{"corpus_id":258187270,"paper_sha":"8b8777a918b1eeff120058b32029b8e2344b289a","doi":"10.48550/arXiv.2304.08990","arxiv_id":"2304.08990","pmid":null,"pmcid":null,"mag_id":null,"dblp_id":"journals/corr/abs-2304-08990","acl_id":null,"title":"A Comparison of Image Denoising Methods","year":2023,"publication_date":"2023-04-18","venue":"arXiv.org","journal":{"name":"ArXiv","pages":null,"volume":"abs/2304.08990"},"journal_issn":null,"journal_title":null,"publication_types":["JournalArticle"],"pubmed_pub_types":null,"s2_fields_of_study":["Computer Science","Engineering"],"reference_count":249,"citation_count":13,"influential_citation_count":0,"is_open_access":true,"arxiv_categories":["eess.IV","cs.CV"],"arxiv_license":"http://creativecommons.org/licenses/by/4.0/","arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":"http://arxiv.org/pdf/2304.08990","s2_open_access_landing_url":"https://www.semanticscholar.org/paper/8b8777a918b1eeff120058b32029b8e2344b289a","s2_open_access_license":null,"s2_open_access_status":"GREEN","pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"The advancement of imaging devices and countless images generated everyday pose an increasingly high demand on image denoising, which still remains a challenging task in terms of both effectiveness and efficiency. To improve denoising quality, numerous denoising techniques and approaches have been proposed in the past decades, including different transforms, regularization terms, algebraic representations and especially advanced deep neural network (DNN) architectures. Despite their sophistication, many methods may fail to achieve desirable results for simultaneous noise removal and fine detail preservation. In this paper, to investigate the applicability of existing denoising techniques, we compare a variety of denoising methods on both synthetic and real-world datasets for different applications. We also introduce a new dataset for benchmarking, and the evaluations are performed from four different perspectives including quantitative metrics, visual effects, human ratings and computational cost. Our experiments demonstrate: (i) the effectiveness and efficiency of representative traditional denoisers for various denoising tasks, (ii) a simple matrix-based algorithm may be able to produce similar results compared with its tensor counterparts, and (iii) the notable achievements of DNN models, which exhibit impressive generalization ability and show state-of-the-art performance on various datasets. In spite of the progress in recent years, we discuss shortcomings and possible extensions of existing techniques. Datasets, code and results are made publicly available and will be continuously updated at https://github.com/ZhaomingKong/Denoising-Comparison.","claims":[{"public_id":"cl_5b5c8f3ead970bca4bd72285f42f8648","status":"active","text":"A new dataset for benchmarking denoising methods is introduced, together with publicly available datasets, code, and results.","confidence":0.97,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_5b5c8f3ead970bca4bd72285f42f8648"},{"public_id":"cl_43bcde5d303f83719c48a721963aa65c","status":"active","text":"A simple matrix-based algorithm can produce results similar to its tensor counterparts.","confidence":0.9,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_43bcde5d303f83719c48a721963aa65c"},{"public_id":"cl_e83137005275dae48ee97e1ec03d13c0","status":"active","text":"Deep neural network models achieve impressive generalization ability and state-of-the-art performance on various datasets.","confidence":0.95,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_e83137005275dae48ee97e1ec03d13c0"},{"public_id":"cl_62d97b09b3cfbe4d2e269f02f0c6d82d","status":"active","text":"Representative traditional denoisers are effective and efficient across various denoising tasks.","confidence":0.92,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_62d97b09b3cfbe4d2e269f02f0c6d82d"}],"concepts":[{"public_id":"co_1593900a97d44f17e5efe96fa92a94c9","status":"active","name":"computational cost","description":"The amount of computation required by a denoising method.","types":["evaluation metric"],"aliases":["runtime cost","efficiency cost"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_1593900a97d44f17e5efe96fa92a94c9"},{"public_id":"co_1653f05fd81ff46fe20d293bc05c3481","status":"active","name":"matrix-based algorithm","description":"A denoising algorithm formulated using matrix representations.","types":["method"],"aliases":["simple matrix-based algorithm"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_1653f05fd81ff46fe20d293bc05c3481"},{"public_id":"co_1a636254184823173e533e67bfa76bfc","status":"active","name":"denoising tasks","description":"Image restoration problems focused on removing noise from images.","types":["task"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_1a636254184823173e533e67bfa76bfc"},{"public_id":"co_2ed6777560cdee287676cf267c6f788f","status":"active","name":"traditional denoisers","description":"Conventional image denoising techniques that are not based on deep neural networks.","types":["method"],"aliases":["traditional denoising techniques"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_2ed6777560cdee287676cf267c6f788f"},{"public_id":"co_735391a0194fdc3cee0827168d2c5948","status":"active","name":"deep neural network models","description":"Denoising models built with deep neural network architectures.","types":["method"],"aliases":["DNN models","DNN architectures"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_735391a0194fdc3cee0827168d2c5948"},{"public_id":"co_77ad25b602d04c5e2b734920e534aaa5","status":"active","name":"human ratings","description":"Subjective assessments provided by people to judge denoising results.","types":["evaluation criterion"],"aliases":["subjective ratings"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_77ad25b602d04c5e2b734920e534aaa5"},{"public_id":"co_8afb2dde069b8b7556c424956c560e93","status":"active","name":"tensor counterparts","description":"Denoising methods that use tensor representations as the comparison target for the matrix-based algorithm.","types":["method"],"aliases":["tensor-based methods"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_8afb2dde069b8b7556c424956c560e93"},{"public_id":"co_9623afaadc432b60d9f0513cad01a4ae","status":"active","name":"visual effects","description":"Visual appearance of denoised images used as an evaluation perspective.","types":["evaluation criterion"],"aliases":["visual quality"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_9623afaadc432b60d9f0513cad01a4ae"},{"public_id":"co_99c898f8c0d402241f3da9eab51fd30a","status":"active","name":"image denoising methods","description":"Algorithms and models designed to remove noise from images while preserving useful content.","types":["method"],"aliases":["denoising methods"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_99c898f8c0d402241f3da9eab51fd30a"},{"public_id":"co_9db0691cc53ac5ec5b77319176b73901","status":"active","name":"datasets, code and results","description":"The study artifacts released publicly and stated to be continuously updated.","types":["resource"],"aliases":["publicly available resources"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_9db0691cc53ac5ec5b77319176b73901"},{"public_id":"co_b0627ca4482712920b7151b181a93fc2","status":"active","name":"benchmarking dataset","description":"A newly introduced dataset intended for standardized comparison of denoising methods.","types":["dataset"],"aliases":["new dataset for benchmarking"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_b0627ca4482712920b7151b181a93fc2"},{"public_id":"co_c0141f8899ff52ab70e4654045ba4cf7","status":"active","name":"quantitative metrics","description":"Numerical measures used to evaluate denoising performance.","types":["evaluation metric"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_c0141f8899ff52ab70e4654045ba4cf7"},{"public_id":"co_c11bf1541aeb2279aac598ee04f4e42d","status":"active","name":"various datasets","description":"Multiple evaluation datasets spanning different denoising settings.","types":["dataset"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_c11bf1541aeb2279aac598ee04f4e42d"},{"public_id":"co_c78f34602aae4ce03285287bd1c02298","status":"active","name":"synthetic datasets","description":"Artificially generated datasets used to evaluate denoising performance.","types":["dataset"],"aliases":["synthetic data"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_c78f34602aae4ce03285287bd1c02298"},{"public_id":"co_da5085f21c5446dd053c2c4f01cc2626","status":"active","name":"real-world datasets","description":"Datasets containing real images or real acquisition noise used for evaluation.","types":["dataset"],"aliases":["real data"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_da5085f21c5446dd053c2c4f01cc2626"}],"external_ids":{"DOI":"10.48550/arXiv.2304.08990","ArXiv":"2304.08990","PubMed":null,"PubMedCentral":null,"MAG":null,"DBLP":"journals/corr/abs-2304-08990","ACL":null},"open_access":{"is_open_access":true,"pdf_url":"http://arxiv.org/pdf/2304.08990","landing_url":"https://www.semanticscholar.org/paper/8b8777a918b1eeff120058b32029b8e2344b289a","source":"semantic_scholar","pdf_url_source":"semantic_scholar_open_access_pdf","license":null,"status":"GREEN","reason":null},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":true,"full_text_source":"arxiv","count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":637406,"paper_uid":"74d03a0f-4c06-401b-a089-56c579f13d35","canonical_identity":{"paper_id":637406,"paper_uid":"74d03a0f-4c06-401b-a089-56c579f13d35","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/258187270"}