Loading…

Single Image Denoising via a New Lightweight Learning-Based Model

Restoring a high-quality image from a noisy version poses a significant challenge in computer vision, particularly in today's context where high-resolution and large-sized images are prevalent. As such, fast and efficient techniques are required to address noise reduction in such images effecti...

Full description

Saved in:

Bibliographic Details
Published in:	IEEE access 2024, Vol.12, p.121077-121092
Main Authors:	Rezvani, Sadjad, Soleymani Siahkar, Fatemeh, Rezvani, Yasin, Alavi Gharahbagh, Abdorreza, Abolghasemi, Vahid
Format:	Article
Language:	English
Subjects:	Deep learning downsampling Image denoising Noise level Noise measurement Noise reduction self-supervision Training Transformers Zero-shot learning
Citations:	Items that this one cites
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

cited_by
cites	cdi_FETCH-LOGICAL-c1762-7621c5241a1faf594da520b7deeafc0d4b5a234006bf4789124b1665708da2cb3
container_end_page	121092
container_issue
container_start_page	121077
container_title	IEEE access
container_volume	12
creator	Rezvani, Sadjad Soleymani Siahkar, Fatemeh Rezvani, Yasin Alavi Gharahbagh, Abdorreza Abolghasemi, Vahid
description	Restoring a high-quality image from a noisy version poses a significant challenge in computer vision, particularly in today's context where high-resolution and large-sized images are prevalent. As such, fast and efficient techniques are required to address noise reduction in such images effectively. Deep CNN-based image-denoising algorithms have gained popularity due to the rapid growth of deep learning and convolutional neural networks (CNNs). However, many existing deep learning models require paired clean/noisy images for training, limiting their utility in real-world denoising scenarios. In this paper, we propose a fast residual denoising framework (FRDF) designed based on zero-shot learning to address this issue. The FRDF first employs a novel downsampling technique to generate six different images from the noisy input, which are then fed into a lightweight residual network with 23K parameters. The network effectively utilizes a hybrid loss function, including residual, regularization, and guidance losses, to produce high-quality denoised images. Our innovative downsampling approach incorporates zero-shot learning principles, enabling our framework to generalize to unseen noise types and adapt to diverse noise conditions without needing labelled data. Extensive experiments conducted on synthetic and real images confirm the superiority of our proposed approach over existing dataset-free methods. Extensive experiments conducted on synthetic and real images show that our method achieves up to 2 dB improvements in PSNR on the McMaster and Kodak24 datasets. This renders our approach applicable in scenarios with limited data availability and computational resources.
doi_str_mv	10.1109/ACCESS.2024.3450842
format	article
fullrecord	<record><control><sourceid>doaj_ieee_</sourceid><recordid>TN_cdi_ieee_primary_10654252</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>10654252</ieee_id><doaj_id>oai_doaj_org_article_87107460d3b94aff933c82501a5c1f27</doaj_id><sourcerecordid>oai_doaj_org_article_87107460d3b94aff933c82501a5c1f27</sourcerecordid><originalsourceid>FETCH-LOGICAL-c1762-7621c5241a1faf594da520b7deeafc0d4b5a234006bf4789124b1665708da2cb3</originalsourceid><addsrcrecordid>eNpNkN1Kw0AQhRdRsGifQC_yAqn7v8lljVULUS-q18tkdzZuSRtJisW3NzVFOjA_nOF8F4eQG0ZnjNH8bl4Ui9VqximXMyEVzSQ_IxPOdJ4KJfT5yX1Jpn2_pkNlg6TMhMxXcVs3mCw3UGPygNs29oOSfEdIIHnFfVLG-nO3x8NMSoRuO7zTe-jRJy-tx-aaXARoepwe9xX5eFy8F89p-fa0LOZl6pjRPB2aOcUlAxYgqFx6UJxWxiNCcNTLSgEXklJdBWmynHFZMa2VoZkH7ipxRZYj17ewtl9d3ED3Y1uI9k9ou9pCt4uuQZsZRo3U1IsqlxBCLoTLuKIMlGOBm4ElRpbr2r7vMPzzGLWHUO0Yqj2Eao-hDq7b0RUR8cShleSKi18HGnDL</addsrcrecordid><sourcetype>Open Website</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Single