Deep Residual Neural Network Design for Super-Resolution Imaging

Wei Ting Chen, Pei-Yin Chen, Bo Chen Lin

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Convolution neural network recently confirmed the high-quality reconstruction for single-image super-resolution (SR). In this paper we present a Deep Level Residual Network (DLNR), a low-memory effective neural network to reconstruct super-resolution images. This neural network also has the following characteristics. (1) Ability to perform different convolution size operations on the image which can achieve more comprehensive feature extraction effects. (2) Using residual learning to expand the depth of the network and increase the capacity of learning. (3) Taking the skill of parameter sharing between the network module to reduce the number of parameters. After the experiment, we find that DLNR can achieve 37.78 in PSNR and 0.975 in SSIM when using Manga109 as testing set for 2× SR.

Original languageEnglish
Title of host publicationNew Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers
EditorsChuan-Yu Chang, Chien-Chou Lin, Horng-Horng Lin
PublisherSpringer Verlag
Pages95-105
Number of pages11
ISBN (Print)9789811391897
DOIs
Publication statusPublished - 2019 Jan 1
Event23rd International Computer Symposium, ICS 2018 - Yunlin, Taiwan
Duration: 2018 Dec 202018 Dec 22

Publication series

NameCommunications in Computer and Information Science
Volume1013
ISSN (Print)1865-0929

Conference

Conference23rd International Computer Symposium, ICS 2018
CountryTaiwan
CityYunlin
Period18-12-2018-12-22

Fingerprint

Super-resolution
Network Design
Imaging
Neural Networks
Neural networks
Convolution
Imaging techniques
Image resolution
Feature extraction
Feature Extraction
Expand
Data storage equipment
Sharing
Testing
Module
Experiments
Experiment
Learning

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Mathematics(all)

Cite this

Chen, W. T., Chen, P-Y., & Lin, B. C. (2019). Deep Residual Neural Network Design for Super-Resolution Imaging. In C-Y. Chang, C-C. Lin, & H-H. Lin (Eds.), New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers (pp. 95-105). (Communications in Computer and Information Science; Vol. 1013). Springer Verlag. https://doi.org/10.1007/978-981-13-9190-3_9
Chen, Wei Ting ; Chen, Pei-Yin ; Lin, Bo Chen. / Deep Residual Neural Network Design for Super-Resolution Imaging. New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers. editor / Chuan-Yu Chang ; Chien-Chou Lin ; Horng-Horng Lin. Springer Verlag, 2019. pp. 95-105 (Communications in Computer and Information Science).
@inproceedings{43ed1d1045f94d3e90be758fbde9108a,
title = "Deep Residual Neural Network Design for Super-Resolution Imaging",
abstract = "Convolution neural network recently confirmed the high-quality reconstruction for single-image super-resolution (SR). In this paper we present a Deep Level Residual Network (DLNR), a low-memory effective neural network to reconstruct super-resolution images. This neural network also has the following characteristics. (1) Ability to perform different convolution size operations on the image which can achieve more comprehensive feature extraction effects. (2) Using residual learning to expand the depth of the network and increase the capacity of learning. (3) Taking the skill of parameter sharing between the network module to reduce the number of parameters. After the experiment, we find that DLNR can achieve 37.78 in PSNR and 0.975 in SSIM when using Manga109 as testing set for 2× SR.",
author = "Chen, {Wei Ting} and Pei-Yin Chen and Lin, {Bo Chen}",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-981-13-9190-3_9",
language = "English",
isbn = "9789811391897",
series = "Communications in Computer and Information Science",
publisher = "Springer Verlag",
pages = "95--105",
editor = "Chuan-Yu Chang and Chien-Chou Lin and Horng-Horng Lin",
booktitle = "New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers",
address = "Germany",

}

Chen, WT, Chen, P-Y & Lin, BC 2019, Deep Residual Neural Network Design for Super-Resolution Imaging. in C-Y Chang, C-C Lin & H-H Lin (eds), New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers. Communications in Computer and Information Science, vol. 1013, Springer Verlag, pp. 95-105, 23rd International Computer Symposium, ICS 2018, Yunlin, Taiwan, 18-12-20. https://doi.org/10.1007/978-981-13-9190-3_9

Deep Residual Neural Network Design for Super-Resolution Imaging. / Chen, Wei Ting; Chen, Pei-Yin; Lin, Bo Chen.

New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers. ed. / Chuan-Yu Chang; Chien-Chou Lin; Horng-Horng Lin. Springer Verlag, 2019. p. 95-105 (Communications in Computer and Information Science; Vol. 1013).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Deep Residual Neural Network Design for Super-Resolution Imaging

AU - Chen, Wei Ting

AU - Chen, Pei-Yin

AU - Lin, Bo Chen

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Convolution neural network recently confirmed the high-quality reconstruction for single-image super-resolution (SR). In this paper we present a Deep Level Residual Network (DLNR), a low-memory effective neural network to reconstruct super-resolution images. This neural network also has the following characteristics. (1) Ability to perform different convolution size operations on the image which can achieve more comprehensive feature extraction effects. (2) Using residual learning to expand the depth of the network and increase the capacity of learning. (3) Taking the skill of parameter sharing between the network module to reduce the number of parameters. After the experiment, we find that DLNR can achieve 37.78 in PSNR and 0.975 in SSIM when using Manga109 as testing set for 2× SR.

AB - Convolution neural network recently confirmed the high-quality reconstruction for single-image super-resolution (SR). In this paper we present a Deep Level Residual Network (DLNR), a low-memory effective neural network to reconstruct super-resolution images. This neural network also has the following characteristics. (1) Ability to perform different convolution size operations on the image which can achieve more comprehensive feature extraction effects. (2) Using residual learning to expand the depth of the network and increase the capacity of learning. (3) Taking the skill of parameter sharing between the network module to reduce the number of parameters. After the experiment, we find that DLNR can achieve 37.78 in PSNR and 0.975 in SSIM when using Manga109 as testing set for 2× SR.

UR - http://www.scopus.com/inward/record.url?scp=85069722534&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069722534&partnerID=8YFLogxK

U2 - 10.1007/978-981-13-9190-3_9

DO - 10.1007/978-981-13-9190-3_9

M3 - Conference contribution

AN - SCOPUS:85069722534

SN - 9789811391897

T3 - Communications in Computer and Information Science

SP - 95

EP - 105

BT - New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers

A2 - Chang, Chuan-Yu

A2 - Lin, Chien-Chou

A2 - Lin, Horng-Horng

PB - Springer Verlag

ER -

Chen WT, Chen P-Y, Lin BC. Deep Residual Neural Network Design for Super-Resolution Imaging. In Chang C-Y, Lin C-C, Lin H-H, editors, New Trends in Computer Technologies and Applications - 23rd International Computer Symposium, ICS 2018, Revised Selected Papers. Springer Verlag. 2019. p. 95-105. (Communications in Computer and Information Science). https://doi.org/10.1007/978-981-13-9190-3_9