ENEE631 Digital Image Processing (Spring'09) Data Hiding in Images Spring ’09 Instructor: Min Wu Electrical and Computer Engineering Department, University of Maryland, College Park   bb.eng.umd.edu (select ENEE631 S’09)   ENEE631 Spring’09 Lecture 21 (4/20/2009) UMCP ENEE631 Slides (created by M.Wu © 2004)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [2] Overview and Logistics Last Time: –Overview of content-based image retrieval –A quick guide on video communications –Introduction on data embedding in images Today: –Continue on data hiding in images UMCP ENEE631 Slides (created by M.Wu © 2004)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [3] Example: Data Embedding by Replacing LSBs Downloaded from steganography/image_downgrading/ UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [4] Example: LSB Replacement (cont’d) Replace LSB with Pentagon’s MSB UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [5] Using Higher LSB Bitplanes for Embedding ……Issues to consider: how much distortion by embedding? how much resilience to minor changes?  smaller distortion  original pixel value  replace 2 nd LSB

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [6] Example: LSB Replacement of Higher Bitplanes Replace 6 LSBs with Pentagon’s 6 MSBs UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [7] Review: Pixel Depth –“Contour” artifacts for low pixel depth at gradual transition areas –Human eyes distinguish about 50 gray levels => 5~6 bits/pixel 8 bits / pixel 4 bits / pixel 2 bits / pixel

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [8] Recall: Embedding Basics – Two Simple Tries Data Hiding: To put secondary data in host signal (1) Replace LSB (2) Round a pixel value to closest even or odd numbers u Both equivalent to reduce effective pixel depth for representing host image u Detection scheme is same as LSB, but embedding brings less distortion in the quantized case, esp. for higher LSB bitplane + Simple embedding;  Fragile to even minor changes even “0” odd “1” pixel value odd-even mapping lookup table mapping … … UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [9] How to Improve the Robustness? Introduce quantization to embedding process –Make features being odd/even multiple of Q Tradeoff between embedding distortion and robustness Larger Q => Higher resilience to minor changes => Higher average changes required to embed data Questions: What’s the expected embedding distortion? Relation with distortion by quantization alone? feature value 2kQ (2k+1)Q (2k+2)Q (2k+3)Q odd-even mapping lookup table mapping … …

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [10] Distortion from Quantization-based Embedding Uniform quantization with step size Q (Assume source’s distribution within each interval is approx. constant) MSE = Q 2 / 12 Odd-even embedding with quantization MSE = ½ * (Q 2 /12) + ½ * (7Q 2 /12) = Q 2 / 3 u MSE equiv. to quantize with 2Q step size! u “Predistort” via quantization to gain resilience [-Q/2, Q/2] feature value 2kQ (2k+1)Q (2k+2)Q (2k+3)Q odd-even mapping Q -Q/2 + Q/2 +Q -Q/2 + Q/2 1/Q

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [11] Two Views of Quantization-based Embedding From decoder’s view –Partition the signal space into two subsets labeled “0” & “1” –Decode according to which subset a sample belongs to u Embedder picks watermarked sample from the subset labeled with to- be-embedded bit, and tries to minimize the amount of changes From embedder’s view –Design two quantizers “#0”, “#1”: step size 2Q, offset by Q –Embedder perform quantization using the quantizer labeled with to-be-embedded bit => “Quantization Index Modulation (QIM)” u Decoder looks for closest representative feature value 2kQ (2k+1)Q (2k+2)Q (2k+3)Q “1” “0”

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [12] Tampering Detection by Pixel-domain Fragile Wmk Downloaded from ICIP’97 CD-ROM paper by Yeung-Mintzer UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [13] Fight Against Forging Tamper-Detection Watermark? If using LSB to embed a fragile watermark for tampering detection, adversary can alter image but retain LSB [Solution 1] Add uncertainty to the embedding mapping –through a random look-up table with controlled run length [Solution 2] Make watermark securely depend on host content E.g. embed a robust/content-base hash of host image feature value 2kQ (2k+1)Q (2k+2)Q (2k+3)Q odd-even mapping lookup table mapping … …

