上海交通大学硕士论文-基于视频印记的同源视频检测

上海交通大学硕士学位论文基于视频印记的同源视频检测姓名：徐波申请学位级别：硕士专业：通信与信息系统指导教师：孙军20080101Content-basedCopyDetectionVideoRetrievalDETECTIONOFCODERIVATIVEVIDEOUSINGVIDEOSIGNATURESABSTRACTWiththerapiddevelopmentandwide-spreadinguseofnetworkingandmultimediatechniques,theproduction,duplication,transmissionandpublicationofdigitalvideosarebecomingeasierandeasier,causingcopyrightprotectionofdigitalvideoasevererprobleminthepastfewyears.Althoughsometechniquesofcontentprotectionformultimediahavebeeninvestigated,likephysicalcopydetectionanddigitalwaterprint,thecopyrightidentificationisstillanissueformostcontentsuppliers.Thevideoscomingfromthesamesourcearecalledcoderivativevideo,andthisthesisisfocusingonthedetectionandidentificationofcoderivativevideo.ThedetectionofcoderivativevideoisanewapplicationpatternofCBCDforvideos(Content-BasedCopyDetection),itincludestechniqueslikeacquirementofcharacteristicinformationrelatedtovideocontentemployingofpropercharacteristic-representingmethodsdesignofappropriatealignmentstrategyandthecoderivativityjudgingbasedonthealignmentresults.Currently,manytechniquesusedinVideoRetrievalarebeusedincoderivativevideodetection,withnot-so-promisingeffect,causedbymanynoisesintroducedbythetransformingandeditingofvideos.Inaccordancewiththesensitivitytodegradationofvideosbycurrentcharacteristicsextractionmethodsandalignmentstrategy,thisthesisproposesseveralsteadycharacteristicsaquiringmethodswhicharesuitabletocoderevitivitydetection,usingtheformofvideosignatures,anddesignsanaccordinglyalignmentalgorithmforthesevideosignatures.Itshowspromisingeffecttomostcommonvariationofvideoparameters.Themajorworkandachievementsofthisthesisincludethefollowingaspects:Afterreviewingmostbasictheoriesrelatedtocoderivativevideodetection,includingthesimilaritymeasureofbothframeandvideosequence,themostdifficultandimportantpointsofthisprocedureareproposedwiththestudyofachievementsinrelativefieldsbothhomeandabroad.Aimingatcoderevitivitydetection,adetectionframeworkusingvideosignaturesiselaboratedbasedondeeplyanalysisofcurrentvideosimilaritymodel,whichisprovedbyexperimentsinthefollowingchapters.Astothesignaturechoosingandproduction,fourvideosignaturesareproposedwiththeregardofsensitivityofknownvideosignaturestodegradationofvideoqualityincoderevitivitydetection,whichallshowtheireffectivenessandrobustnessbymassiveexperimentsbothbetweencoderivativevideosandunrelatedvideos.Concerningthealignmentofvideosignaturesproposedinthisthesis,famoussequencealignmentalgorithmsinbiologyresearchareintroduced,aspecificsequencealignmentalgorithmisdesigned,alongwithseveralscoringfunctionsbasedontherealmeaningofthevideosignaturesthisthesisuses.Agreatamountofexperimentdatashowthatacombinationofvideosignaturesandthesequencealignmentalgorithmproposedbythisthesiswillleadtoamorepromisingcoderivativedetectionresult,comparedtocurrentcoderivativevideodetectionmethods.KEYWORDSContentProtectionVideoSignatureCoderivativeVideoSequenceAlignmentCBVRContent-BasedVideoRetrievalCCVColorCoherenceVector1.................................................................................................92SVS.......................................................................................203IVS........................................................................................224VoronoiVVS.............................................................................245Voronoi..............................................................................256...............................................................................317...........................................................................328...................................................................................379Seq1Seq2...........................................................3910Seq1Seq1’.............................3911Seq1Seq2.....................4012Seq3Seq4.............................................................4313Seq3Seq3’.................................4414Seq3Seq4.................................4415.....................................................................................................4516.............................................................................................4717.............................................................................................4918.............................................................................................6219.....................................................................................7520.............................................................7721.............................................................7822.....................................................................................851...........................................................................422........................................................................483...........................................................................524...............................525.........................................