際際滷

際際滷Share a Scribd company logo

Isit1

S
shilpa kadam

This document provides an overview of fundamentals of watermarking and data hiding. It discusses basic techniques, performance analysis in terms of error probabilities and capacity, and applications to images and advanced topics. The document outlines the tutorial, which covers data hiding, watermarking and steganography applications, embedding data in images, basic properties like fidelity and robustness, attack models, and theoretical concepts from information theory, game theory, coding theory and cryptography that are relevant to understanding fundamental limits.

1 of 24
Download to read offline
Fundamentals of Watermarking and Data Hiding Pierre Moulin University of Illinois at UrbanaChampaign Dept of Electrical and Computer Engineering moulin@ifp.uiuc.edu July 9, 2006 ISIT Tutorial, Seattle c 2006 by Pierre Moulin. All rights reserved. 1
Outline 1. Overview 2. Basic Techniques 3. Binning Schemes and QIM Codes 4. Performance Analysis: Error Probabilities 5. Performance Analysis: Capacity 6. Applications to Images & Advanced Topics 2
SESSION 1: OVERVIEW Data hiding, watermarking, steganography Basic properties: 鍖delity, payload, robustness, security 3
Some Reading Books: Digital Watermarking, by I. Cox, M. Miller, J. Bloom, Morgan-Kaufmann, 2002 Information Hiding Techniques for Steganography and Digital Watermarking, by S. Katzenbeisser and F. Petitcolas, Eds., Artech House, 2000 Information Hiding: Steganography and Watermarking, by N. Johnson, Z. Duric and S. Jajodia, Kluwer, 2000 4
New IEEE Transactions on Information Forensics and Security (quarterly, inaugural issue in March 2006) Special issues of various IEEE journals, 1999 2005 Annual Information Hiding Workshops Watermarking newsletter: www.watermarkingworld.org www.ifp.uiuc.edu/moulin Tutorial paper Data Hiding Codes by P. Moulin and R. Koetter, Proceedings IEEE, December 2005. 5
Multimedia Security Dissemination of digital documents Owner identi鍖cation Forgery detection Identi鍖cation of illegal copies Intellectual protection 6
Authentication 7
8
Media Elements Audio Images Video Graphics Documents Computer programs 9
Nonadversarial Applications Database annotation Information embedding, e.g., audio in images, text in host signals (movie subtitles, 鍖nancial data, synchronization signals) 10
Data Hiding Embed data in covertext (high payload) Perceptual similarity requirement Multimedia database management Covert communications (military, spies, etc.) Steganography ( 粒留僚 粒留, covert writing): conceal existence of hidden message 11
Watermarking Hide a few bits of information Original and modi鍖ed signals should be perceptually similar Application to digital cameras, TV, DVD video, audio Authentication Transaction tracking Broadcast monitoring 12
Fingerprinting Fingerprinter marks several copies of original and distributes copies to users 1, 2, 揃 揃 揃 , L Each mark is di鍖erent Users may collude to remove watermarks Applications: copy control, traitor tracing 13
Summary of Applications Applications Watermarking authentication, copyright protection Data hiding covert communications, database annotation, information embedding Steganography covert communications Fingerprinting copy control, traitor tracing 14
A Brief History Tattoo hidden message on head of slave (ancient Greeks) Invisible ink Secret point patterns Watermarks in paper (Italy, 13th century) Digital watermarking: early 1990s Standardization attempts: SDMI (music), ISO (MPEG video) 15
Hiding Data in Images secret key k Encoder original image S watermarked image X Picture taken by Alice on January 1, 2000. This message is going to be embedded forever in this picture. I challenge you to remove the message without substantially altering the picture. 1001001101001110100...............101 binary representation Decoder Picture taken by Alice on January 1, 2000. This message is going to be embedded forever in this picture. I challenge you to remove the message without substantially altering the picture. Decoded message 1001001101001110100...............101 Decoded binary message secret key k Attack Pirate 11011000...01 16
Decoders Task 17
Attacks on Images Original JPEG, QF=10 4 4 median 鍖ltering Gaussian 鍖lter ( = 3) Rotated by 10 degrees Random bend 18
Basic Properties Fidelity (in terms of signal distortion metric) Payload (number of transmitted bits) Robustness (against adversary) Security (cryptanalysis of randomized code) Detectability (by steganalyzers/eavesdroppers) 19
System Issues System complexity Does decoder know host signal? (public vs private watermarking) Security level? Reliance on private or public cryptographic system? 20
Attack Models No attack Deterministic attacks (reversible & irreversible) Stochastic attacks (memoryless & stationary) Code breaking System attacks (e.g., ambiguity, sensitivity & scrambling) Benchmarking (e.g., Stirmark) 21
Attacks Attack Type Examples Memoryless independent noise, random pixel replacement Blockwise memoryless JPEG compression Attacks with stationary noise, statistical regularity spatially invariant 鍖ltering, some estimation attacks Deterministic compression, format changes Arbitrary attacks cropping, permutations, desynchronization, nonstationary noise 22
Basic Theoretical Concepts Information theory Game theory Detection and estimation theory Coding theory Cryptography 23
Purposes of an Information-Theoretic Approach make appropriate simplifying assumptions to understand fundamental limits of IH and optimally design algorithms provide new insights into IH provide a precise framework for evaluating any IH algorithm develop approach that generalizes easily to related problems Caution: cost of mismodeling may be severe in game with opponent! 24

