Introduction to Introduction to Visible Visible

Watermarking Watermarking IPR Course: TA LectureIPR Course: TA Lecture

2002/12/182002/12/18

NTU CSIE R105NTU CSIE R105

OutlineOutline

IntroductionIntroduction State-of-the-ArtState-of-the-Art Characteristics of Visible Characteristics of Visible

Watermarking SchemesWatermarking Schemes Attacking Visible Watermarking Attacking Visible Watermarking

SchemesSchemes Discussions and ConclusionsDiscussions and Conclusions

Classifying Watermarking Classifying Watermarking SchemesSchemes

Data hiding

Fragilewatermarking

Watermarking

Imperceptiblewatermarking

Visiblewatermarking

Robustwatermarking

Visibledata embedding

Steganography

Non-robust data embedding

Robustdata embedding

Imperceptibledata embedding

Visible WatermarkingVisible Watermarking

+

• IPR protection schemes for images and video that have to be released for certain purposes• Unobtrusive copyright patterns can be recognized on embedded contents

Invisible Watermarking v.s. Invisible Watermarking v.s. Visible WatermarkingVisible Watermarking

Invisible Watermarking Visible Watermarking

Watermark Perceptibility

Protection

Robustness

Imperceptible distortion

Visibly Meaningful pattern

Intentional attacks and

common signal processing

User-intervention based

watermark removal

Passive Active

Current Research Status

Hot Only few papers

ExtractionExplicit

extraction module

Direct viewing

Requirements of Requirements of Visible WatermarkingVisible Watermarking

Perceptibility of host image details Perceptibility of host image details Contents should not be rendered useless after Contents should not be rendered useless after

being visibly watermarked being visibly watermarked Visibility of watermark patterns in Visibility of watermark patterns in

embedded contents embedded contents No explicit watermark extraction techniques No explicit watermark extraction techniques

are requiredare required RobustnessRobustness

Difficult to remove unless exhaustive and Difficult to remove unless exhaustive and expensive human interventions are involvedexpensive human interventions are involved

A General Model of A General Model of Visible WatermarkingVisible Watermarking

I’: the watermarked contentI’: the watermarked content I: the un-watermarked original contentI: the un-watermarked original content W: the watermark pattern W: the watermark pattern KKii: the weighting factor: the weighting factor D: a distance function measuring the perceptual difference of its two paraD: a distance function measuring the perceptual difference of its two para

metersmeters EEii: image feature extraction operators: image feature extraction operators ThresholdThresholdII: the largest allowable distortion of image details that observers : the largest allowable distortion of image details that observers

can tolerate and, at the same time, the signature of can be maintained. can tolerate and, at the same time, the signature of can be maintained. ThresholdThresholdWW: the largest allowable distortion of the embedded watermark p: the largest allowable distortion of the embedded watermark p

attern that the copyright information can be clearly recognized. attern that the copyright information can be clearly recognized.

WKIKI **' 21

III ThresholdIEIED ))(),'((

WWW ThresholdWEIED ))( ),'((

State-of-the-ArtState-of-the-Art G. Braudaway, K.A. Magerlein, and F. Mintzer, "Protecting Publicly G. Braudaway, K.A. Magerlein, and F. Mintzer, "Protecting Publicly

Available Images with a Visible Image Watermark," Proceedings of Available Images with a Visible Image Watermark," Proceedings of the SPIE International Conference on Electronic Imaging, San Jose, the SPIE International Conference on Electronic Imaging, San Jose, CA, Feb.,1996 CA, Feb.,1996

J. Meng and S. F. Chang, “Embedding visible watermarks in the coJ. Meng and S. F. Chang, “Embedding visible watermarks in the compressed domain,” Proc. of ICIP 98.mpressed domain,” Proc. of ICIP 98.

M. S. Kankanhalli, Rajmohan and J. R. Ramakrishnan, “Adaptive VM. S. Kankanhalli, Rajmohan and J. R. Ramakrishnan, “Adaptive Visible Watermarking of Images,” IEEE International Conference oisible Watermarking of Images,” IEEE International Conference on Multimedia Computing and Systems, 1999n Multimedia Computing and Systems, 1999

S. P. Mohanty, J. R. Ramakrishnan, and M. S. Kankanhalli, “A DCT S. P. Mohanty, J. R. Ramakrishnan, and M. S. Kankanhalli, “A DCT domain visible watermarking technique for images,” Proc. of ICMdomain visible watermarking technique for images,” Proc. of ICME 2000.E 2000.

S. P. Mohanty, J. R. Ramakrishnan, and M. S. Kankanhalli, “A Dual S. P. Mohanty, J. R. Ramakrishnan, and M. S. Kankanhalli, “A Dual

Watermarking Technique for Images, “ Proc. ACM, pp. 49-51, 1999Watermarking Technique for Images, “ Proc. ACM, pp. 49-51, 1999

The Scheme Proposed by G. BraudawaThe Scheme Proposed by G. Braudaway et al y et al

*)(667.38

)(' 3/2,,

,, LY

YYYY

w

mnw

A

mnmnmn

An approximately uniform color space is used, such as the CIE An approximately uniform color space is used, such as the CIE 1976 (L*u*v*) space and the CIE 1976(L*a*b*)-space, so amoun1976 (L*u*v*) space and the CIE 1976(L*a*b*)-space, so amounts of brightness increasing and decreasing are perceptually eqts of brightness increasing and decreasing are perceptually equal for a fixed change occurred everywhere in the color spaceual for a fixed change occurred everywhere in the color space

DefinitionsDefinitions YYn,mn,m and Y’ and Y’n,mn,m: the brightness values of each pixel in the unmark: the brightness values of each pixel in the unmark

ed original and the watermarked imageed original and the watermarked image YYww: the brightness of the “scene white” : the brightness of the “scene white”

Other Enhancing Other Enhancing SchemesSchemes

[Meng and Chang] [Meng and Chang] The same embedding model is extended to the DCT domain by simple statThe same embedding model is extended to the DCT domain by simple stat

istic model approximation for the convenience of processing directly in thistic model approximation for the convenience of processing directly in the MPEG-compressed domain.e MPEG-compressed domain.

