{"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/deep-speckle-correlation-a-deep-learning","title":"Deep speckle correlation: a deep learning approach towards scalable imaging through scattering media","arxiv_id":"1806.04139","date":"2018-06-11","proceeding":null,"authors":["Yunzhe Li","Yujia Xue","Lei Tian"],"abstract":"Imaging through scattering is an important, yet challenging problem.\nTremendous progress has been made by exploiting the deterministic input-output\nrelation for a static medium. However, this approach is highly susceptible to\nspeckle decorrelations - small perturbations to the scattering medium lead to\nmodel errors and severe degradation of the imaging performance. In addition,\nthis is complicated by the large number of phase-sensitive measurements\nrequired for characterizing the input-output `transmission matrix'. Our goal\nhere is to develop a new framework that is highly scalable to both medium\nperturbations and measurement requirement. To do so, we abandon the traditional\ndeterministic approach, instead propose a statistical framework that permits\nhigher representation power to encapsulate a wide range of statistical\nvariations needed for model generalization. Specifically, we develop a\nconvolutional neural network (CNN) that takes intensity-only speckle patterns\nas input and predicts unscattered object as output. Importantly, instead of\ncharacterizing a single input-output relation of a fixed medium, we train our\nCNN to learn statistical information contained in several scattering media of\nthe same class. We then show that the CNN is able to generalize over a\ncompletely different set of scattering media from the same class, demonstrating\nits superior adaptability to medium perturbations. In our proof of concept\nexperiment, we first train our CNN using speckle patterns captured on diffusers\nhaving the same macroscopic parameter (e.g. grits); the trained CNN is then\nable to make high-quality reconstruction from speckle patterns that were\ncaptured from an entirely different set of diffusers of the same grits. Our\nwork paves the way to a highly scalable deep learning approach for imaging\nthrough scattering media.","url_abs":"http://arxiv.org/abs/1806.04139v1","url_pdf":"http://arxiv.org/pdf/1806.04139v1.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":"deep-speckle-correlation-a-deep-learning","repo_url":"https://github.com/bu-cisl/Deep-Speckle-Correlation","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.04139","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}