Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations

Single Image Super-Resolution (SISR) aims to generate a high-resolution (HR) image of a given low-resolution (LR) image. The most of existing convolutional neural network (CNN) based SISR methods usually take an assumption that a LR image is only bicubicly down-sampled version of an HR image. Howeve...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2019-09
Hauptverfasser: Rao, Muhammad Umer, esti, Gian Luca, Micheloni, Christian
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Rao, Muhammad Umer
esti, Gian Luca
Micheloni, Christian
description Single Image Super-Resolution (SISR) aims to generate a high-resolution (HR) image of a given low-resolution (LR) image. The most of existing convolutional neural network (CNN) based SISR methods usually take an assumption that a LR image is only bicubicly down-sampled version of an HR image. However, the true degradation (i.e. the LR image is a bicubicly downsampled, blurred and noisy version of an HR image) of a LR image goes beyond the widely used bicubic assumption, which makes the SISR problem highly ill-posed nature of inverse problems. To address this issue, we propose a deep SISR network that works for blur kernels of different sizes, and different noise levels in an unified residual CNN-based denoiser network, which significantly improves a practical CNN-based super-resolver for real applications. Extensive experimental results on synthetic LR datasets and real images demonstrate that our proposed method not only can produce better results on more realistic degradation but also computational efficient to practical SISR applications.
doi_str_mv 10.48550/arxiv.1909.03748
format Article
fullrecord <record><control><sourceid>proquest_arxiv</sourceid><recordid>TN_cdi_arxiv_primary_1909_03748</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2287844426</sourcerecordid><originalsourceid>FETCH-LOGICAL-a526-5f11cd70a452c6150c6756a29ca1c59a1417e5779b5b8715d67adb1fac1599383</originalsourceid><addsrcrecordid>eNplkEtPAjEcxBsTEwnyATzZxPNiX_8-jgZUSIgmwM3DpnS7WFzYtd0V_fby8OZpDvPLZGYQuqFkKDQAubfxO3wNqSFmSLgS-gL1GOc004KxKzRIaUMIYVIxAN5Db2PvG7zoGh-zuU911bWh3uEX3-7r-IHLOuJF2K0rj6dbu_b_yX1o3_Hc2yqkNjg89utoC3u00jW6LG2V_OBP-2j59LgcTbLZ6_N09DDLLDCZQUmpKxSxApiTFIiTCqRlxlnqwFgqqPKglFnBSisKhVS2WNHSOgrGcM376PYce1qeNzFsbfzJjw_kpwcOxN2ZaGL92fnU5pu6i7tDp5wxrbQQgkn-CwbHXnM</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2287844426</pqid></control><display><type>article</type><title>Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations</title><source>arXiv.org</source><source>Free E- Journals</source><creator>Rao, Muhammad Umer ; esti, Gian Luca ; Micheloni, Christian</creator><creatorcontrib>Rao, Muhammad Umer ; esti, Gian Luca ; Micheloni, Christian</creatorcontrib><description>Single Image Super-Resolution (SISR) aims to generate a high-resolution (HR) image of a given low-resolution (LR) image. The most of existing convolutional neural network (CNN) based SISR methods usually take an assumption that a LR image is only bicubicly down-sampled version of an HR image. However, the true degradation (i.e. the LR image is a bicubicly downsampled, blurred and noisy version of an HR image) of a LR image goes beyond the widely used bicubic assumption, which makes the SISR problem highly ill-posed nature of inverse problems. To address this issue, we propose a deep SISR network that works for blur kernels of different sizes, and different noise levels in an unified residual CNN-based denoiser network, which significantly improves a practical CNN-based super-resolver for real applications. Extensive experimental results on synthetic LR datasets and real images demonstrate that our proposed method not only can produce better results on more realistic degradation but also computational efficient to practical SISR applications.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.1909.03748</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Artificial neural networks ; Computer Science - Computer Vision and Pattern Recognition ; Ill posed problems ; Image degradation ; Image resolution ; Inverse problems ; Noise levels ; Resolvers</subject><ispartof>arXiv.org, 2019-09</ispartof><rights>2019. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,780,881,27904</link.rule.ids><backlink>$$Uhttps://doi.org/10.48550/arXiv.1909.03748$$DView paper in arXiv$$Hfree_for_read</backlink><backlink>$$Uhttps://doi.org/10.1145/3349801.3349823$$DView published paper (Access to full text may be restricted)$$Hfree_for_read</backlink></links><search><creatorcontrib>Rao, Muhammad Umer</creatorcontrib><creatorcontrib>esti, Gian Luca</creatorcontrib><creatorcontrib>Micheloni, Christian</creatorcontrib><title>Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations</title><title>arXiv.org</title><description>Single Image Super-Resolution (SISR) aims to generate a high-resolution (HR) image of a given low-resolution (LR) image. The most of existing convolutional neural network (CNN) based SISR methods usually take an assumption that a LR image is only bicubicly down-sampled version of an HR image. However, the true degradation (i.e. the LR image is a bicubicly downsampled, blurred and noisy version of an HR image) of a LR image goes beyond the widely used bicubic assumption, which makes the SISR problem highly ill-posed nature of inverse problems. To address this issue, we propose a deep SISR network that works for blur kernels of different sizes, and different noise levels in an unified residual CNN-based denoiser network, which significantly improves a practical CNN-based super-resolver for real applications. Extensive experimental results on synthetic LR datasets and real images demonstrate that our proposed method not only can produce better results on more realistic degradation but also computational efficient to practical SISR applications.</description><subject>Artificial neural networks</subject><subject>Computer Science - Computer Vision and Pattern Recognition</subject><subject>Ill posed problems</subject><subject>Image degradation</subject><subject>Image resolution</subject><subject>Inverse problems</subject><subject>Noise levels</subject><subject>Resolvers</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><sourceid>GOX</sourceid><recordid>eNplkEtPAjEcxBsTEwnyATzZxPNiX_8-jgZUSIgmwM3DpnS7WFzYtd0V_fby8OZpDvPLZGYQuqFkKDQAubfxO3wNqSFmSLgS-gL1GOc004KxKzRIaUMIYVIxAN5Db2PvG7zoGh-zuU911bWh3uEX3-7r-IHLOuJF2K0rj6dbu_b_yX1o3_Hc2yqkNjg89utoC3u00jW6LG2V_OBP-2j59LgcTbLZ6_N09DDLLDCZQUmpKxSxApiTFIiTCqRlxlnqwFgqqPKglFnBSisKhVS2WNHSOgrGcM376PYce1qeNzFsbfzJjw_kpwcOxN2ZaGL92fnU5pu6i7tDp5wxrbQQgkn-CwbHXnM</recordid><startdate>20190909</startdate><enddate>20190909</enddate><creator>Rao, Muhammad Umer</creator><creator>esti, Gian Luca</creator><creator>Micheloni, Christian</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20190909</creationdate><title>Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations</title><author>Rao, Muhammad Umer ; esti, Gian Luca ; Micheloni, Christian</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a526-5f11cd70a452c6150c6756a29ca1c59a1417e5779b5b8715d67adb1fac1599383</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Artificial neural networks</topic><topic>Computer Science - Computer Vision and Pattern Recognition</topic><topic>Ill posed problems</topic><topic>Image degradation</topic><topic>Image resolution</topic><topic>Inverse problems</topic><topic>Noise levels</topic><topic>Resolvers</topic><toplevel>online_resources</toplevel><creatorcontrib>Rao, Muhammad Umer</creatorcontrib><creatorcontrib>esti, Gian Luca</creatorcontrib><creatorcontrib>Micheloni, Christian</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection><collection>arXiv Computer Science</collection><collection>arXiv.org</collection><jtitle>arXiv.org</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Rao, Muhammad Umer</au><au>esti, Gian Luca</au><au>Micheloni, Christian</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations</atitle><jtitle>arXiv.org</jtitle><date>2019-09-09</date><risdate>2019</risdate><eissn>2331-8422</eissn><abstract>Single Image Super-Resolution (SISR) aims to generate a high-resolution (HR) image of a given low-resolution (LR) image. The most of existing convolutional neural network (CNN) based SISR methods usually take an assumption that a LR image is only bicubicly down-sampled version of an HR image. However, the true degradation (i.e. the LR image is a bicubicly downsampled, blurred and noisy version of an HR image) of a LR image goes beyond the widely used bicubic assumption, which makes the SISR problem highly ill-posed nature of inverse problems. To address this issue, we propose a deep SISR network that works for blur kernels of different sizes, and different noise levels in an unified residual CNN-based denoiser network, which significantly improves a practical CNN-based super-resolver for real applications. Extensive experimental results on synthetic LR datasets and real images demonstrate that our proposed method not only can produce better results on more realistic degradation but also computational efficient to practical SISR applications.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.1909.03748</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2019-09
issn 2331-8422
language eng
recordid cdi_arxiv_primary_1909_03748
source arXiv.org; Free E- Journals
subjects Artificial neural networks
Computer Science - Computer Vision and Pattern Recognition
Ill posed problems
Image degradation
Image resolution
Inverse problems
Noise levels
Resolvers
title Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-26T13%3A17%3A44IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_arxiv&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Deep%20Super-Resolution%20Network%20for%20Single%20Image%20Super-Resolution%20with%20Realistic%20Degradations&rft.jtitle=arXiv.org&rft.au=Rao,%20Muhammad%20Umer&rft.date=2019-09-09&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.1909.03748&rft_dat=%3Cproquest_arxiv%3E2287844426%3C/proquest_arxiv%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2287844426&rft_id=info:pmid/&rfr_iscdi=true