[Kankanhalli et al] [Kankanhalli et al] Local features related to the degree of distortion tolerances, such as edge lLocal features related to the degree of distortion tolerances, such as edge l

ocations, texture distributions and luminance sensitivity, are taken into coocations, texture distributions and luminance sensitivity, are taken into consideration so that more unobtrusive watermarked images can be generatnsideration so that more unobtrusive watermarked images can be generated. ed.

Simple statistics of block-DCT coefficients are calculated and analyzed to Simple statistics of block-DCT coefficients are calculated and analyzed to decide the watermark embedding energy of each block. decide the watermark embedding energy of each block.

Edge integrity will be preserved, in these approaches, since the edge inEdge integrity will be preserved, in these approaches, since the edge information is essential to maintain the image quality. formation is essential to maintain the image quality.

And the energy of the embedded watermark is larger in highly textureAnd the energy of the embedded watermark is larger in highly textured areas than in smooth ones due to different noise sensitivity. d areas than in smooth ones due to different noise sensitivity.

In additions, the watermark energy of mid-gray regions is also smaller In additions, the watermark energy of mid-gray regions is also smaller than other areas since the noises are more visible against a mid-gray bthan other areas since the noises are more visible against a mid-gray background ackground

[S. P. Mohanty et al][S. P. Mohanty et al] in addition to the visibly embedded watermark, a fragile invisible watermin addition to the visibly embedded watermark, a fragile invisible waterm

ark is also adopted to check if the visible watermark is altered or not ark is also adopted to check if the visible watermark is altered or not

Important observations Important observations (1/4)(1/4)

Attacking visible watermarking Attacking visible watermarking scheme means successfully recover scheme means successfully recover the watermarked area.the watermarked area.

Implication:Implication: Similar image processing techniques Similar image processing techniques

can be adoptedcan be adopted Image recoveryImage recovery Object removalObject removal

Important Important observations(2/4)observations(2/4)

To clearly recognize the copyright To clearly recognize the copyright patterns, the patterns, the contours of contours of embedded patternsembedded patterns must be must be preserved. preserved.

Implication:Implication: An attacking scheme is effective if An attacking scheme is effective if

1.1.The pattern is completely removed The pattern is completely removed

2.2.The shape is seriously distorted without The shape is seriously distorted without seriously degrading visual quality.seriously degrading visual quality.

Important Important observations(3/4)observations(3/4)

The perceptibility of the host image The perceptibility of the host image details within watermarked area details within watermarked area depends on the depends on the preservation of preservation of edge informationedge information..

Implication:Implication: Available information while attackingAvailable information while attacking

Surrounding pixels around watermarked Surrounding pixels around watermarked area. area.

Edge information within watermarked area Edge information within watermarked area is available while attacking. is available while attacking.

Important Important observations(4/4)observations(4/4)

The robustness lies in the The robustness lies in the inevitability of exhaustive and inevitability of exhaustive and expensive laborsexpensive labors..

Implication:Implication: Only minimum user intervention should Only minimum user intervention should

be adopted during attacking be adopted during attacking User selection of watermarked areas User selection of watermarked areas

Averaging AttacksAveraging Attacks

Refill the watermarked areas by Refill the watermarked areas by averaging surrounding pixels.averaging surrounding pixels. Good approximations for small areas.Good approximations for small areas. Blurring effects across object boundariesBlurring effects across object boundaries

Image InpaintingImage Inpainting M. Bertalmio, V. Caselles, and C. Ballester, “M. Bertalmio, V. Caselles, and C. Ballester, “II

mage inpaintingmage inpainting,” ,” SIGGRAPH 2000SIGGRAPH 2000, Aug. 2000, Aug. 2000

Image inpainting Image inpainting is an is an iterativeiterative image recovery technique. image recovery technique. prolongs the approaching isophotes into damaged aprolongs the approaching isophotes into damaged a

reas.reas. successfully reconstruct the edges of damaged area.successfully reconstruct the edges of damaged area.

),(),,(

),,(),(),( jiI

njiN

njiNjiLjiI nnn

t

Basic Inpainting AttacksBasic Inpainting Attacks

Attacks against visible watermarking are regarded as Attacks against visible watermarking are regarded as common image recovery problems.common image recovery problems.

Good results can be obtained for areas composed of Good results can be obtained for areas composed of thinthin copyright patterns, but areas composed of copyright patterns, but areas composed of thickthick patterns patterns cannot be successfully recovered. cannot be successfully recovered.

General AttacksGeneral Attacks

Classifying flat areas within watermarked area Classifying flat areas within watermarked area by analyzing remaining edge information of host by analyzing remaining edge information of host imagesimages

Directly extend colors of surrounding flat areas Directly extend colors of surrounding flat areas into watermarked areasinto watermarked areas

Edge Area

Watermark Area

Further ImprovementFurther Improvement

Experimental Results (I)Experimental Results (I)

Experimental Results (II)Experimental Results (II)

Experimental Results Experimental Results (III)(III)