Image Dehazing Algorithm Based on Deep Transfer Learning and Local Mean Adaptation

Image Dehazing Algorithm Based on Deep Transfer Learning and Local Mean Adaptation | 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 Article Image Dehazing Algorithm Based on Deep Transfer Learning and Local Mean Adaptation Dongyang Shi, Sheng Huang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6502922/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 31 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted 8 You are reading this latest preprint version Abstract Background In recent years, haze has significantly hindered the quality and efficiency of daily tasks, reducing the visual perception range. Various approaches have emerged to address image dehazing, including image enhancement, image restoration, and deep learning-based dehazing methods. While these methods have improved dehazing performance to some extent, they often struggle in bright regions of the image, leading to distortions and suboptimal dehazing results. Moreover, dehazing models generally exhibit weak noise resistance, with the PSNR value of dehazed images typically falling below 30 dB. Residual noise remains in the processed images, leading to degraded visual quality. Currently, it is challenging for dehazing models to simultaneously ensure effective dehazing in bright regions while maintaining strong noise suppression capabilities. Methods To address both issues simultaneously, we propose an image dehazing algorithm based on deep transfer learning and local mean adaptation. The framework consists of several key modules: an atmospheric light estimation module based on deep transfer learning, a transmission map estimation module utilizing local mean adaptation, a haze-free image reconstruction module, an image enhancement module, and a noise reduction module. This design not only ensures the stable and accurate estimation of atmospheric light but also enables the model to effectively process different regions of hazy images, preventing distortion artifacts. Furthermore, to enrich the details of the dehazed images and enhance the dehazing performance while improving the model’s noise resistance, we incorporate an image enhancement module and a noise reduction module into the proposed dehazing framework. Results To validate the effectiveness of the proposed algorithm, we conducted dehazing experiments on a self-constructed hazy dataset, the SOTS (outdoor) dataset, and the NH-HAZE dataset. Experimental results demonstrate that the proposed dehazing model achieves superior performance across all three datasets. The dehazed images exhibit no color distortion, and the PSNR values consistently exceed 30 dB, indicating high-quality dehazed images. The dehazed images also demonstrate a significant advantage in SSIM performance compared to mainstream dehazing algorithms, consistently achieving a similarity of over 85%. This indicates that the proposed dehazing model effectively mitigates distortion while enhancing noise resistance, exhibiting strong generalization capabilities across different datasets. Conclusion The experimental results confirm that the proposed dehazing algorithm effectively handles bright regions such as the sky while significantly improving the model’s noise resistance, reducing residual noise in the dehazed images. Both aspects demonstrate strong performance, validating the effectiveness and superiority of the proposed dehazing model. Furthermore, the algorithm achieves consistently good dehazing performance across all three hazy datasets, demonstrating its generalization capability. This study introduces a novel dehazing method and theoretical framework, which can be effectively applied to scenarios such as autonomous driving and intelligent surveillance systems. The proposed model provides a new approach to image dehazing, contributing to advancements in related fields and fostering further development in haze removal technologies. Physical sciences/Mathematics and computing/Computer science Physical sciences/Mathematics and computing/Software Physical sciences/Mathematics and computing/Information technology Image Dehazing Transfer Learning Local Mean Image Enhancement Image Denoising Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 31 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted Reviews received at journal 21 May, 2025 Reviewers agreed at journal 20 May, 2025 Reviewers agreed at journal 20 May, 2025 Reviewers invited by journal 20 May, 2025 Editor assigned by journal 05 May, 2025 Editor invited by journal 29 Apr, 2025 Submission checks completed at journal 29 Apr, 2025 First submitted to journal 22 Apr, 2025 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-6502922","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":459753208,"identity":"652c5cc1-98bf-45f5-a4b4-04ffae783729","order_by":0,"name":"Dongyang Shi","email":"","orcid":"","institution":"Chongqing University of Posts and Telecommunications","correspondingAuthor":false,"prefix":"","firstName":"Dongyang","middleName":"","lastName":"Shi","suffix":""},{"id":459753209,"identity":"619cc3fa-5132-4ed7-80ec-533b8078d399","order_by":1,"name":"Sheng Huang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIiWNgGAWjYBACef7G5sc/Ktjk7NubDxCnxXDG4TZjhjN8xgY8xxKItOZAeoM0Y5tc4gaJHAPidDA2HGwwLmwzMzaXyPl44w2DnZxuAwEt7MyNDY9nnEuTs+x5u9lyDkOysdkBImwx4Ck7ZsxwPHebNA/DgcRthLQA1TRI8LD9T2w4kPOMeC3SPG1siRtO5LARp8VwxsE2wxln2Iwle44ZW84xIMIv8vztjx98AEYlP3vzwxtvKuzkCGpBARI8REYNshZSdYyCUTAKRsGIAACwlUh9kYelXQAAAABJRU5ErkJggg==","orcid":"","institution":"Chongqing University of Posts and Telecommunications","correspondingAuthor":true,"prefix":"","firstName":"Sheng","middleName":"","lastName":"Huang","suffix":""}],"badges":[],"createdAt":"2025-04-22 09:53:36","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6502922/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6502922/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-13613-z","type":"published","date":"2025-07-31T16:12:52+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":88268094,"identity":"690c4c29-716b-48c6-a5b6-e1e035b26004","added_by":"auto","created_at":"2025-08-04 16:48:49","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2029950,"visible":true,"origin":"","legend":"","description":"","filename":"ImageDehazingAlgorithmBasedonDeepTransferLearningandLocalMeanAdaptation.