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [14] Pixel-domain Table-lookup Embedding Pixel-domain Table-lookup Embedding (Yeung-Mintzer ICIP’97) –Simple to implement; be able to localize alteration extracted wmk from altered image

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [15] Yeung’s Fragile Watermark for Tampering Detection Basic idea: –enforce certain relationship to embed data –minimize distortion: nearest neighbour, constrained runs –diffuse error incurred to surrounding pixels v’=v+d1+d2: LUT(v’)=b original image marked image lookup table generator LUT( ) seed data to be embedded table lookup test image extracted data LUT( ) visualize &decide embed detect d1: diffused error

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [16] LUT Embedding: Distortion/Security/Robustness What’s new compared with odd-even embedding? –Mapping from feature to embedded bit is less predictable –Adjacent intervals may be mapped to the same bit value How much security gained with proprietary LUT? => => –Proprietary LUT brings uncertainty and makes it difficult for attackers to embed specific data at his/her will How much MSE introduced by embedding? => => –Larger than odd-even embedding How much resilience gained? => => –Moving away by Q/2 step may not trigger detection error u Due to possible continuous run in LUT Ref: M. Wu: "Joint Security and Robustness Enhancement for Quantization Based Embedding," IEEE Trans. on Circuits and Systems for Video Technology, vol. 13, no. 8, pp , August (see ICIP’03 for shorter conf. version)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [17] Security and Distortion by LUT Embedding Security against forgery –Quantify LUT security by the entropy rate of a Markovian model u For run <= 2: entropy is 0.67 bits/entry; or tables for 256-entry Distortion incurred by embedding –With fixed quantization step, long run of LUT brings higher distortion MSE| run=1 = Q 2 / 3 MSE| run  2 = Q 2 / 2

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [18]

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [19] Robustness of LUT Embedding LUT with ≥ 2 run can bring higher robustness –Noise dragging a sample out of the enforced interval does not necessarily incur detection errors Study error probability at the same WNR level –WNR: power ratio of wmk to noise –Increased run gives smaller P e than run=1 except at high WNR

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [20] Case Study: Mintzer-Yeung Patent on Fragile Wmk F. Mintzer and M.M. Yeung: “Invisible Image Watermark for Image Verification,” U.S. Patent 5,875,249, issued Feb Acquire knowledge on latest art in industry from patent –Especially useful when industry don’t publish all key techniques (but they often aggressively patent these “IP”) u Watermark is a good example: Digimarc, Verance, IBM, NEC … –Complementary to literature search of journal/conf. papers For details on how to patent your novel ideas –Talk to your supervisor & lawyers, and check univ./company policies –Resource u Online workshop on Patent 101 (see also the patent handout) u US Patent Office (full-text patent search and patent doc. images) u ENEE698T – Technology Laws (first offered F’08; see future offering)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [21] A Glimpse at the Patent System Intended to add "the fuel of interest to the fire of genius" (Abraham Lincoln) –In exchange for disclosing an invention to the public, the inventor receives the exclusive right to control exploitation of the invention and to realize any profits for a specific length of time Common types of U.S. patent: –Utility patents  of most interest to ECEer u A term up to 20 years from the date the patent application was filed u Granted to anyone who invents a new and useful process, machine, manufacture, or composition of matter, or a new and useful improvement thereof –Design patents u new, original & ornamental design for 14 years’ protection –Plant patents

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [22] Patent Process Idea –Make sure your idea is new and practical/useful Document –Keep records that document your discovery –File “Invention Disclosure” & Be careful with public disclosure Research –Search existing literature & patents related to your invention –Analyze existing patents and literature Apply –Prepare and file the patent application documents –Review by PTO examiner; amend your patent claims if needed

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [23] Structure of A Utility Patent Title Page –Title, Patent Number, File & Issue Dates –Inventors, Assignees, Patent examiners –Related patents and references –Abstract, # of claims, # of drawings, representative drawing Drawings Main text –Field of the Invention (usually in one sentence) –Background –Summary –Brief description of drawings –Detailed description of the preferred embodiment –Claims

