Recent Progress in Info ForensicsRecent Progress in Info Forensics Min Wu MediaandSecurityTeamMAST Media ECE Department UMIACSUniversity of Maryland College Park Include joint work with WeiH ID: 109885
Download Pdf The PPT/PDF document "Seeing the Invisibles:Seeing the Invisib..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Seeing the Invisibles:Seeing the Invisibles: Recent Progress in Info. ForensicsRecent Progress in Info. Forensics Min Wu MediaandSecurityTeam(MAST) Media ECE Department / UMIACSUniversity of Maryland, College Park Include joint work with Wei-Hong Chuang, Ravi Garg, HongmeiGou, AdiHajj-Ahmad, K.J. Ray Liu, HuiSu, AshwinSwaminathan, and AvinashVarna. Questions about Digital Images Questions about Digital Images and Video Photographed? Scanned? Or computer generated? When and where captured? Post on YouTube or not? Min Wu (UMD): Seeing Invisibles -Spring 2013 Illustration by The New York Times; photo via AgenceFrance-Presse Many Forms of Digital FingerprintsMany Forms of Digital Fingerprints Many types of fingerprints for multimedia protection & management I C EI C E I . . . I . . . track individual copy and trace unauthorized use Ctt bd Fiit C on b d Fi t Compact content signature for content identification, and also useful for watermarking and content authentication Examine inherent traces left on multimedia by device or processing Providenon intrusiveforensicstodetermineoriginintegrityetc Min Wu (UMD): Seeing Invisibles -Spring 2013 - , Many Forms of Digital FingerprintsMany Forms of Digital Fingerprints Many types of fingerprints for multimedia protection & management I C EI C E I . . . I . . . track individual copy and trace unauthorized use Ctt bd Fiit C on b d Fi t Compact content signature for content identification, and also useful for watermarking and content authentication WhiH Alicew1 Leak Examine inherent traces left on multimedia by device or processing Providenon intrusiveforensicstodetermineoriginintegrityetc Whi te Leak Min Wu (UMD): Seeing Invisibles -Spring 2013 - , Many Forms of Digital FingerprintsMany Forms of Digital Fingerprints Many types of fingerprints for multimedia protection & management I C EI C E I . . . I . . . Shazamapp track individual copy and trace unauthorized use Ctt bd Fiit Shazam for iPhone C on b d Fi t Compact content signature for content identification, and also useful for watermarking and content authentication Examine inherent traces left on multimedia by device or processing Providenon intrusiveforensicstodetermineoriginintegrityetc Min Wu (UMD): Seeing Invisibles -Spring 2013 - , Many Forms of Digital FingerprintsMany Forms of Digital Fingerprints Many types of fingerprints for multimedia protection & management I C EI C E I . . . I . . . track individual copy and trace unauthorized use Ctt bd Fiit C on b d Fi t Compact content signature for content identification, and also useful for watermarking and content authentication Examine inherent traces left on multimedia by device or processing Providenon intrusiveforensicstodetermineoriginintegrityetc Min Wu (UMD): Seeing Invisibles -Spring 2013 - , Intrinsic Traces in Images and VideoIntrinsic Traces in Images and Video Digital / software components of device or processing system f Ensemble properties o f analog components: e.g. statistical noise R ? ? ? ? ? ? Candidate CFApattern CFA Interpolation CFA Fitting errorFitting error 7 Scannermodel 2 Intrinsic Traces in Images and VideoIntrinsic Traces in Images and Video Digital / software components of device or processing system f Ensemble properties o f analog components: e.g. statistical noise individualityofcapturingdeviceorenvironment Unreproducible/ unclonable