Image Denoising via a New Lightweight Learning-Based Model</title><source>IEEE Xplore Open Access Journals</source><creator>Rezvani, Sadjad ; Soleymani Siahkar, Fatemeh ; Rezvani, Yasin ; Alavi Gharahbagh, Abdorreza ; Abolghasemi, Vahid</creator><creatorcontrib>Rezvani, Sadjad ; Soleymani Siahkar, Fatemeh ; Rezvani, Yasin ; Alavi Gharahbagh, Abdorreza ; Abolghasemi, Vahid</creatorcontrib><description>Restoring a high-quality image from a noisy version poses a significant challenge in computer vision, particularly in today's context where high-resolution and large-sized images are prevalent. As such, fast and efficient techniques are required to address noise reduction in such images effectively. Deep CNN-based image-denoising algorithms have gained popularity due to the rapid growth of deep learning and convolutional neural networks (CNNs). However, many existing deep learning models require paired clean/noisy images for training, limiting their utility in real-world denoising scenarios. In this paper, we propose a fast residual denoising framework (FRDF) designed based on zero-shot learning to address this issue. The FRDF first employs a novel downsampling technique to generate six different images from the noisy input, which are then fed into a lightweight residual network with 23K parameters. The network effectively utilizes a hybrid loss function, including residual, regularization, and guidance losses, to produce high-quality denoised images. Our innovative downsampling approach incorporates zero-shot learning principles, enabling our framework to generalize to unseen noise types and adapt to diverse noise conditions without needing labelled data. Extensive experiments conducted on synthetic and real images confirm the superiority of our proposed approach over existing dataset-free methods. Extensive experiments conducted on synthetic and real images show that our method achieves up to 2 dB improvements in PSNR on the McMaster and Kodak24 datasets. This renders our approach applicable in scenarios with limited data availability and computational resources.</description><identifier>ISSN: 2169-3536</identifier><identifier>EISSN: 2169-3536</identifier><identifier>DOI: 10.1109/ACCESS.2024.3450842</identifier><identifier>CODEN: IAECCG</identifier><language>eng</language><publisher>IEEE</publisher><subject>Deep learning ; downsampling ; Image denoising ; Noise level ; Noise measurement ; Noise reduction ; self-supervision ; Training ; Transformers ; Zero-shot learning</subject><ispartof>IEEE access, 2024, Vol.12, p.121077-121092</ispartof><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c1762-7621c5241a1faf594da520b7deeafc0d4b5a234006bf4789124b1665708da2cb3</cites><orcidid>0009-0000-5086-127X ; 0000-0003-0863-1977 ; 0000-0002-2151-5180 ; 0000-0002-3779-0726</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/10654252$$EHTML$$P50$$Gieee$$Hfree_for_read</linktohtml><link.rule.ids>315,786,790,4043,27666,27956,27957,27958,55284</link.rule.ids></links><search><creatorcontrib>Rezvani, Sadjad</creatorcontrib><creatorcontrib>Soleymani Siahkar, Fatemeh</creatorcontrib><creatorcontrib>Rezvani, Yasin</creatorcontrib><creatorcontrib>Alavi Gharahbagh, Abdorreza</creatorcontrib><creatorcontrib>Abolghasemi, Vahid</creatorcontrib><title>Single Image Denoising via a New Lightweight Learning-Based Model</title><title>IEEE access</title><addtitle>Access</addtitle><description>Restoring a high-quality image from a noisy version poses a significant challenge in computer vision, particularly in today's context where high-resolution and large-sized images are prevalent. As such, fast and efficient techniques are required to address noise reduction in such images effectively. Deep CNN-based image-denoising algorithms have gained popularity due to the rapid growth of deep learning and convolutional neural networks (CNNs). However, many existing deep