欢迎来到天天文库
浏览记录
ID:36444069
大小:4.55 MB
页数:72页
时间:2019-05-10
《彩色图像水印算法研究》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、辽宁师范大学硕士学位论文彩色图像水印算法研究姓名:贝依林申请学位级别:硕士专业:计算机软件与理论指导教师:闫德勤20080501ABSTRACTInrecentyears,thespreadoftheInternetandtheadvanceindigitaltechniquesfortransmittingandstoringinformationhavemad.eitpossibletocopyandeditimages,audio,videoandothertypesofmultimediad
2、ata.Multimediadatabydigitalformatcanbeduplicatedwithoutlossofqualityandchangedeasilyatlowcost.Thesesituationscausetheillicitactionssuchasinterception,duplication,misuseandunauthorizedmodificationofinformation.Oneapproachtoaddresstheproblemiswatermarkin
3、gtechnique.Theconceptofdigitalwatermarking·isassociatedwiththedata—hidingtechnique.DigitalwatermarkinghasbeenproposedasawaytoidentifytheSOUrCe,creator,owner,distributor,orauthorizedconsumerofmultimediadata.Inthisthesis,thecurrentwatermarkdevelopingstat
4、us,basiccharactersandproblemsofdigitalwatermarktechnologywereanalyzed,andthegeneralframeofdigitalwatermarkprocessingsystemwasdiscussedandthreewatermarkingschemesareproposed.Inthefirstscheme,presentsacolorimagewatermarkingmethodwithblinddetectionbasedon
5、amplitudemodulation.Onthebasisofthevisualmaskingofhumanvisualsystem,theembeddedpositionswerechosenadaptively.Basically,thewatermarkwasembeddedbymodifyingthepixelvaluesinthebluechannelofanimage,whilethewatermarkretrievalisachieved‘byusingapredictiontech
6、niquebasedonlinearcombinationofnearbypixelvaluesaroundtheembeddedpixels.Weuseatechniqueofgaussianweightingmarksbyaveragingtheluminanceofitsownpixelanditsneighboringpixels,anduseitasanadditionalscalingfactortotunethestrengthofembeddingwatermark.Experime
7、ntalresultsshowthatthepresentedschemeisinvisibleandrobustagainstvarioussignalprocessingsuchascropping,compression,filtering,noiseadding,imageenhancementandgeometricalattacks.Inthesecondscheme,anewalgorithmisproposed;thewatermarklIwasembeddedbymodifying
8、thepixelvaluesinthebluechannelofanimage,whilethewatermarkretrievalisachievedbyusingapredictiontechniquebasedonsupportvectorregressionmodel.Themodelistrainedbylearningtherelationshipbetweentheselectedpixelandstatisticalfeatureofneighbori
此文档下载收益归作者所有