Introduction to Visible Watermarking

8/8/2019 Introduction to Visible Watermarking

1/22

Introduction toIntroduction to

Visible WatermarkingVisible WatermarkingIPR Course: TA LectureIPR Course: TA Lecture

2002/12/182002/12/18

NTU CSIE R105NTU CSIE R105


2/22

OutlineOutline

nn IntroductionIntroduction

nn StateState--ofof--thethe--ArtArt

nn Characteristics of Visible WatermarkingCharacteristics of Visible Watermarking

SchemesSchemes

nn Attacking Visible Watermarking SchemesAttacking Visible Watermarking Schemes

nn Discussions and ConclusionsDiscussions and Conclusions


3/22

Classifying Watermarking SchemesClassifying Watermarking Schemes

Data hiding

Fragile

watermarking

Watermarking

Imperceptible

watermarking

Visible

watermarking

Robust

watermarking

Visible

data embedding

Steganography

Non-robust

data embedding

Robust

data embedding

Imperceptible

data embedding


4/22

Visible WatermarkingVisible Watermarking

+

IPR protection schemes for images and video that haveto be released for certain purposes

Unobtrusive copyright patterns can be recognized on

embedded contents


5/22

Invisible WatermarkingInvisible Watermarkingv.sv.s..


Invisible Watermarking Visible Watermarking

Watermark

Perceptibility

Protection

Robustness

Imperceptible distortion Visibly Meaningful pattern

Intentional attacks andcommon signal processing

User-intervention basedwatermark removal

Passive Active

Current

Research StatusHot Only few papers

Extraction

Explicit extraction

moduleDirect viewing


6/22

Requirements ofRequirements of


nn Perceptibility of host image detailsPerceptibility of host image details

nn Contents should not be rendered useless after being visiblyContents should not be rendered useless after being visibly

watermarkedwatermarked

nn Visibility of watermark patterns in embedded contentsVisibility of watermark patterns in embedded contents

nn No explicit watermark extraction techniques are requiredNo explicit watermark extraction techniques are required

nn RobustnessRobustness

nn Difficult to remove unless exhaustive and expensive humanDifficult to remove unless exhaustive and expensive human

interventions are involvedinterventions are involved


7/22

A General Model ofA General Model of


nn II: the watermarked content: the watermarked content

nn I: the unI: the un--watermarked original contentwatermarked original content

nn W: the watermark patternW: the watermark pattern

nn KKii: the weighting factor: the weighting factor

nn D: a distance function measuring the perceptual difference of itD: a distance function measuring the perceptual difference of it s two parameterss two parameters

nn EEii: image feature extraction operators: image feature extraction operatorsnn ThresholdThresholdII: the largest allowable distortion of image details that observe: the largest allowable distortion of image details that observers can toleraters can tolerate

and, at the same time, the signature of can be maintained.and, at the same time, the signature of can be maintained.

nn ThresholdThresholdWW: the largest allowable distortion of the embedded watermark pat: the largest allowable distortion of the embedded watermark pat tern thattern thatthe copyright information can be clearly recognized.the copyright information can be clearly recognized.

WKIKI **' 21 +=

III ThresholdIEIED


8/22

StateState--ofof--thethe--ArtArt

nn G.G. BraudawayBraudaway, K.A., K.A. MagerleinMagerlein, and F., and F. MintzerMintzer, "Protecting Publicly Available, "Protecting Publicly Available

Images with a Visible Image Watermark," Proceedings of the SPIEImages with a Visible Image Watermark," Proceedings of the SPIE

International Conference on Electronic Imaging, San Jose, CA, FeInternational Conference on Electronic Imaging, San Jose, CA, Feb.,1996b.,1996

nn J.J. MengMengand S. F. Chang,and S. F. Chang, Embedding visible watermarks in the compressedEmbedding visible watermarks in the compressed

domain,domain, Proc. of ICIP 98.Proc. of ICIP 98.

nn M. S.M. S. KankanhalliKankanhalli,, RajmohanRajmohan and J. R.and J. R. RamakrishnanRamakrishnan,, Adaptive VisibleAdaptive VisibleWatermarking of Images,Watermarking of Images, IEEE International Conference on MultimediaIEEE International Conference on Multimedia

Computing and Systems, 1999Computing and Systems, 1999

nn S. P.S. P. MohantyMohanty, J. R., J. R. RamakrishnanRamakrishnan, and M. S., and M. S. KankanhalliKankanhalli,, A DCT domainA DCT domain

visible watermarking technique for images,visible watermarking technique for images, Proc. of ICME 2000.Proc. of ICME 2000.

nn S. P.S. P. MohantyMohanty, J. R., J. R. RamakrishnanRamakrishnan, and M. S., and M. S. KankanhalliKankanhalli,, A DualA Dual

