If you can't read please download the document
Upload
truonglien
View
223
Download
0
Embed Size (px)
Citation preview
Forensic Methods for Detecting Image Manipulation -Copy Move
Anthony SutardjaOmar RamadanYan Zhao
Electrical Engineering and Computer SciencesUniversity of California at Berkeley
Technical Report No. UCB/EECS-2015-84http://www.eecs.berkeley.edu/Pubs/TechRpts/2015/EECS-2015-84.html
May 13, 2015
Copyright 2015, by the author(s).All rights reserved.
Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and thatcopies bear this notice and the full citation on the first page. To copyotherwise, to republish, to post on servers or to redistribute to lists,requires prior specific permission.
Table of Contents For the UC Berkeley Master of Engineering Capstone Project 1. Introduction 2 2. Trends, Market, & Strategy 3 3. Industry 6 4. Intellectual Property Law 12 5. Works Cited 17 6. Appendix 21 7. Individual Technical Contribution - Anthony 25 8. Individual Concluding Reflection - Anthony 42
Forensic Methods for Image Manipulation Detection 1
Introduction Problem Statement Forthepastcentury,photographshaveservedasreliableprimarysourcesofevidence,butthatis
quicklychanging.Photomanipulationtoolshavebecomewidespreadanditiseasytomanipulate
images.PhotomanipulationstoolssuchasAdobePhotoShopaffordgreaterartisticexpression,and
enableuserstocreatemanipulationsthatchallengethelimitsofournaturalperception.Thedifference
betweenauthenticandmanipulatedphotoshasbecomehardertodistinguish,andcanonlybedetected
bydigitalforensicexperts.Theimageforensicscapstoneprojectaimstocreateanonlinesoftware
servicethatperformstheworkofforensicanalysts,andvisualizesandanalyzesthepossible
manipulationsthatmayhavebeenperformedonanimage.Ourgoalistoempoweranybodytoperform
imageforensicanalysis,andhelpincreaseourfaithindigitalcontent.
Ourcapstonehascompletedaprototypeofthisimagemanipulationdetectionwebservicethathelpsa
useridentifyfraudulentfeatureswithinanimage.Althoughfullyfunctional,ourserviceisnotyetready
forcommercializationasmanyofthetechniquesarestillunderresearch.Inthefollowingsections,we
performamarket,industry,andintellectualpropertyanalysisonhowwewouldhypotheticallytakeour
onlinesoftwareservicetomarket.
Forensic Methods for Image Manipulation Detection 2
Trends, Market, and Strategy Trends Therearestrongtechnologicalandregulatorytrendsthatrevealagrowingneedforeffectiveimage
forensicssolutions.Smartphonesandthecamerasembeddedwithinthemhaverapidlybecomethemost
popularmethodofphotocapture.AccordingtothePewResearchCenter,thepercentageofU.S.adults
whoownsmartphonesjumpedfrom35%in2011to56%in2013(Smith,SmartphoneOwnership
2013).Thisrelentlessgrowthinsmartphoneadoptionwillcontinue,especiallyinregionsoutsideof
theUnitedStates(Dulaney5).Morepeoplehavetheabilitytotakedigitalphotos,asseventysix
percentofsmartphoneuserssaytheyusethesmartphonescamerastotakephotos(Smith,Mobile
Access2010).Thisinstantandconvenientphotosnappingabilityisbecomingrecognizedbymany
institutionsasawayofdocumentingincidentsandfilingclaims(HowtoDocumentAutoAccident
Damage.)(Thomson3).However,asweincreaseourrelianceondigitalphotography,webecome
susceptibletofraud.Digitalimageeditingsoftwareiseasytoobtainandisgrowinginutilizationas
well(Kahn9).Withthegrowingtrendinsmartphoneadoptionandtheriseinaccesstoimageediting
software,digitallymanipulatedimageswillbecomecommonplace.
Duetothegrowingtrendofdigitalimagedocumentation,theregulatorylandscapeisalsochanging
withnewcourtprecedentsthatarebeginningtorequireverifyingtheauthenticityofdigitalimages.In
Israel,anewlawmakesitillegaltouseimagesinadvertisementsthathavebeenretouchedtomake
modelslookthinnerwithoutprintingadisclosureonthepicture(PictureImperfect).IntheUnited
States,theFederalRulesofEvidencenowrequiretheauthenticationofdigitalimagesbeforetheyare
usedasevidenceincourt(Thomson3).Thegrowingtrendinregulatorymeasuresforverifyingdigital
imageswillonlycontinuetopushtheneedforimageforensicsservice.
Hence,identifyingthesetrendshashelpedusrealizethecustomersoftheimageforensicsindustry.In
journalism,weseetheneedtoverifydigitalimagesforjournalisticintegrity.Inthecourtroom,wesee
theneedtoverifydigitalimagestoestablishevidentiaryauthenticity.Ininsurance,weseetheneedto
identifyfraudulentdigitalinsuranceclaims.Asmoreservicesbegintorelyondigitaldocumentation,
theneedforsuchmediaverificationserviceswillincrease.
Forensic Methods for Image Manipulation Detection 3
Market Potential Markets Thereareseveralmarketsinwhichanimageforensicssolutionisgreatlyneeded.