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [24] Useful Contents for Technical Studies Detailed descriptions of invention (process/method/apparatus) –Along with drawings (and background/summary) –They are often intended to be written in an easily accessible way Technical discussions on “Preferred embodiment(s)” in Mintzer-Yeung’s patent –Image stamping via LUT embedding –Image verification via Table lookup and visualization –Error diffusion to alleviate visual distortion incurred by embedding –Apply the process to DC-image for embedding data in JPEG image

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [25] Claims: Crucial Part for Business Values Not always “fun” to read –Many legally speaking terms and wordings Usually prefer broad (and allowable) claims Claims are the hot spot examined by USPTO –Determines whether (1) the proposed claims have been claimed by other patents? (2) anticipated by other already issued patents? (3) straightforward combination or extensions of existing methods for similar purposes by those “skilled in the art” 28 Claims in Mintzer-Yeung patent –“Root” claims and “child” claims u Claim Tree: useful tree visualization to illustrate relations between claims

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [26] Reading Assignment (1) M. M. Yeung, F. Mintzer: “An Invisible Watermarking Technique for Image Verification", Proc. of the IEEE Int’l Conf. on Image Processing (ICIP'97), Oct (2) F. Mintzer and M.M. Yeung: “Invisible Image Watermark for Image Verification,” U.S. Patent 5,875,249, issued Feb For further exploration (3) M. Wu and B. Liu: "Watermarking for Image Authentication", ICIP'98, vol.2. (4) M. Wu: "Joint Security and Robustness Enhancement for Quantization Based Embedding," IEEE Trans. on Circuits and Systems for Video Technology, vol. 13, no. 8, pp , August (see ICIP’03 for shorter conf. version)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [27] From Fragile to Robust Watermark UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [28] From Fragile/Semi-Fragile to Robust Watermark Applications of fragile/semi-fragile watermark –Tampering detection –Secret communications => “Steganography” (covert writing) –Convey side info. in a seamless way: lyric, director’s notes Situations demanding higher robustness –Protect ownership (copyright label), prevent leak (digital fingerprint) –Desired robustness against compression, filtering, etc. How to make it robust? –Use “quantization” from signal processing –Use error correcting coding –Borrow theories from signal detection & telecommunications u “Spread Spectrum Watermark”: use “noise” as watermark and add it to the host signal for improved invisibility and robustness UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [29] Background: Robust Watermark / Data-Embedding u Embedding domain tailored to media characteristics & application requirement … © Copyright …

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [30] Spread Spectrum Watermark: Cox et al (NECI) What to use as watermark? Where to put it? –Place wmk in perceptually significant spectrum (for robustness) u Modify by a small amount below Just-noticeable-difference (JND) –Use long random noise-like vector as watermark u for robustness/security against jamming+removal & imperceptibility Embedding v’ i = v i +  v i w i = v i (1+  w i ) –Perform DCT on entire image and embed wmk in DCT coeff. –Choose N=1000 largest AC coeff. and scale {v i } by a random factor 2D DCTsort v’=v (1+  w) IDCT & normalize Original image N largest coeff. other coeff. marked image random vector generator wmk seed UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [31] Watermarking Example by Cox et al. Original Cox Difference between whole image DCTmarked & original Embed in 1000 largest coeff. UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [32] Cox et al’s Scheme (cont’d): Detection  Subtract original image from the test one before feeding to detector (“non-blind detection”)  Correlation-based detection u a correlator normalized by |Y| in Cox et al. paper DCT compute similarity threshold test image decision wmk DCTselect N largest original unmarked image select N largest preprocess – – orig X test X’ X’=X+W+N ? X’=X+N ? To think UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [33] Performance of Cox et al’s Scheme Robustness –(claimed) scaling, JPEG, dithering, cropping, “printing-xeroxing- scanning”, multiple watermarking –No big surprise with high robustness u equivalent to sending just 1-bit {0,1} with O(10 3 ) samples Comment –Must store orig. unmarked image  “private wmk”, “non-blind” detection –Perform image registration if necessary –Adjustable parameters: N and  UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [34] Comments on Cox et al’s Schemes “1000 largest coeff.” before and after embedding –May not be identical (and order may also changes) –Solutions: use original as ref; use “embeddable” mask to maintain synch. Detection without using original/host image –Treat host image as part of the noise/interference ~ Blind detection u need long wmk signal to combat severe host interference [Zeng-Liu] –Can do better than blind detection, as embedder knows the host signal => “Embedding with Side Info.” H0: H1: = vs. H0: = H1: =