individual properties~ e.g. individual variation from analog part of sensors due to manufacturingvariability manufacturing Min Wu (UMD): Seeing Invisibles -Spring 2013 Samsung i760 Apple iPhone 3G Intrinsic Traces in Images and VideoIntrinsic Traces in Images and Video Digital / software components of device or processing system f Ensemble properties o f analog components: e.g. statistical noise individualityofcapturingdeviceorenvironment Unreproducible/ unclonable individual properties~ e.g. individual variation from analog part of sensors due to manufacturingvariability manufacturing Unique time-varying location-dependent conditions during capture DttiCt dtti t on vs. C oun r - d e t Determine integrity, origin, time/location, processing history. etc.Remove detectable/inferable traces for privacy sanitization Min Wu (UMD): Seeing Invisibles -Spring 2013 Questions on Time and Placeon Time and Place 500 600 - -80 300 400 500 e (in seconds) -120 -110 -100 10 104 108 100 200 -150 -140 -130 was the video actually shot? And Wasthe soundtrackcapturedatthesametime asthe 9 . 6 . 4 . 8 Frequency (in Hz) Was picture? Or super-imposed afterward? fingerprintinfluencedbypowergrid ontosensor Min Wu (UMD): Seeing Invisibles -Spring 2013 Ubiquitous Forensic Fingerprints from Power GridUbiquitous Forensic Fingerprints from Power Grid 400 500 nds) -40 -20 0.7 0.8 0.9 efficient 500 600 nds) -80 100 200 300 Time (in sec o -100 -80 -60 0.3 0.4 0.5 0.6 Correlation co 200 300 Time (in seco -150 -140 -130 -120 -110 49.5 50 50.5 51.5 Frequency (in Hz) - - - 0 Time frame lag ENF matchin g result demonstratin g similar variations in the ENF Video ENF signalPower ENF signal 9.6 10 10.4 10.8 Frequency (in Hz) NetworkFrequency signal extracted from video and from power signal recorded in India Variesslightlyovertrendsconsistentseenheardrecordings determinerecordingtime/locationdetecttamperingetc Min Wu (UMD): Seeing Invisibles -Spring 2013 determine recording time/location , detect tampering , etc . Ref: Ravi-Varna-Wu paper in ACM Multimedia 2011 Tampering DetectionTampering Detection ENF signal from Video matchingresultdemonstratingdetectiontamperingtraces 10.2 10.3 quency (in Hz)Inserted 480 800 Time (in seconds) Hz)Ground truth ENF signal 480 800 Ti(id) Frequency (i n a itidttd n secon d s ) p i a l b e d e t d b y ev correspondingMin Wu (UMD): Seeing Invisibles -Spring 2013 Forensic Binding of Audio and Visual TracksForensic Binding of Audio and Visual Tracks correlation (3) applications:Min Wu (UMD): Seeing Invisibles -Spring 2013 Ref: Chuang-Ravi-Wu paper in ACM CCS 2012 Explore Machine Learning to Infer LocationExplore Machine Learning to Infer Location estimation from sensing signals containing ENF traces Preliminarily identified useful features for average 85% accuracy m ates (Hz) m ates (Hz) m ates (Hz) r equency Esti m r equency Esti m e quency Esti m F r F r Time (secs) Time (secs) Time (secs) Min Wu (UMD): Seeing Invisibles -Spring 2013 Can ENF Pinpoint to Locations Within a Grid?Can ENF Pinpoint to Locations Within a Grid? Main trend of ENF is known to be same in a gridMicroscopic traces due to localized effect Small variations arent felt between places far apartDynamic distributed control to stabilize power grid has a response p ro g ation s p eed of about 500 miles p er second ppg pp Our multi-location studies in U.S. east and west grids Min Wu (UMD): Seeing Invisibles -Spring 2013 (a) ENF signals from different locations of US Eastern grid after high-pass filtering Include joint work with Wei-Hong Chuang, Ravi Garg, HongmeiGou, AdiHajj-Ahmad, K.J. Ray Liu, HuiSu, AshwinSwaminathan, and AvinashVarna. Min Wu (UMD): Seeing Invisibles -Spring 2013