欢迎来到天天文库
浏览记录
ID:36777823
大小:224.94 KB
页数:4页
时间:2019-05-15
《基于TV正则化和局部约束的遥感图像恢复》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、第38卷第6期光子学报Vo1.38No.62009年6月ACTAPH()T()NICASINICAJune2009RemoteSensingImageRestorationBasedonTVRegularizati0nandLocalConstraintsDINGHalyong‘。.BIANZheng—fu(1Co/legDEll'uipotlmeTltalidSpatialhUorDlaticso厂CUMT,Xuzhou,JtangSH221008,(、^ina)(2S(hoolo./RemoteSensing,NanjingLuersityo/htformali(m5(iencPandT
2、eeknologY,Nanjing210044.(、^ina)Abstract:Imagerestorationalgorithmsbasedontotalvariationtheoryandpixelsintensityloca1constrainsmethodwereintroduced.Toobtainabetterrestoredeffect,anovelhybridalgorithmwaspresentedtoestimatetheoriginalimagefromthedegradedone.Totalvariation(TV)regularizationtermwasadded
3、intotheobectfunctiontoformtheiterationformula.And,1ocalmeanandloea1variancewereusedtoconstrainttheso1ution.Innumerica1experiment,thedegradedimageswererestoredusingtheTVmethodandtheproposedmethod.Theresultsshowthatbetterperformanceoftheproposedapproachcanbeobtained.Keywords:ImageRestoration;TotalVar
4、iation;LocalConslraints;EdgePreservationCLCN:TP751DocumentCode:AArticleID:1004l213(2009)061577—4processing【915]0Introduction.ImagepixelsintensityconstraintisanotherRemotesensingimagesareoftencorruptedbyimagerestorationmethodwhichisusedtotherelativemotionbetweentheopticalsystemandregularizetheill—co
5、nditioningoftherestorationtheobjects,andatmosphereturbulence.Imagesproblem.Inreference[16],fixedintensityboundsaredegradedfromblurringandadditivenoise.wereappliedtoblindimagerestoration,whichImagerestorationorimagedeconvolutionreferstoproducesoversmoothedtextureandedgeregions.thetechniquestoobtaint
6、hehighqualityimageInreference[1719],localmeanandvariancefirefromthedegradationimages.Thechallengeinestimatedduringtheiterationprocess,whichimagerestorationproblemisthattheyareill—posedproduceseffectiveestinlationresults.InthisinthesenseofHadamard。I.paper,theTVregularizationmethodandlocalItisdifficu
7、lttoestimatetheoriginalimageconstraintsmethodareusedtoestimatetheblurredfronltheobservationimage-andsomeprioriremotesensingimage.informationaboutimagemustbeintroducedastheThispaperisorganizedasfollows.Insec
此文档下载收益归作者所有