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [35] Improve Invisibility and Robustness on Cox scheme Apply better Human Perceptual Model –Global scaling factor is not suitable for all coefficients –More explicitly compute just-noticeable-difference (JND) u JND ~ max amount each coefficient can be modified invisibly u Employ human visual model: freq. sensitivity, masking, … –Use more localized transform => fine tune wmk for each region u block-based DCT; wavelet transform Improve robustness: detection performance depends on ||s|| /  d u Add a watermark as strong as JND allows u Embed in as many “embeddable” coefficients => improve robustness Block-DCT schemes: Podichuk-Zeng; Swanson-Zhu-Tewfik ’97 –Leverage existing visual model for block DCT from JPEG UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [36] Perceptual Comparison: Cox vs. Podilchuk OriginalCox Podilchuk whole image DCT block-DCT Embed in 1000 largest coeff. Embed to all “embeddables” UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [37] Compare Cox & Podilchuk Schemes (cont’d) CoxPodilchuk Amplified pixel-wise difference between marked and original (gray~0) UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [38] Comments on Cox & Podilchuk’s SS Wmk Robustness –Very robust against additive noise (seen from detection theory) –Sensitive to synchronization errors, esp. under blind detection u jitters (line dropping/addition) u geometric distortion (rotation, scale, translation) Question: How to improving synchronization resilience? => add registration pattern; embed in RST-invariant domain; …

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [39] Suggested Readings 1.I. Cox, J. Kilian, T. Leighton, T. Shamoon: “Secure Spread Spectrum Watermarking for Multimedia'', IEEE Trans. on Image Proc., vol.6, no.12, pp , M. M. Yeung and F. Mintzer: “An Invisible Watermarking Technique for Image Verification", Proc. of the IEEE Int’l Conf. on Image Processing (ICIP), Oct M. Wu and B. Liu: "Data Hiding in Image and Video: Part-I -- Fundamental Issues and Solutions", IEEE Trans. on Image Proc., vol.12, no.6, pp , June M. Wu, W. Trappe, Z.J. Wang, and K.J.R. Liu: “Collusion-resistant fingerprinting for Multimedia,” IEEE Signal Proc Magazine, March M. Wu and B. Liu: Multimedia Data Hiding, Springer-Verlag, I. Cox, M. Miller, and J. Bloom: Digital Watermarking, Morgan Kauffman, And the related references cited by these publications. UMCP ENEE631 Slides (created by M.Wu © 2004)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [40] Localized Embedding: Double-Edge Sword “Innocent Tools” exploited by attackers: block concealment Recovery of lost blocks –for resilient multimedia transmission of JPEG/MPEG –good quality by edge-directed interpolation: Jung et al; Zeng-Liu Remove robust watermark by block replacement edge estimation edge-directed interpolation UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [41] Block Replacement Attack

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [42] Attack effective on block-DCT based spread-spectrum watermark marked original (no distortion)JPEG 10%after proposed attack 512x512 lenna Threshold: 3 ~ 6 Recall: claimed high robustness&quality by fine tuning wmk strength for each region

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [43]