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6502922/v1_covered_7640b259-74a3-4040-8f97-7f2325a4da7a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Image Dehazing Algorithm Based on Deep Transfer Learning and Local Mean Adaptation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Image Dehazing, Transfer Learning, Local Mean, Image Enhancement, Image Denoising","lastPublishedDoi":"10.21203/rs.3.rs-6502922/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6502922/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eIn recent years, haze has significantly hindered the quality and efficiency of daily tasks, reducing the visual perception range. Various approaches have emerged to address image dehazing, including image enhancement, image restoration, and deep learning-based dehazing methods. While these methods have improved dehazing performance to some extent, they often struggle in bright regions of the image, leading to distortions and suboptimal dehazing results. Moreover, dehazing models generally exhibit weak noise resistance, with the PSNR value of dehazed images typically falling below 30 dB. Residual noise remains in the processed images, leading to degraded visual quality. Currently, it is challenging for dehazing models to simultaneously ensure effective dehazing in bright regions while maintaining strong noise suppression capabilities.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eTo address both issues simultaneously, we propose an image dehazing algorithm based on deep transfer learning and local mean adaptation. The framework consists of several key modules: an atmospheric light estimation module based on deep transfer learning, a transmission map estimation module utilizing local mean adaptation, a haze-free image reconstruction module, an image enhancement module, and a noise reduction module. This design not only ensures the stable and accurate estimation of atmospheric light but also enables the model to effectively process different regions of hazy images, preventing distortion artifacts. Furthermore, to enrich the details of the dehazed images and enhance the dehazing performance while improving the model\u0026rsquo;s noise resistance, we incorporate an image enhancement module and a noise reduction module into the proposed dehazing framework.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eTo validate the effectiveness of the proposed algorithm, we conducted dehazing experiments on a self-constructed hazy dataset, the SOTS (outdoor) dataset, and the NH-HAZE dataset. Experimental results demonstrate that the proposed dehazing model achieves superior performance across all three datasets. The dehazed images exhibit no color distortion, and the PSNR values consistently exceed 30 dB, indicating high-quality dehazed images. The dehazed images also demonstrate a significant advantage in SSIM performance compared to mainstream dehazing algorithms, consistently achieving a similarity of over 85%. This indicates that the proposed dehazing model effectively mitigates distortion while enhancing noise resistance, exhibiting strong generalization capabilities across different datasets.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThe experimental results confirm that the proposed dehazing algorithm effectively handles bright regions such as the sky while significantly improving the model\u0026rsquo;s noise resistance, reducing residual noise in the dehazed images. Both aspects demonstrate strong performance, validating the effectiveness and superiority of the proposed dehazing model. Furthermore, the algorithm achieves consistently good dehazing performance across all three hazy datasets, demonstrating its generalization capability. This study introduces a novel dehazing method and theoretical framework, which can be effectively applied to scenarios such as autonomous driving and intelligent surveillance systems. The proposed model provides a new approach to image dehazing, contributing to advancements in related fields and fostering further development in haze removal technologies.\u003c/p\u003e","manuscriptTitle":"Image Dehazing Algorithm Based on Deep Transfer Learning and Local Mean Adaptation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-22 09:16:22","doi":"10.21203/rs.3.rs-6502922/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2025-05-21T08:17:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"75836756830766825305204961125880289026","date":"2025-05-20T15:56:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"212661228501369032190876369949913808990","date":"2025-05-20T15:23:20+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-20T14:16:57+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-05T14:31:07+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-04-29T11:37:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-04-29T07:30:13+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-04-22T09:44:23+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"861b0a6b-0611-4fc7-a25d-a3b4bdf19ace","owner":[],"postedDate":"May 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":48837333,"name":"Physical sciences/Mathematics and computing/Computer science"},{"id":48837334,"name":"Physical sciences/Mathematics and computing/Software"},{"id":48837335,"name":"Physical sciences/Mathematics and computing/Information technology"}],"tags":[],"updatedAt":"2025-08-04T16:38:59+00:00","versionOfRecord":{"articleIdentity":"rs-6502922","link":"https://doi.org/10.1038/s41598-025-13613-z","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-07-31 16:12:52","publishedOnDateReadable":"July 31st, 2025"},"versionCreatedAt":"2025-05-22 09:16:22","video":"","vorDoi":"10.1038/s41598-025-13613-z","vorDoiUrl":"https://doi.org/10.1038/s41598-025-13613-z","workflowStages":[]},"version":"v1","identity":"rs-6502922","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6502922","identity":"rs-6502922","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}