Ourtechnologyhasseriousimplicationsforthepressimagemarket.Theexplodinggrowthofthe
Internetanddigitalcontenthasempoweredcitizenjournalismandtransformedthepressindustrywith
exclusiverichmediacollections.Withthisgrowth,presscompaniesnowhavetheadditionalchallenge
ofmaintainingtheircredibilitybyvalidatingmediafromnewunverifiedsources.Photoeditorsatsuch
institutionsworktoensurethatonlygenuinephotosarepublished,butwithadvancededitingtools,
evenanexperiencedanalystcanfailtoidentifyimagemanipulations.The$32billiondollarnewspaper
industrytakesphotomanipulationasaseriousthreatbecauseofthedamageitbringstotheircredibility
(McKenna).Theirzerotolerancepolicytowardsphotomanipulationfrequentlyresultsinhundredsof
thousandsofdollarsinlosseseachmanipulatedphotoisissuedamandatorykillthatalertsall
affectedcustomerstocontroldamages(Lum).Theinvolvedphotographersareusuallyterminatedsuch
asinthecaseofGettyfreelancerMarcFeldman(Lum),L.ATimesstaffphotographerColinCrawford
(Irby),orReutersfreelancerAdnanHajj(AlteredImagesPromptPhotographersFiring).
Anotherpotentialmarketresidesinthecourtoflaw.Admittingmanipulatedphotographicevidencein
courtisbothagrowingtrendandagrowingconcern.Perjuryisnotuncommon,andpoliceofficersare
estimatedtocommitperjurytwentytofiftypercentofthetimeonfourthamendmentissues(Slobogin).
Lawyersalsocommitperjury,andevengoodlawyersandareuntruthfulintheirprofession.Asit
stands,theauthenticityofphotographsiscontingentonthehonestyofthewitnesses.Animage
forensicsexpertisonlycalleduponwhentherearenowitnessestotestifyinalawsuitorwhenthe
photographicevidencehasitsintegritychallenged(Kashi).Otherwise,thereisnostandardprocedureto
havethephotoevidencethoroughlyvetted.The$400billiondollarlegalindustryneedsabetterwayto
admitphotographicevidenceinacourtoflawthatismorerobustthanasingletestamentofthetruth
(Novet).Weseeourimageforensicsoftwareasanopportunitytochangethewayphotoevidenceis
processed.
Forensic Methods for Image Manipulation Detection 4
Segmented Market Entry Thelastmarketthatwehaveidentifiediswithinsurancecompanies.Ourcapstoneteamhasdecidedto
focusoncateringourimageforensicssoftwareservicetowardstheinsurancesegmentduetothegreat
valuewewillprovidetoinsurancecompanies.Thisvaluestemsfromthemajorlossesinsurance
companiesfaceduetofraudulentclaims.Whiledigitallydocumentedphotographyanddocumentation
hasmadeitmorecosteffectivefortheinsuranceindustrytosurveydamages,ithasexposedthemto
significantlosses.Thisshifttowardsdigitalevidencehasmadeiteasiertotamperwithphotographsto
exaggerateorfabricateinsuranceclaims.Onestudyfoundthatoneinsevenhundredgeneralinsurance
claimshasbeendigitallyaltered,andoneinseventyfivepropertydamageinsuranceclaimshasbeen
digitallyaltered(PictureImperfect).Thesedigitallyalteredinsuranceclaimsaccountforasignificant
portionofthe$40billiondollarslostbytheinsuranceindustryeveryyearintheUnitedStates(FBI).
Thesemonumentallossesindicatethatinsurancecompanieshavestrongneedsforinnovativewaysto
combatfraudulentclaims.Ourimageforensicssoftwareseekstoaddressthisproblembyhelping
insurancecompaniesidentifyfraudulentclaimsaccurately,andinacosteffectivemanner.
Marketing Strategy Product Thedetectionofmanipulatedimagesisanimportantproblemthatrequirestheconsultationofhighly
specializedforensicexperts.Giventhegrowingnatureofthisproblem,weproposeaservicethatputs
theanalyticalpowerofimageforensicmethodsinthehandsofthecommonuser.Wegeneratedetailed
reportsthatcanbeinterpretedbynonexperts,effectivelyallowinglowerlevelanalyststocreatemore
valuewhilerelievingtheworkloadsofbusyforensicexperts.
Ourproductincreasesthecapacityofanalystsitprovidesthemwithforensicinformationthathelps
themmakeinformeddecisionswhenevaluatingclaims.Whileimagessubmittedwithaclaimareonlya
portionoftheevidencethatanalystsusetobasetheirdecisions,theinformationweprovidehelpsthem
identifyandflagsuspiciousclaimsforfurtherreview,ultimatelyenablingthemtopreventmorefraud.
Forensic Methods for Image Manipulation Detection 5
Value Guarantee Thelargestriskforourcustomersisanegativereturnoninvestment,inthatthecostofourplatformis
notexceededbythevalueoffraudthatwestop.Toreducethisrisk,andpromotethesaleofour
service,ourstrategyistominimizethesoftwareacquisitionrisksforourcustomers.Wecanreducethe
switchingcostsbysupplyingourownforwarddeployedengineerstointegratewithourcustomers
softwarestacksandourownsupportteamtoprovidetrainingservices.Furthermore,withefficiencyour
serviceoffersanditseffectivenessincombatingfraud,wecanofferaguaranteedreturnoninvestment. Customer Outreach Customeracquisitionisthelargestchallengethatwewillface.Giventhebusinesstobusinessnatureof
ourbusinessmodelandthatcustomersarelikelytohavealargebureaucracywithseveraldecision
makers,oursalestimelinewilllikelytakemonths.Togenerateleads,weplantoattenddigital