More Related Content

Isit1

  • 1. Fundamentals of Watermarking and Data Hiding Pierre Moulin University of Illinois at UrbanaChampaign Dept of Electrical and Computer Engineering moulin@ifp.uiuc.edu July 9, 2006 ISIT Tutorial, Seattle c 2006 by Pierre Moulin. All rights reserved. 1
  • 2. Outline 1. Overview 2. Basic Techniques 3. Binning Schemes and QIM Codes 4. Performance Analysis: Error Probabilities 5. Performance Analysis: Capacity 6. Applications to Images & Advanced Topics 2
  • 3. SESSION 1: OVERVIEW Data hiding, watermarking, steganography Basic properties: 鍖delity, payload, robustness, security 3
  • 4. Some Reading Books: Digital Watermarking, by I. Cox, M. Miller, J. Bloom, Morgan-Kaufmann, 2002 Information Hiding Techniques for Steganography and Digital Watermarking, by S. Katzenbeisser and F. Petitcolas, Eds., Artech House, 2000 Information Hiding: Steganography and Watermarking, by N. Johnson, Z. Duric and S. Jajodia, Kluwer, 2000 4
  • 5. New IEEE Transactions on Information Forensics and Security (quarterly, inaugural issue in March 2006) Special issues of various IEEE journals, 1999 2005 Annual Information Hiding Workshops Watermarking newsletter: www.watermarkingworld.org www.ifp.uiuc.edu/moulin Tutorial paper Data Hiding Codes by P. Moulin and R. Koetter, Proceedings IEEE, December 2005. 5
  • 6. Multimedia Security Dissemination of digital documents Owner identi鍖cation Forgery detection Identi鍖cation of illegal copies Intellectual protection 6
  • 8. 8
  • 9. Media Elements Audio Images Video Graphics Documents Computer programs 9
  • 10. Nonadversarial Applications Database annotation Information embedding, e.g., audio in images, text in host signals (movie subtitles, 鍖nancial data, synchronization signals) 10
  • 11. Data Hiding Embed data in covertext (high payload) Perceptual similarity requirement Multimedia database management Covert communications (military, spies, etc.) Steganography ( 粒留僚 粒留, covert writing): conceal existence of hidden message 11
  • 12. Watermarking Hide a few bits of information Original and modi鍖ed signals should be perceptually similar Application to digital cameras, TV, DVD video, audio Authentication Transaction tracking Broadcast monitoring 12
  • 13. Fingerprinting Fingerprinter marks several copies of original and distributes copies to users 1, 2, 揃 揃 揃 , L Each mark is di鍖erent Users may collude to remove watermarks Applications: copy control, traitor tracing 13
  • 14. Summary of Applications Applications Watermarking authentication, copyright protection Data hiding covert communications, database annotation, information embedding Steganography covert communications Fingerprinting copy control, traitor tracing 14
  • 15. A Brief History Tattoo hidden message on head of slave (ancient Greeks) Invisible ink Secret point patterns Watermarks in paper (Italy, 13th century) Digital watermarking: early 1990s Standardization attempts: SDMI (music), ISO (MPEG video) 15
  • 16. Hiding Data in Images secret key k Encoder original image S watermarked image X Picture taken by Alice on January 1, 2000. This message is going to be embedded forever in this picture. I challenge you to remove the message without substantially altering the picture. 1001001101001110100...............101 binary representation Decoder Picture taken by Alice on January 1, 2000. This message is going to be embedded forever in this picture. I challenge you to remove the message without substantially altering the picture. Decoded message 1001001101001110100...............101 Decoded binary message secret key k Attack Pirate 11011000...01 16
  • 18. Attacks on Images Original JPEG, QF=10 4 4 median 鍖ltering Gaussian 鍖lter ( = 3) Rotated by 10 degrees Random bend 18
  • 19. Basic Properties Fidelity (in terms of signal distortion metric) Payload (number of transmitted bits) Robustness (against adversary) Security (cryptanalysis of randomized code) Detectability (by steganalyzers/eavesdroppers) 19
  • 20. System Issues System complexity Does decoder know host signal? (public vs private watermarking) Security level? Reliance on private or public cryptographic system? 20
  • 21. Attack Models No attack Deterministic attacks (reversible & irreversible) Stochastic attacks (memoryless & stationary) Code breaking System attacks (e.g., ambiguity, sensitivity & scrambling) Benchmarking (e.g., Stirmark) 21
  • 22. Attacks Attack Type Examples Memoryless independent noise, random pixel replacement Blockwise memoryless JPEG compression Attacks with stationary noise, statistical regularity spatially invariant 鍖ltering, some estimation attacks Deterministic compression, format changes Arbitrary attacks cropping, permutations, desynchronization, nonstationary noise 22
  • 23. Basic Theoretical Concepts Information theory Game theory Detection and estimation theory Coding theory Cryptography 23
  • 24. Purposes of an Information-Theoretic Approach make appropriate simplifying assumptions to understand fundamental limits of IH and optimally design algorithms provide new insights into IH provide a precise framework for evaluating any IH algorithm develop approach that generalizes easily to related problems Caution: cost of mismodeling may be severe in game with opponent! 24