Image Super-resolution Reconstruction Based on Deep Residual Network

DOI

10.2991/icaita-18.2018.8How to use a DOI?

Keywords

super-resolution reconstruction; deep learning; residual network

Abstract

Having powerful ability to learn and represent features, deep convolutional neural networks (CNN) can get better results in image super-resolution reconstruction. However, deep networks also exist some problems. For example, the gradient will gradually vanish through the network, which makes the training difficult to converge. To address these problems, this paper presents a deep network model that is suitable for super-resolution reconstruction. By applying residual network modules, the model realizes the transmission of information across one or more layers. This residual network structure can not only avoid the loss of effective information in the network, but also speed up the training convergence. Compared with the previous classical methods, the proposed model converges faster, and the subjective and objective evaluations have been improved to a certain extent.

Copyright

© 2018, 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  - Bin Sun
AU  - Jian Lu
AU  - Xiaopeng Wei
PY  - 2018/03
DA  - 2018/03
TI  - Image Super-resolution Reconstruction Based on Deep Residual Network
BT  - Proceedings of the 2018 2nd International Conference on Artificial Intelligence: Technologies and Applications (ICAITA 2018)
PB  - Atlantis Press
SP  - 29
EP  - 32
SN  - 1951-6851
UR  - https://doi.org/10.2991/icaita-18.2018.8
DO  - 10.2991/icaita-18.2018.8
ID  - Sun2018/03
ER  -