learning models require paired clean/noisy images for training, limiting their utility in real-world denoising scenarios. In this paper, we propose a fast residual denoising framework (FRDF) designed based on zero-shot learning to address this issue. The FRDF first employs a novel downsampling technique to generate six different images from the noisy input, which are then fed into a lightweight residual network with 23K parameters. The network effectively utilizes a hybrid loss function, including residual, regularization, and guidance losses, to produce high-quality denoised images. Our innovative downsampling approach incorporates zero-shot learning principles, enabling our framework to generalize to unseen noise types and adapt to diverse noise conditions without needing labelled data. Extensive experiments conducted on synthetic and real images confirm the superiority of our proposed approach over existing dataset-free methods. Extensive experiments conducted on synthetic and real images show that our method achieves up to 2 dB improvements in PSNR on the McMaster and Kodak24 datasets. This renders our approach applicable in scenarios with limited data availability and computational resources.</description><subject>Deep learning</subject><subject>downsampling</subject><subject>Image denoising</subject><subject>Noise level</subject><subject>Noise measurement</subject><subject>Noise reduction</subject><subject>self-supervision</subject><subject>Training</subject><subject>Transformers</subject><subject>Zero-shot learning</subject><issn>2169-3536</issn><issn>2169-3536</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>ESBDL</sourceid><sourceid>DOA</sourceid><recordid>eNpNkN1Kw0AQhRdRsGifQC_yAqn7v8lljVULUS-q18tkdzZuSRtJisW3NzVFOjA_nOF8F4eQG0ZnjNH8bl4Ui9VqximXMyEVzSQ_IxPOdJ4KJfT5yX1Jpn2_pkNlg6TMhMxXcVs3mCw3UGPygNs29oOSfEdIIHnFfVLG-nO3x8NMSoRuO7zTe-jRJy-tx-aaXARoepwe9xX5eFy8F89p-fa0LOZl6pjRPB2aOcUlAxYgqFx6UJxWxiNCcNTLSgEXklJdBWmynHFZMa2VoZkH7ipxRZYj17ewtl9d3ED3Y1uI9k9ou9pCt4uuQZsZRo3U1IsqlxBCLoTLuKIMlGOBm4ElRpbr2r7vMPzzGLWHUO0Yqj2Eao-hDq7b0RUR8cShleSKi18HGnDL</recordid><startdate>2024</startdate><enddate>2024</enddate><creator>Rezvani, Sadjad</creator><creator>Soleymani Siahkar, Fatemeh</creator><creator>Rezvani, Yasin</creator><creator>Alavi Gharahbagh, Abdorreza</creator><creator>Abolghasemi, Vahid</creator><general>IEEE</general><scope>97E</scope><scope>ESBDL</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>DOA</scope><orcidid>https://orcid.org/0009-0000-5086-127X</orcidid><orcidid>https://orcid.org/0000-0003-0863-1977</orcidid><orcidid>https://orcid.org/0000-0002-2151-5180</orcidid><orcidid>https://orcid.org/0000-0002-3779-0726</orcidid></search><sort><creationdate>2024</creationdate><title>Single Image Denoising via a New Lightweight Learning-Based Model</title><author>Rezvani, Sadjad ; Soleymani Siahkar, Fatemeh ; Rezvani, Yasin ; Alavi Gharahbagh, Abdorreza ; Abolghasemi, Vahid</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c1762-7621c5241a1faf594da520b7deeafc0d4b5a234006bf4789124b1665708da2cb3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Deep learning</topic><topic>downsampling</topic><topic>Image denoising</topic><topic>Noise level</topic><topic>Noise measurement</topic><topic>Noise reduction</topic><topic>self-supervision</topic><topic>Training</topic><topic>Transformers</topic><topic>Zero-shot learning</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Rezvani, Sadjad</creatorcontrib><creatorcontrib>Soleymani Siahkar, Fatemeh</creatorcontrib><creatorcontrib>Rezvani, Yasin</creatorcontrib><creatorcontrib>Alavi Gharahbagh, Abdorreza</creatorcontrib><creatorcontrib>Abolghasemi, Vahid</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE Xplore Open Access Journals</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998–Present</collection><collection>IEEE Xplore</collection><collection>CrossRef</collection><collection>DOAJ Directory of Open Access Journals</collection><jtitle>IEEE access</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Rezvani, Sadjad</au><au>Soleymani Siahkar, Fatemeh</au><au>Rezvani, Yasin</au><au>Alavi Gharahbagh, Abdorreza</au><au>Abolghasemi, Vahid</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Single Image Denoising via a New Lightweight Learning-Based Model</atitle><jtitle>IEEE access</jtitle><stitle>Access</stitle><date>2024</date><risdate>2024</risdate><volume>12</volume><spage>121077</spage><epage>121092</epage><pages>121077-121092</pages><issn>2169-3536</issn><eissn>2169-3536</eissn><coden>IAECCG</coden><abstract>Restoring a high-quality image from a noisy version poses a significant challenge in computer vision, particularly in today's context where high-resolution and large-sized images are prevalent. As such, fast and efficient techniques are required to address noise reduction in such images effectively. Deep CNN-based image-denoising algorithms have gained popularity due to the rapid growth of deep learning and convolutional neural networks (CNNs). However, many existing deep learning models require paired clean/noisy images for training, limiting their utility in real-world denoising scenarios. In this paper, we propose a fast residual denoising framework (FRDF) designed based on zero-shot learning to address this issue. The FRDF first employs a novel downsampling technique to generate six different images from the noisy input, which are then fed into a lightweight residual network with 23K parameters. The network effectively utilizes a hybrid loss function, including residual, regularization, and guidance losses, to produce high-quality denoised images. Our innovative downsampling approach incorporates zero-shot learning principles, enabling our framework to generalize to unseen noise types and adapt to diverse noise conditions without needing labelled data. Extensive experiments conducted on synthetic and real images confirm the superiority of our proposed approach over existing dataset-free methods. Extensive experiments conducted on synthetic and real images show that our method achieves up to 2 dB improvements in PSNR on the McMaster and Kodak24 datasets. This renders our approach applicable in scenarios with limited data availability and computational resources.</abstract><pub>IEEE</pub><doi>10.1109/ACCESS.2024.3450842</doi><tpages>16</tpages><orcidid>https://orcid.org/0009-0000-5086-127X</orcidid><orcidid>https://orcid.org/0000-0003-0863-1977</orcidid><orcidid>https://orcid.org/0000-0002-2151-5180</orcidid><orcidid>https://orcid.org/0000-0002-3779-0726</orcidid><oa>free_for_read</oa></addata></record>
fulltext	fulltext
identifier	ISSN: 2169-3536
ispartof	IEEE access, 2024, Vol.12, p.121077-121092
issn	2169-3536 2169-3536
language	eng
recordid	cdi_ieee_primary_10654252
source	IEEE Xplore Open Access Journals
subjects	Deep learning downsampling Image denoising Noise level Noise measurement Noise reduction self-supervision Training Transformers Zero-shot learning
title	Single Image Denoising via a New Lightweight Learning-Based Model
url	http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-09-22T13%3A33%3A37IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-doaj_ieee_&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Single%20Image%20Denoising%20via%20a%20New%20Lightweight%20Learning-Based%20Model&rft.jtitle=IEEE%20access&rft.au=Rezvani,%20Sadjad&rft.date=2024&rft.volume=12&rft.spage=121077&rft.epage=121092&rft.pages=121077-121092&rft.issn=2169-3536&rft.eissn=2169-3536&rft.coden=IAECCG&rft_id=info:doi/10.1109/ACCESS.2024.3450842&rft_dat=%3Cdoaj_ieee_%3Eoai_doaj_org_article_87107460d3b94aff933c82501a5c1f27%3C/doaj_ieee_%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c1762-7621c5241a1faf594da520b7deeafc0d4b5a234006bf4789124b1665708da2cb3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=10654252&rfr_iscdi=true