using improved imaging sensor pattern noise for source camera identification

《using improved imaging sensor pattern noise for source camera identification》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、USINGIMPROVEDIMAGINGSENSORPATTERNNOISEFORSOURCECAMERAIDENTIFICATIONYongjianHu1,ChaoJian2,andChang-TsunLi31SchoolofElectronicandInformationEngineering,2CollegeofAutomationScienceandEngineering,SouthChinaUniversityofTechnology,Guangzhou510641,P.R.China3DepartmentofCom

2、puterScience,UniversityofWarwick,CoventryCV47AL,UKE-mail:eeyjhu@scut.edu.cn;chao.axl.jian@gmail.com;c-t.li@warwick.ac.ukABSTRACTtakenbythesamecamera.ButnshouldbeverylargeinthisTheimagingsensor(e.g.,CCD)patternnoiseisanoise-likecase.Theyproposedtospeeduptheextraction

3、processthroughspread-spectrumsignalinherentlycastontoeverydigitalimagesuppressingimagecontentbeforeperformingtheaveragingop-byeachimagingdevice.Ithasbeenrecognizedasareliableeration.Thewaytheyremovedimagecontentistosubtractthedevicefingerprintforsourcecamera/scanneri

4、dentificationanddenoisedversionfromtheoriginalimage.Theeffectisapparentimageoriginverification.However,oneweaknessofcurrentbecausemostsceneisinvisibleandonlystrongedgesandtex-sensorpatternnoise-basedcamerafingerprintsisthattheircon-turesleavesomeweakmarksonresidualimag

5、es.Ontheotherstructiononlyusesinformationfromsinglecolorchannelim-hand,thesuperimpositionoftheseresidualimagescausesin-ages.Theresultantfingerprintcannotcomprehensivelyreflectdividualindependentadditivenoisecomponents(e.g.,read-outcharacteristicsofthecamerasensorarray

6、.Takingintoaccountnoise)atthesamespatiallocationcounteractedwitheachother,thecharacteristicsoftheCFA(colorfilterarray)structure,thisandthusgreatlyreducestheeffectofirrelevantnoisecompo-workproposesanewwaytoconstructcamerafingerprintsusingnents.Onedisadvantageofthissch

7、emeisthatthedenoisinginformationfromallofthethreesinglecolorchannelimages.filterwouldbringshapingnoiseintoresidualimages.ShapingGooddetectionresultshavebeenachievedondifferentcameranoiseisamultiplicativenoise,andishardtoberemovedbyav-models.eraging.Tosolvethisproblem

8、,thesameauthors[3]adoptedamaximum-likelihoodestimatortoextractamoreaccurateformKeywords—Digitalimageforensics,cameraidentification,ofthePRN