Enhancing Learned Image Compression with Gaussian Mixture Models and Deep Neural Networks

Enhancing Learned Image Compression with Gaussian Mixture Models and Deep Neural Networks | 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 Enhancing Learned Image Compression with Gaussian Mixture Models and Deep Neural Networks Mingyu Zhu, Tianhao Zhao, Ziyi Huang, Hang Zhou This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6282807/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 Traditional image compression standards such as JPEG, JPEG2000, and BPG have achieved notable success, yet struggle to meet the low-latency and adaptive demands of modern wireless transmission. Their fixed transform coding frameworks are ill-suited for dynamic wireless environments. Recent advances in deep learning, particularly Convolutional Neural Networks and Recurrent Neural Networks, have enabled end-to-end nonlinear modeling for improved image compression. This study investigates the integration of CNN- and RNN-based architectures into wireless image transmission systems, targeting two key challenges: reducing perceptual distortion and optimizing computational efficiency. A core contribution lies in introducing Gaussian Mixture Models (GMMs) into these DL frameworks, enabling probabilistic modeling of latent features to support adaptive bit allocation. By comparing with traditional and existing DL-based methods, the proposed approach offers dual optimization in compression performance and channel adaptability. Experimental results show that GMM-enhanced DL models significantly improve robustness and compression quality under fluctuating channel conditions, offering a promising direction for deploying adaptive, efficient image compression schemes in resource-constrained wireless networks. Deep learning Learned image compression Gaussian Mixture Model Rate-distortion optimization 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-6282807","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":433000345,"identity":"53909f71-2dca-4a83-aba6-d4003e52f582","order_by":0,"name":"Mingyu Zhu","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Mingyu","middleName":"","lastName":"Zhu","suffix":""},{"id":433000346,"identity":"17691c28-179a-42f1-8936-58b85d6240d1","order_by":1,"name":"Tianhao Zhao","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Tianhao","middleName":"","lastName":"Zhao","suffix":""},{"id":433000347,"identity":"b7464fe8-8fb9-4f2e-b80c-ad190351b965","order_by":2,"name":"Ziyi Huang","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Ziyi","middleName":"","lastName":"Huang","suffix":""},{"id":433000348,"identity":"f3c62684-be42-499f-a0dc-42e0f8990686","order_by":3,"name":"Hang Zhou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4UlEQVRIiWNgGAWjYBACAxDB2MDAwC/BkABjE6lFcgZDYgNpWgxuQFQT1mLOfvbwy587bBI33254/piHwUZ2wwHmZw/wabHsyUuzkDyTlrjtzoHEZh6GNOMNB9jMDfA67ECOmYFh2+HEbTcSQFoOJ244wMMmgVfL+TdmBolALZtngLX8J0LLjRzjBweBWjZIgLUcIEbLGzPGxrY04xlAv8ycY5BsPPMwmxkBh+UYf/zZZiPbP7sn4cObCjvZvuPNz/BqAQKYM3gSINHETEA9SMkHCM1+gLDaUTAKRsEoGJEAAFT1UlzcCUUIAAAAAElFTkSuQmCC","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":true,"prefix":"","firstName":"Hang","middleName":"","lastName":"Zhou","suffix":""}],"badges":[],"createdAt":"2025-03-22 09:53:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6282807/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6282807/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100357168,"identity":"fbfd5e8f-dcca-4a0f-a331-ac6af67f522c","added_by":"auto","created_at":"2026-01-16 07:19:10","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":648379,"visible":true,"origin":"","legend":"","description":"","filename":"EnhancingLearnedImageCompressionwithGaussianMixtureModelsandDeepNeuralNetworks.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6282807/v1_covered_5d0051e7-fba7-4184-84e6-9eaac82cf513.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Enhancing Learned Image Compression with Gaussian Mixture Models and Deep Neural Networks","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":"Deep learning, Learned image compression, Gaussian Mixture Model, Rate-distortion optimization","lastPublishedDoi":"10.21203/rs.3.rs-6282807/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6282807/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTraditional image compression standards such as JPEG, JPEG2000, and BPG have achieved notable success, yet struggle to meet the low-latency and adaptive demands of modern wireless transmission. Their fixed transform coding frameworks are ill-suited for dynamic wireless environments. Recent advances in deep learning, particularly Convolutional Neural Networks and Recurrent Neural Networks, have enabled end-to-end nonlinear modeling for improved image compression. This study investigates the integration of CNN- and RNN-based architectures into wireless image transmission systems, targeting two key challenges: reducing perceptual distortion and optimizing computational efficiency. A core contribution lies in introducing Gaussian Mixture Models (GMMs) into these DL frameworks, enabling probabilistic modeling of latent features to support adaptive bit allocation. By comparing with traditional and existing DL-based methods, the proposed approach offers dual optimization in compression performance and channel adaptability. Experimental results show that GMM-enhanced DL models significantly improve robustness and compression quality under fluctuating channel conditions, offering a promising direction for deploying adaptive, efficient image compression schemes in resource-constrained wireless networks.\u003c/p\u003e","manuscriptTitle":"Enhancing Learned Image Compression with Gaussian Mixture Models and Deep Neural Networks","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-25 15:49:28","doi":"10.21203/rs.3.rs-6282807/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":"1b5adc43-4831-494a-aa6f-eaaa4b21ba73","owner":[],"postedDate":"March 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-01-09T02:39:01+00:00","versionOfRecord":[],"versionCreatedAt":"2025-03-25 15:49:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6282807","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6282807","identity":"rs-6282807","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}