欢迎来到天天文库
浏览记录
ID:20606804
大小:11.12 MB
页数:59页
时间:2018-10-14
《基于加权红-黑小波变换的dr图像增强方法研究》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、万方数据硕士学位论文basedonhybridoptimizationalgorithmandSOon.Thesemethodscanenhancethecontrastoftheimageandimagedetailstosomeextent,buttherearesomeshortcomings,suchastheintroductionofhaloartifactsattheedges,amplifyingnoise,fuzzydetailsandslowspeedandothershortcomings.Anenhancemen
2、talgorithmbasedonweightedred·blackwavelets(WRB)transformispresentedfortheoriginalDRimageswitlltheproblemsofdetailsburiedandwidedynamicrange.Withthepropertyofmulti-resolutionanalysis,classicalwavelettransformCandecomposeanimagetothevarioussub-bands,duetoitssub—bandtransfo
3、rmcoefficientshavegoodmaneuverability,andclassicalhasalong-termdevelopmentinthefieldofimageenhancement.However,sameastheFouriertransform,classicalwavelettransformbelongtothefrequencydomainanalysis,thoughithastherapidcalculationmethod,itstillhastheweaknessoflargeamountofc
4、alculation,largememo巧space,floatingpointcalculationsandSOon.Secondgenerationwavelettransform(alsocalledliRingwavelettransform),comparedtothefirstgenerationwavelettransform,nolongerdependentontheFouriertransform,butinheritsthefirstgenerationwavelet’Scharacteristicsofmulti
5、-resolutionanalysis,itusein-situoperations,hasfastcomputingspeedandwithouttakingupadditionalstoragespace,andthereisnorestrictiononthesizeoftheimage.Onthisbasis,GeertUytterhoevenproposedanon-separablewavelet-red-blackWavelet(red·blackwavelet,RBW)basedontwo·dimensionallift
6、ingscheme.Thewavelethastwofullstepsofthehorizontal/verticalanddiagonallifting;eachliftingwavelettransformiscomposedbythreestepsofsplit,predictandupdate.Inthisthesis,weproposedanewweightedred-blackwavelettransformbasedonclassicalred-blackwavelet,andapplyittotheDRimageenha
7、ncement.Classicalred.blackWaveletTransformselectthelinearpredictionoperatorandupdateoperator,butmostoftheimageshavenon-contiguousareas(suchasedges),the万方数据ABSTRACTpixelvalues’differenceisbigneartheedge,usinglinearinterpolationformulacannotgetagoodpredictionandupdate,wesh
8、ouldchoosethepredictionoperatorandupdateoperatorbasedonthelocalfeaturesofimages,SOthatthepredictionoper
此文档下载收益归作者所有