2017)
PB  - Atlantis Press
SP  - 126
EP  - 131
SN  - 2352-5401
UR  - https://doi.org/10.2991/mecae-17.2017.23
DO  - 10.2991/mecae-17.2017.23
ID  - Yuan2017/03
ER  -