Single Image Reflection Removal With Physically-Based Training Images

Cited 49 time in webofscience Cited 21 time in scopus
  • Hit : 145
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorKim, Soominko
dc.contributor.authorYuchi Houko
dc.contributor.authorYoon, Sung-Euiko
dc.date.accessioned2020-10-22T23:55:32Z-
dc.date.available2020-10-22T23:55:32Z-
dc.date.created2020-10-14-
dc.date.created2020-10-14-
dc.date.issued2020-06-17-
dc.identifier.citationConference on Computer Vision and Pattern Recognition (CVPR), pp.5164 - 5173-
dc.identifier.issn1063-6919-
dc.identifier.urihttp://hdl.handle.net/10203/276911-
dc.description.abstractRecently, deep learning-based single image reflection separation methods have been exploited widely. To benefit the learning approach, a large number of training image-pairs (i.e., with and without reflections) were synthesized in various ways, yet they are away from a physically-based direction. In this paper; physically based rendering is used for faithfully synthesizing the required training images, and a corresponding network structure and loss term are proposed. We utilize existing RGBD/RGB images to estimate meshes, then physically simulate the light transportation between meshes, glass, and lens with path tracing to synthesize training data, which successfully reproduce the spatially variant anisotropic visual effect of glass reflection. For guiding the separation better, we additionally consider a module, backtrack network (BT-net) for backtracking the reflections, which removes complicated ghosting, attenuation, blurred and defocused effect of glass/lens. This enables obtaining a priori information before having the distortion. The proposed method considering additional a priori information with physically simulated training data is validated with various real reflection images and shows visually pleasant and numerical advantages compared with state-of-the-art techniques.-
dc.languageEnglish-
dc.publisherIEEE Computer Society and the Computer Vision Foundation (CVF)-
dc.titleSingle Image Reflection Removal With Physically-Based Training Images-
dc.typeConference-
dc.identifier.wosid000620679505044-
dc.identifier.scopusid2-s2.0-85094829097-
dc.type.rimsCONF-
dc.citation.beginningpage5164-
dc.citation.endingpage5173-
dc.citation.publicationnameConference on Computer Vision and Pattern Recognition (CVPR)-
dc.identifier.conferencecountryUS-
dc.identifier.conferencelocationVirtual-
dc.identifier.doi10.1109/CVPR42600.2020.00521-
dc.contributor.localauthorYoon, Sung-Eui-
dc.contributor.nonIdAuthorYuchi Hou-
Appears in Collection
CS-Conference Papers(학술회의논문)
Files in This Item
There are no files associated with this item.
This item is cited by other documents in WoS
⊙ Detail Information in WoSⓡ Click to see webofscience_button
⊙ Cited 49 items in WoS Click to see citing articles in records_button

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0