一种在单图像超分辨率中保持边缘信息的技术.pdfVIP

Availableonlineat

ScienceDirect

ProcediaComputerScience85(2016)100–108

InternationalConferenceonComputationalModelingandSecurity(CMS2016)

ATechniquetoPreserveEdgeInformationinSingleImage

SuperResolution

a*b

AmishaJShah,SuryakantBGupta

aAssistantProfessor,C.K.P.C.E.T.,Surat,India.

bFCIPT,InstituteforPlasmaResearch,Gandhinagar,India

Abstract

Goalofimagesuperresolutionistoenhancethesizeofanimagewithoutupsettingtheinheritedinformation.Thequalityofan

enhancedimageisconservediftheinformationaroundallkindofedgesispreserved.Theproposednovelapproachpottedthe

informationaroundcurvatureedgesandhardedges(abrupttransitioninintensity)usingNonSub-SampledContourletTransform

(NSCT)basedlearningprocess.Furthermorethesmoothnessofsmoothedges(gradualtransitioninintensity)ispreservedby

usingsoftedgesmoothnesspriorasaregularizingparameter.Thevalidityoftheproposedapproachisproventhroughsimulation

onseveralimages.

©2016TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense

©2016AmishaJShah,SuryakantBGupta.PublishedbyElsevierB.V.Peer-reviewunderresponsibilityoforganizing

(/licenses/by-nc-nd/4.0/).

committeeofthe2016InternationalConferenceonComputationalModelingandSecurity(CMS2016).

Peer-reviewunderresponsibili