{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/rethinking-the-csc-model-for-natural-images","title":"Rethinking the CSC Model for Natural Images","arxiv_id":"1909.05742","date":"2019-09-12","proceeding":"NeurIPS 2019 12","authors":["Dror Simon","Michael Elad"],"abstract":"Sparse representation with respect to an overcomplete dictionary is often used when regularizing inverse problems in signal and image processing. In recent years, the Convolutional Sparse Coding (CSC) model, in which the dictionary consists of shift-invariant filters, has gained renewed interest. While this model has been successfully used in some image processing problems, it still falls behind traditional patch-based methods on simple tasks such as denoising. In this work we provide new insights regarding the CSC model and its capability to represent natural images, and suggest a Bayesian connection between this model and its patch-based ancestor. Armed with these observations, we suggest a novel feed-forward network that follows an MMSE approximation process to the CSC model, using strided convolutions. The performance of this supervised architecture is shown to be on par with state of the art methods while using much fewer parameters.","url_abs":"https://arxiv.org/abs/1909.05742v1","url_pdf":"https://arxiv.org/pdf/1909.05742v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"rethinking-the-csc-model-for-natural-images","repo_url":"https://github.com/drorsimon/CSCNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"color-image-denoising","task_name":"Color Image Denoising"},{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/color-image-denoising-on-bsd68-sigma15","task":"Color Image Denoising","dataset":"BSD68 sigma15","model":"CSCNet","rank_in_archive_order":3,"of":4,"metrics":{"PSNR":"33.83"},"uses_additional_data":false},{"leaderboard":"/sota/color-image-denoising-on-bsd68-sigma25","task":"Color Image Denoising","dataset":"BSD68 sigma25","model":"CSCNet","rank_in_archive_order":2,"of":4,"metrics":{"PSNR":"31.18"},"uses_additional_data":false},{"leaderboard":"/sota/color-image-denoising-on-bsd68-sigma75","task":"Color Image Denoising","dataset":"BSD68 sigma75","model":"CSCNet","rank_in_archive_order":1,"of":1,"metrics":{"PSNR":"26.32"},"uses_additional_data":false},{"leaderboard":"/sota/color-image-denoising-on-cbsd68-sigma50","task":"Color Image Denoising","dataset":"CBSD68 sigma50","model":"CSCNet","rank_in_archive_order":11,"of":18,"metrics":{"PSNR":"28.00"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1909.05742","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}