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [44] Watermark Attacks: What and Why? Attacks: intentionally obliterate watermarks –remove a robust watermark –make watermark undetectable (e.g., miss synchronization) –uncertainty in detection (e.g., multiple ownership claims) –forge a valid (fragile) watermark –bypass watermark detector Why study attacks? –identify weaknesses –propose improvement –understand pros and limitation of tech. solution To win each campaign, a general should know both his troop and the opponent’s as well as possible. -- Sun Tzu, The Art of War, 500 B.C. UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [45] Summary: Spread Spectrum Embedding Main ideas –Place wmk in perceptually significant spectrum (for robustness) u Modify by a small amount below Just-noticeable-difference (JND) –Use long random vector of low power as watermark to avoid artifacts (for imperceptibility, robustness, and security) Cox’s approach –Perform DCT on entire image & embed wmk in large DCT AC coeff. –Embedding: v’ i = v i +  v i w i = v i (1+  w i ) –Detection: subtract original and perform correlation w/ wmk Podilchuk’s improvement –Embed in many “embeddable” coeff. in block-DCT domain –Adjust watermark strength by explicitly computing JND

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [46] Suggested Readings 1.I. Cox, J. Kilian, T. Leighton, T. Shamoon: “Secure Spread Spectrum Watermarking for Multimedia'', IEEE Trans. on Image Proc., vol.6, no.12, pp , M. M. Yeung, F. Mintzer: “An Invisible Watermarking Technique for Image Verification", Proc. of the IEEE Int’l Conf. on Image Processing (ICIP), Oct M. Wu and B. Liu: "Data Hiding in Image and Video: Part-I -- Fundamental Issues and Solutions", IEEE Trans. on Image Proc., vol.12, no.6, pp , June M. Wu, W. Trappe, Z.J. Wang, and K.J.R. Liu: “Collusion-resistant fingerprinting for Multimedia,” IEEE Signal Proc Magazine, March M. Wu and B. Liu: Multimedia Data Hiding, Springer-Verlag, I. Cox, M. Miller, and J. Bloom: Digital Watermarking, Morgan Kauffman, And the related references cited by these publications. UMCP ENEE631 Slides (created by M.Wu © 2004)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [47] Summary: Type-I Additive Embedding Add secondary signal in host media Representative: spread spectrum embedding –Add a noise-like signal and detection via correlation –Good tradeoff between security, imperceptibility & robustness –Limited capacity: host signal often appears as major interferer modulation data to be hidden  X original source X’ = X +  marked copy = UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [48] Type-II Relationship Enforcement Embedding Deterministically enforcing relationship –Secondary information carried solely in watermarked signal –Typical relationship: parity/modulo in quantized features Representative: odd-even (quantized) embedding –Alternative view: switching between two quantizers w/ step size 2Q u “Quantization Index Modulation” –Robustness achieved by quantization or tolerance zone –High capacity but limited robustness even “0” odd “1” feature value 2kQ (2k+1)Q (2k+2)Q (2k+3)Q odd-even mapping lookup table mapping … … UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [49] Robustness vs. Payload Blind/non-coherent detection ~ original copy unavailable Robustness and payload tradeoff Advanced embedding: quantization w/ distortion-compensation –Combining the two types with techniques suggested by info. theoryRobustnessPayload Imperceptibility noise stronger noise weaker UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [50] Distortion Compensated Quantization Embedding Distortion compensation technique –Increase quantization step by a factor  for higher robustness –Compensate the extra distortion by dragging the enforced feature toward the original feature value Overall embedding distortion unchanged Choose alpha to maximize a distortion-compensation SNR odd/even mapping UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [51] Improved Robustness by Distortion Compensation –ICS (ideal Costa’s scheme) –SS (spread spectrum additive embedding) –binary DM (odd-even quantized embedding) –binary SCS (odd-even quantized embedding with distortion compensation)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [52] Issues to Address Tradeoff among conflicting requirements –Imperceptibility –Robustness & security –Capacity Key elements of data hiding –Perceptual model –Embedding one bit –Multiple bits –Uneven embedding capacity –Robustness and security –What data to embed Upper Layers Uneven capacity equalization Error correction Security …… Lower Layers Imperceptible embedding of one bit Multiple-bit embedding Coding of embedded data Robustness Capacity Imperceptibility UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [53] Robust Wmk Application for Tracing Traitors Leak of information as well as alteration and repackaging poses serious threats to government operations and commercial markets –e.g., pirated content or classified document Promising countermeasure: robustly embed digital fingerprints –Insert ID or “fingerprint” (often through conventional watermarking) to identify each user –Purpose: deter information leakage; digital rights management (DRM) –Challenge: imperceptibility, robustness, tracing capability studio The Lord of the Ring Alice Bob Carl w1 w2 w3Sell UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [54] Embedded Fingerprint for Tracing Traitors Insert special signals to identify recipients –Deter leak of proprietary documents –Complementary protection to encryption –Consider imperceptibility, robustness, traceability –Attacks mounted by single and multiple users Multi-user Attacks Traitor Tracing President Satellite Image Alice Bob Carl w1 w2 w3 Leak

