Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations
Although SISR (Single Image Super Resolution) problem can be effectively solved by deep learning based methods, the training phase often considers single degradation type such as bicubic interpolation or Gaussian blur with fixed variance. These priori hypotheses often fail and lead to reconstruction...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2020-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9066958/ |
id |
doaj-749bdb2739994a0bbf1beaa55b4fbb06 |
---|---|
record_format |
Article |
spelling |
doaj-749bdb2739994a0bbf1beaa55b4fbb062021-03-30T01:40:51ZengIEEEIEEE Access2169-35362020-01-018741957420410.1109/ACCESS.2020.29879119066958Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian DegradationsWenyi Wang0https://orcid.org/0000-0003-1619-0294Guangyang Wu1https://orcid.org/0000-0002-9537-2944Weitong Cai2https://orcid.org/0000-0001-7726-4387Liaoyuan Zeng3https://orcid.org/0000-0002-7524-5227Jianwen Chen4School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, ChinaAlthough SISR (Single Image Super Resolution) problem can be effectively solved by deep learning based methods, the training phase often considers single degradation type such as bicubic interpolation or Gaussian blur with fixed variance. These priori hypotheses often fail and lead to reconstruction error in real scenario. In this paper, we propose an end-to-end CNN model RPSRMD to handle SR problem in multiple Gaussian degradations by extracting and using as side information a shared image prior that is consistent in different Gaussian degradations. The shared image prior is generated by an AED network RPGen with a rationally designed loss function that contains two parts: consistency loss and validity loss. These losses supervise the training of AED to guarantee that the image priors of one image with different Gaussian blurs to be very similar. Afterwards we carefully designed a SR network, which is termed as PResNet (Prior based Residual Network) in this paper, to efficiently use the image priors and generate high quality and robust SR images when unknown Gaussian blur is presented. When we applied variant Gaussian blurs to the low resolution images, the experiments prove that our proposed RPSRMD, which includes RPGen and PResNet as two core components, is superior to many state-of-the-art SR methods that were designed and trained to handle multi-degradation.https://ieeexplore.ieee.org/document/9066958/Single image super-resolutionconvolution neural networkGaussian blurmultiple degradations |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Wenyi Wang Guangyang Wu Weitong Cai Liaoyuan Zeng Jianwen Chen |
spellingShingle |
Wenyi Wang Guangyang Wu Weitong Cai Liaoyuan Zeng Jianwen Chen Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations IEEE Access Single image super-resolution convolution neural network Gaussian blur multiple degradations |
author_facet |
Wenyi Wang Guangyang Wu Weitong Cai Liaoyuan Zeng Jianwen Chen |
author_sort |
Wenyi Wang |
title |
Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations |
title_short |
Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations |
title_full |
Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations |
title_fullStr |
Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations |
title_full_unstemmed |
Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations |
title_sort |
robust prior-based single image super resolution under multiple gaussian degradations |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2020-01-01 |
description |
Although SISR (Single Image Super Resolution) problem can be effectively solved by deep learning based methods, the training phase often considers single degradation type such as bicubic interpolation or Gaussian blur with fixed variance. These priori hypotheses often fail and lead to reconstruction error in real scenario. In this paper, we propose an end-to-end CNN model RPSRMD to handle SR problem in multiple Gaussian degradations by extracting and using as side information a shared image prior that is consistent in different Gaussian degradations. The shared image prior is generated by an AED network RPGen with a rationally designed loss function that contains two parts: consistency loss and validity loss. These losses supervise the training of AED to guarantee that the image priors of one image with different Gaussian blurs to be very similar. Afterwards we carefully designed a SR network, which is termed as PResNet (Prior based Residual Network) in this paper, to efficiently use the image priors and generate high quality and robust SR images when unknown Gaussian blur is presented. When we applied variant Gaussian blurs to the low resolution images, the experiments prove that our proposed RPSRMD, which includes RPGen and PResNet as two core components, is superior to many state-of-the-art SR methods that were designed and trained to handle multi-degradation. |
topic |
Single image super-resolution convolution neural network Gaussian blur multiple degradations |
url |
https://ieeexplore.ieee.org/document/9066958/ |
work_keys_str_mv |
AT wenyiwang robustpriorbasedsingleimagesuperresolutionundermultiplegaussiandegradations AT guangyangwu robustpriorbasedsingleimagesuperresolutionundermultiplegaussiandegradations AT weitongcai robustpriorbasedsingleimagesuperresolutionundermultiplegaussiandegradations AT liaoyuanzeng robustpriorbasedsingleimagesuperresolutionundermultiplegaussiandegradations AT jianwenchen robustpriorbasedsingleimagesuperresolutionundermultiplegaussiandegradations |
_version_ |
1724186553959317504 |