[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-512779-105":59,"doc-detail-512779-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","when-awgn-based-denoiser-meets-real-noises","When AWGN-Based Denoiser Meets Real Noises","","Discriminative learning image denoisers perform well on synthetic Additive White Gaussian Noise (AWGN), yet struggle on real noises that are spatially/channel-correlated and spatially/channel-variant, causing a strong domain gap. This work trains a deep denoiser using a noise estimator plus a denoiser with mixed AWGN and Random Value Impulse Noise (RVIN). It further introduces Pixel-shuffle Down-sampling (PD) to adapt the trained model to real noise. Extensive experiments validate effectiveness and generalization, achieving state-of-the-art results on DND real sRGB images when trained with synthetic noises.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/when-awgn-based-denoiser-meets-real-noises/512779/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/when-awgn-based-denoiser-meets-real-noises/512779.png","ImageObject",300,407,{"name":92,"@type":93},"Angel","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-07","2026-10-06",true,{"@type":102,"interactionType":103,"userInteractionCount":44},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why do AWGN-trained denoisers perform poorly on real noisy images?","Question",{"text":112,"@type":113},"Real noises are typically spatially/channel-correlated and spatially/channel-variant, while most prior training uses pixel-independent AWGN. This mismatch creates a domain gap, leading to degraded performance on real images.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does the proposed method improve robustness beyond AWGN?",{"text":117,"@type":113},"It trains a deep model composed of an explicit noise estimator and a conditional denoiser using mixed AWGN and Random Value Impulse Noise (RVIN), making the basis network more flexible for non-AWGN conditions.",{"name":119,"@type":110,"acceptedAnswer":120},"What is Pixel-shuffle Down-sampling (PD) used for?",{"text":121,"@type":113},"PD adapts the trained model to real noises by transforming spatially-correlated real noise into a form closer to spatially-variant but pixel-independent noise, enabling an AWGN-based denoiser to be applied effectively.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},512779,1791403505,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":44,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":39,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":143,"read_time":46,"preview_page_count":19,"preview_urls":144},687207412472,"https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d","When AWGN-Based Denoiser Meets Real Noises  \nYuqian Zhou, 1 Jianbo Jiao,2∗ Haibin Huang,3 Yang Wang,4 Jue Wang,3 Honghui Shi, 1 Thomas Huang 1  \n1IFP Group, UIUC, 2University of Oxford, 3Megvii Research, 4 Stony Brook University  \n{yuqian2, [t-huang1](t-huang1}@illinois.edu)[}](t-huang1}@illinois.edu)[@illinois.edu](t-huang1}@illinois.edu), {jiaojianbo.i, [shihonghui3](shihonghui3}@gmail.com)[}](shihonghui3}@gmail.com)[@gmail.com](shihonghui3}@gmail.com),  \n{huanghaibin, [wangjue](wangjue}@megvii.com)[}](wangjue}@megvii.com)[@megvii.com](wangjue}@megvii.com), [wang33@cs.stonybrook.edu](wang33@cs.stonybrook.edu)  \nAbstract  \nDiscriminative learning based image denoisers have achieved promising performance on synthetic noises such as Additive White Gaussian Noise (AWGN) . The synthetic noises adopted in most previous work are pixel-independent, but real noises are mostly spatially/channel-correlated and spatially/channel-variant. This domain gap yields unsatisﬁed performance on images with real noises if the model is only trained with AWGN. In this paper, we propose a novel approach to boost the performance of a real image denoiser which is trained only with synthetic pixel-independent noise data dominated by AWGN. First, we train a deep model that consists of a noise estimator and a denoiser with mixed AWGN and Random Value Impulse Noise (RVIN) . We then investigate Pixel-shufﬂe Down-sampling (PD) strategy to adapt the trained model to real noises. Extensive experiments demonstrate the effectiveness and generalization of the proposed approach. Notably, our method achieves state-of-theart performance on real sRGB images in the DND benchmark among models trained with synthetic noises. Codes are available at [https://github.com/yzhouas/PD-Denoising-pytorch](https://github.com/yzhouas/PD-Denoising-pytorch).  \nIntroduction  \nAs a fundamental task in image processing and computer vision, image denoising has been extensively explored in the past several decades even for downstream applications (Zhou, Liu, and Huang 2018; Wang et al. 2019) . Traditional methods including the ones based on image ﬁltering (Dabov et al. 2008), low rank approximation (Gu et al. 2014; Xu et al. 2017; Yair and Michaeli 2018), sparse coding (Elad and Aharon 2006), and image prior (Ulyanov, Vedaldi, and Lempitsky 2017) have achieved satisfactory results on synthetic noise such as Additive White Gaussian Noise (AWGN) . Recently, deep CNN has been applied to this task, and discriminative-learning-based methods such as DnCNN (Zhang et al. 2017a) outperform most traditional methods on AWGN denoising.  \nUnfortunately, while these learning-based methods work well on the same type of synthetic noise that they were  \n∗ Corresponding author  \nCopyright 􀀂c 2020, Association for the Advancement of Artiﬁcial Intelligence ([www.aaai.org](www.aaai.org)). All rights reserved.  \nFigure 1: Basic idea of the proposed adaptation method: Pixel-shufﬂe Down-sampling (PD) . Spatially-correlated real noise (Left) is broken into spatially-variant pixelindependent noise (Middle) to approximate spatially-variant Gaussian noise (Right) . Then an AWGN-based denoiser can be applied to such real noise accordingly.  \ntrained on, their performance degrades rapidly on real images, showing poor generalization ability in real world applications. This indicates that these data-driven denoising models are highly domain-speciﬁc and non-ﬂexible to transfer to other noise types beyond AWGN. To improve model ﬂexibility, the recently-proposed FFDNet (Zhang, Zuo, and Zhang 2018) trains a conditional non-blind denoiser with a manually adjusted noise-level map. By giving high-valued uniform maps to FFDNet, only over-smoothed results can be obtained in real image denoising. Therefore, blind denoising of real images is still very challenging due to the lack of accurate modeling of real noise distribution. These unknown real-world noises are much more complex than pixel-independent AWGN. They can be spatiall","cbCaidEj1hrbpBoH","https://ap.wps.com/l/cbCaidEj1hrbpBoH","pdf",4332420,"English","# Abstract\n# Introduction\n## Background on denoising and domain gap\n## Existing approaches to real-image denoising\n# Proposed method\n## Training with mixed AWGN and RVIN\n## PD adaptation strategy","[{\"question\":\"Why do AWGN-trained denoisers perform poorly on real noisy images?\",\"answer\":\"Real noises are typically spatially/channel-correlated and spatially/channel-variant, while most prior training uses pixel-independent AWGN. This mismatch creates a domain gap, leading to degraded performance on real images.\"},{\"question\":\"How does the proposed method improve robustness beyond AWGN?\",\"answer\":\"It trains a deep model composed of an explicit noise estimator and a conditional denoiser using mixed AWGN and Random Value Impulse Noise (RVIN), making the basis network more flexible for non-AWGN conditions.\"},{\"question\":\"What is Pixel-shuffle Down-sampling (PD) used for?\",\"answer\":\"PD adapts the trained model to real noises by transforming spatially-correlated real noise into a form closer to spatially-variant but pixel-independent noise, enabling an AWGN-based denoiser to be applied effectively.\"}]","When AWGN-Based Denoiser Meets Real Noises | PDF",1791275469,[145,149,153,157],{"page_no":8,"url":146,"width":147,"height":148},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/619f340103d2123c1eb972e1cc62d826.webp",983,1272,{"page_no":14,"url":150,"width":151,"height":152},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/8bacf2766ba3fe4390b8cb53c3a5da9c.webp",982,1270,{"page_no":81,"url":154,"width":155,"height":156},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/1d4cda05d3fec43f65d5f7ba4492bbf5.webp",981,1269,{"page_no":19,"url":158,"width":159,"height":160},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/269111d7bbbe308b9c0124a3caa7408d.webp",980,1268]