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [55] Potential civilian use for digital rights management (DRM) u u Copyright industry – $500+ Billion business ~ 5% U.S. GDP Alleged Movie Pirate Arrested (23 January 2004) – –A real case of a successful deployment of 'traitor-tracing' mechanism in the digital realm – –Use invisible fingerprints to protect screener copies of pre- release movies Carmine CaridiRussellfriends … Internet w1 Last Samurai Hollywood studio traced pirated version Case Study: Tracing Movie Screening Copies UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [56] Collusion Attacks by Multiple Users Collusion: A cost-effective attack against multimedia fingerprints – –Users with same content but different fingerprints come together to produce a new copy with diminished or attenuated fingerprints – –Fairness: Each colluder contributes equal share through averaging, interleaving, and nonlinear combining Collusion by averaging 1/3 AliceChris Bob Colluded copy Originally fingerprinted copies AliceChris Collage

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [57] Key Issues How to construct fingerprints? –Identify individual users –Resist multi-user collusion How to embed fingerprints in media data? –Tailor to media characteristics for robustness & imperceptibility Interaction between choices of fingerprint construction, embedding, and detection –esp. to combat collusion attacks –Analogous to “cross-layer” methods in communications

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [58] Road Map on Media Fingerprinting Research Robust Embedding Orthogonal Fingerprints Represent how many users? Resist how many colluders? “most effective” collusions? Amount of resources used? Group-based FP to exploit Attacker Behavior Coded FP Joint Coding-Embedding Framework overcome prior work’s problems of long code length, low resilience, and limited scalability adapt to media characteristics Combinatorial codes + CDM Error correcting codes + TDM Correlated Fingerprints

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [59] Example: Orthogonal Fingerprint for Curves/Graphics Use (approx.) orthogonal sequences as FPs for different users –Detection by looking for high correlation result Embed in parametric modeling domain of curve –Perturb B-spline parameters according to spread spectrum sequences Detection Statistics Typical threshold is 3~6 for false alarm of ~ Original Curve (captured by TabletPC) Fingerprinted Curve (100 control points)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [60] Fingerprinting Topographic Map –Traditional protection: intentionally alter geospatial content –Embed much less intrusive digital fingerprint for a modern protection 9 long curves are marked; 1331 control points used to carry the fingerprint 1100x1100 Original MapFingerprinted Map

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [61] Collusion-Resistant Fingerprinting of Maps 2-User Interleaving Attack5-User Averaging Attack...   Can survive combined attacks of collusion + print + scan   Can extend to 3-D Digital Elevation Map

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [62] Explore unique issues associated with multimedia – –Consider interaction b/w fingerprint encoding, embedding & detection – –Use appropriate embedding to prevent arbitrary manipulation on code Build correlated fingerprints in two steps – –Use antipodal coded modulation to embed fingerprint codes u u via orthogonal spread spectrum sequences u u shared bits get sustained and used to identify colluders – –Binary Anti-collusion fingerprint codes resist up to K colluders any subset of up to K users take same unique set of positions 1st bit 2nd bit... Spreading + Combinatorial Coded Fingerprint

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [63] 16-bit ACC for Detecting  3 Colluders Out of 20 User-1 ( -1,-1, -1, -1, 1, 1, 1, 1, …, 1 ) ( -1, 1, 1, 1, 1, 1, …, -1, 1, 1, 1 ) User-4 Extracted fingerprint code ( -1, 0, 0, 0, 1, …, 0, 0, 0, 1, 1, 1 ) Collude by Averaging Uniquely Identify User 1 & 4 Embed fingerprint via HVS-based spread spectrum embedding in block-DCT domain

