File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: Computational image speckle suppression using block matching and machine learning

TitleComputational image speckle suppression using block matching and machine learning
Authors
KeywordsMotion compensation
Neural networks
Shrinkage
Speckle
Issue Date2019
PublisherOptical Society of America. The Journal's web site is located at http://ao.osa.org/journal/ao/about.cfm
Citation
Applied Optics, 2019, v. 58 n. 7, p. B39-B45 How to Cite?
AbstractWe develop an image despeckling method that combines nonlocal self-similarity filters with machine learning, which makes use of convolutional neural network (CNN) denoisers. It consists of three major steps: block matching, CNN despeckling, and group shrinkage. Through the use of block matching, we can take advantage of the similarity across image patches as a regularizer to augment the performance of data-driven denoising using a pre-trained network. The outputs from the CNN denoiser and the group coordinates from block matching are further used to form 3D groups of similar patches, which are then filtered through a wavelet-domain shrinkage. The experimental results show that the proposed method achieves noticeable improvement compared with state-of-the-art speckle suppression techniques in both visual inspection and objective assessments.
Persistent Identifierhttp://hdl.handle.net/10722/275025
ISSN
2010 Impact Factor: 1.707

 

DC FieldValueLanguage
dc.contributor.authorZENG, T-
dc.contributor.authorSo, HKH-
dc.contributor.authorLam, EY-
dc.date.accessioned2019-09-10T02:33:55Z-
dc.date.available2019-09-10T02:33:55Z-
dc.date.issued2019-
dc.identifier.citationApplied Optics, 2019, v. 58 n. 7, p. B39-B45-
dc.identifier.issn0003-6935-
dc.identifier.urihttp://hdl.handle.net/10722/275025-
dc.description.abstractWe develop an image despeckling method that combines nonlocal self-similarity filters with machine learning, which makes use of convolutional neural network (CNN) denoisers. It consists of three major steps: block matching, CNN despeckling, and group shrinkage. Through the use of block matching, we can take advantage of the similarity across image patches as a regularizer to augment the performance of data-driven denoising using a pre-trained network. The outputs from the CNN denoiser and the group coordinates from block matching are further used to form 3D groups of similar patches, which are then filtered through a wavelet-domain shrinkage. The experimental results show that the proposed method achieves noticeable improvement compared with state-of-the-art speckle suppression techniques in both visual inspection and objective assessments.-
dc.languageeng-
dc.publisherOptical Society of America. The Journal's web site is located at http://ao.osa.org/journal/ao/about.cfm-
dc.relation.ispartofApplied Optics-
dc.rightsApplied Optics. Copyright © Optical Society of America.-
dc.rights© XXXX [year] Optical Society of America]. One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modifications of the content of this paper are prohibited.-
dc.subjectMotion compensation-
dc.subjectNeural networks-
dc.subjectShrinkage-
dc.subjectSpeckle-
dc.titleComputational image speckle suppression using block matching and machine learning-
dc.typeArticle-
dc.identifier.emailSo, HKH: hso@eee.hku.hk-
dc.identifier.emailLam, EY: elam@eee.hku.hk-
dc.identifier.authoritySo, HKH=rp00169-
dc.identifier.authorityLam, EY=rp00131-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1364/AO.58.000B39-
dc.identifier.pmid30874216-
dc.identifier.scopuseid_2-s2.0-85062210534-
dc.identifier.hkuros304140-
dc.identifier.volume58-
dc.identifier.issue7-
dc.identifier.spageB39-
dc.identifier.epageB45-
dc.publisher.placeUnited States-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats