All-in-One Image Restoration with Quaternion Cross-Attention and Haar Frequency Decomposition

All-in-One Image Restoration with Quaternion Cross-Attention and Haar Frequency Decomposition | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article All-in-One Image Restoration with Quaternion Cross-Attention and Haar Frequency Decomposition An Hung Nguyen, Minh Tuan Pham This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7260641/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Image restoration is a critical task in computer vision, addressing various degradations like Gaussian noise, haze, and rain that compromise image quality and hinder downstream applications. Traditional methods often treat each degradation separately, limiting their effectiveness in complex, real-world scenarios where multiple degradations coexist. While recent "all-in-one" models attempt unified restoration, they still struggle to fully capture inter-channel relationships and leverage frequency-domain information for optimal detail recovery. To overcome these limitations, we propose QuaHaarIR, a novel all-in-one image restoration model. QuaHaarIR employs a quaternion representation for effective inter-channel correlation modeling and integrates frequency-aware modules to precisely handle high- and low-frequency components. Our architecture incorporates the Quaternion Haar Transform for frequency decomposition, Quaternion Channel-Wise Cross Attention (QCCA) for adaptive feature selection, and a Hypernetwork-based Transformer Encoder for dynamic, degradation-aware behavior. Extensive experiments demonstrate QuaHaarIR's superior generalization and state-of-the-art performance across diverse image restoration tasks, eliminating the need for prior degradation classification. Image Restoration Transformer Quaternion Cross-Attention Haar Wavelet All-in-One model Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7260641","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":498203656,"identity":"68feadf6-a152-463e-a7fe-784a8bf85639","order_by":0,"name":"An Hung Nguyen","email":"","orcid":"","institution":"The University of Da Nang – University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"An","middleName":"Hung","lastName":"Nguyen","suffix":""},{"id":498203657,"identity":"852b7481-ea2e-4d86-9dc5-0ce1182bad21","order_by":1,"name":"Minh Tuan Pham","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAsUlEQVRIiWNgGAWjYPACG8YGEMVDnGpmEJHG2MBGopbDJGgxFzt/TOLnjvOyG+43sEm83cGQuJ2QFsvZyWySvWduG284xsAmOfcMQ+LOBgJaDG4ns0nwtt1OBGmR5m1jMDY4QIQWyb9t50jUAlR5AK5FjqAWoF+MrWXbko1nHktstpzbJkFYi7l04sObb9vsZPsOHz54422bDQ9hhyGY4AQgQUA9qpZRMApGwSgYBTgAAO/mPNPaNsvwAAAAAElFTkSuQmCC","orcid":"","institution":"The University of Da Nang – University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Minh","middleName":"Tuan","lastName":"Pham","suffix":""}],"badges":[],"createdAt":"2025-07-31 09:38:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7260641/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7260641/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94473578,"identity":"bfd3a9c4-c7f9-4a59-bce7-5ba63ad3d7f7","added_by":"auto","created_at":"2025-10-27 15:44:54","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6224133,"visible":true,"origin":"","legend":"","description":"","filename":"20250801QuaHaarIRSpringerJournal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7260641/v1_covered_5412c588-83a8-4382-9d78-6836a98f2236.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"All-in-One Image Restoration with Quaternion Cross-Attention and Haar Frequency Decomposition","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Image Restoration, Transformer, Quaternion Cross-Attention, Haar Wavelet, All-in-One model","lastPublishedDoi":"10.21203/rs.3.rs-7260641/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7260641/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eImage restoration is a critical task in computer vision, addressing various degradations like Gaussian noise, haze, and rain that compromise image quality and hinder downstream applications. Traditional methods often treat each degradation separately, limiting their effectiveness in complex, real-world scenarios where multiple degradations coexist. While recent \"all-in-one\" models attempt unified restoration, they still struggle to fully capture inter-channel relationships and leverage frequency-domain information for optimal detail recovery. To overcome these limitations, we propose QuaHaarIR, a novel all-in-one image restoration model. QuaHaarIR employs a quaternion representation for effective inter-channel correlation modeling and integrates frequency-aware modules to precisely handle high- and low-frequency components. Our architecture incorporates the Quaternion Haar Transform for frequency decomposition, Quaternion Channel-Wise Cross Attention (QCCA) for adaptive feature selection, and a Hypernetwork-based Transformer Encoder for dynamic, degradation-aware behavior. Extensive experiments demonstrate QuaHaarIR's superior generalization and state-of-the-art performance across diverse image restoration tasks, eliminating the need for prior degradation classification. \u003c/p\u003e","manuscriptTitle":"All-in-One Image Restoration with Quaternion Cross-Attention and Haar Frequency Decomposition","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-14 14:10:50","doi":"10.21203/rs.3.rs-7260641/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0fec287c-4eab-4711-b636-34c45c5ce9a2","owner":[],"postedDate":"August 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-10-27T14:26:17+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-14 14:10:50","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7260641","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7260641","identity":"rs-7260641","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}