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [64] Suggested Readings 1.I. Cox, J. Kilian, T. Leighton, T. Shamoon: “Secure Spread Spectrum Watermarking for Multimedia'', IEEE Trans. on Image Proc., vol.6, no.12, pp , M. M. Yeung, F. Mintzer: “An Invisible Watermarking Technique for Image Verification", Proc. of the IEEE Int’l Conf. on Image Processing (ICIP), Oct M. Wu and B. Liu: "Data Hiding in Image and Video: Part-I -- Fundamental Issues and Solutions", IEEE Trans. on Image Proc., vol.12, no.6, pp , June M. Wu, W. Trappe, Z.J. Wang, and K.J.R. Liu: “Collusion-resistant fingerprinting for Multimedia,” IEEE Signal Proc Magazine, March M. Wu and B. Liu: Multimedia Data Hiding, Springer-Verlag, I. Cox, M. Miller, and J. Bloom: Digital Watermarking, Morgan Kauffman, And the related references cited by these publications. UMCP ENEE631 Slides (created by M.Wu © 2004)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [65]

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [66] Quick Review on Signal Detection Hypothesis testing –Error-free case: for observation values where only one hypothesis having non-zero probability (or prob. density)

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [67] Quick Review on Signal Detection (cont’d) Minimize probability of detection error –Equal probable prior: maximum likelihood (ML) detection decision rule:arg j max P j (y) –General prior probability: scale by prior probability decision rule: arg j max P(H j ) P j (y) = arg j max P(H j | y) u Maximum A Posteriori (MAP) detection u Likelihood ratio test with proper threshold

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [68] Review on Detection of Signal Sequence in Noise Formulation u Known signals; noise values unknown but statistics known Under equal prior probability and AWGN noise –Equivalent to minimum distance detection –For equal-power signals, equiv. to maximum correlation detection (1) Antipodal signaling: s 1 =  s 0 = s (2) On-Off signaling: s 0 = 0; s 1 = s u detection performance proportional to the square root of ratio between signal energy and noise power

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [69] Conveying One-bit Info. Through Noisy Channel [ Known signal s; noise values unknown but statistics known ] Optimal detection ~ minimize prob. of error MAP ~ maximize posterior probability …P(H ? |y) => ML ~ maximum likelihood detector [for equal prior] … P(y|H ? ) => Minimum distance detector [for i.i.d. Gaussian noise] … ||y – bs|| => Maximum correlation detector [for equal-energy sig.] Detection statistic (under i.i.d. Gaussian noise) [correlator]  i y i s i  normalize variance prob. distribution under each hypothesis ~ N(  ||s|| 2, ||s|| 2  d 2 ) T b=1 b=  1

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [70] Illustration: Anti-podal Signaling & Detection Project onto the deterministic signal and check correlation –Positive high correlation => (s1, s2) being sent with high confidence; –Negative high correlation => (-s1, -s2) being sent Use long signal to boost SNR –Each component may be subject to low power constraints, but overall signal energy can be large by increasing sig. length => two sig. point (red dots) can be widely separately, thus reliably detected We can analyze on-off keying similarly.

ENEE631 Digital Image Processing (Spring'09) Lec 21 – Data Hiding [71] Hypothesis Testing for Ownership Watermark Optimal detection for On-Off Keying (OOK) –under i.i.d. Gaussian noise {d i } u use a correlator-type detector u b  {0,1} represents absence vs. presence of ownership mark –Need to determine how to choose watermark signal {s i } Neyman-Pearson Detection [Poor’s book Sec.2.4] –False-alarm: claiming wmk existence when nothing embedded –Given max. allowed false-alarm, maximize detection probability u Use likelihood ratio as detection statistic u Determine threshold based on false-alarm prob. => Q(3)=10 -3 ; Q(6)=10 -5 –Pd vs. Pfp tradeoff ~ ROC curve UMCP ENEE631 Slides (created by M.Wu © based on Research Talks ’98-’04) T presence absence