Watermarking Technique for Images,Watermarking Technique for Images, Proc. ACM, pp. 49Proc. ACM, pp. 49--51, 199951, 1999


9/22

The Scheme Proposed by G.The Scheme Proposed by G. BraudawayBraudaway et alet al

*)(667.38

)(' 3/2,,,, L

Y

YYYY

w

mnw

A

mn

mnmn D-

-+=

t

t

mm

mm

nn An approximately uniform color space is used, such as the CIE 19An approximately uniform color space is used, such as the CIE 1976 (L*u*v*)76 (L*u*v*)

space and the CIE 1976(L*a*b*)space and the CIE 1976(L*a*b*)--space, so amounts of brightness increasingspace, so amounts of brightness increasing

and decreasing are perceptually equal for a fixed change occurrand decreasing are perceptually equal for a fixed change occurred everywhereed everywherein the color spacein the color space

nn DefinitionsDefinitions

nn YYn,mn,m andandYYn,mn,m: the brightness values of each pixel in the unmarked original a: the brightness values of each pixel in the unmarked original andnd

the watermarked imagethe watermarked image

nn YYww: the brightness of the: the brightness of the scene whitescene white


10/22

Other Enhancing SchemesOther Enhancing Schemes

nn [[MengMengand Chang]and Chang]

nn The same embedding model is extended to the DCT domain by simpleThe same embedding model is extended to the DCT domain by simple statisticstatisticmodel approximation for the convenience of processing directly imodel approximation for the convenience of processing directly in the MPEGn the MPEG--compressed domain.compressed domain.

nn [[KankanhalliKankanhalli et al]et al]

nn Local features related to the degree of distortion tolerances, sLocal features related to the degree of distortion tolerances, such as edge locations,uch as edge locations,texture distributions and luminance sensitivity, are taken intotexture distributions and luminance sensitivity, are taken into consideration so thatconsideration so that

more unobtrusive watermarked images can be generated.more unobtrusive watermarked images can be generated.nn Simple statistics of blockSimple statistics of block--DCT coefficients are calculated and analyzed to decideDCT coefficients are calculated and analyzed to decide

the watermark embedding energy of each block.the watermark embedding energy of each block.

nn Edge integrity will be preserved, in these approaches, since theEdge integrity will be preserved, in these approaches, since the edgeedgeinformation is essential to maintain the image quality.information is essential to maintain the image quality.

nn And the energy of the embedded watermark is larger in highly texAnd the energy of the embedded watermark is larger in highly tex tured areastured areasthan in smooth ones due to different noise sensitivity.than in smooth ones due to different noise sensitivity.

nn

In additions, the watermark energy of midIn additions, the watermark energy of mid--gray regions is also smaller thangray regions is also smaller thanother areas since the noises are more visible against a midother areas since the noises are more visible against a mid--gray backgroundgray background

nn [S. P.[S. P. MohantyMohantyet al]et al]

nn in addition to the visibly embedded watermark, a fragile invisibin addition to the visibly embedded watermark, a fragile invisible watermark is alsole watermark is alsoadopted to check if the visible watermark is altered or notadopted to check if the visible watermark is altered or not


11/22

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

nn Attacking visible watermarking scheme meansAttacking visible watermarking scheme means

successfully recover the watermarked area.successfully recover the watermarked area.

nn Implication:Implication:

nn Similar image processing techniques can be adoptedSimilar image processing techniques can be adoptednn Image recoveryImage recovery

nn Object removalObject removal


12/22

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

nn To clearly recognize the copyright patterns, theTo clearly recognize the copyright patterns, the

contours of embedded patternscontours of embedded patterns must bemust be

preserved.preserved.

nn Implication:Implication:nnAn attacking scheme is effective ifAn attacking scheme is effective if

1.1.The pattern is completely removedThe pattern is completely removed

2.2.The shape is seriously distorted without seriouslyThe shape is seriously distorted without seriously

degrading visual quality.degrading visual quality.


13/22


nn The perceptibility of the host image detailsThe perceptibility of the host image detailswithin watermarked area depends on thewithin watermarked area depends on thepreservation of edge informationpreservation of edge information..

nn

Implication:Implication:nnAvailable information while attackingAvailable information while attacking

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

nn Edge information within watermarked area is availableEdge information within watermarked area is availablewhile attacking.while attacking.


14/22


nn The robustness lies in theThe robustness lies in the inevitability ofinevitability of

exhaustive and expensive laborsexhaustive and expensive labors..

nn Implication:Implication:

nn Only minimum user intervention should be adoptedOnly minimum user intervention should be adoptedduring attackingduring attacking

nn User selection of watermarked areasUser selection of watermarked areas


15/22


16/22

ImageImage InpaintingInpainting

nn M.M. BertalmioBertalmio, V., V. CasellesCaselles, and C., and C. BallesterBallester,, ImageImageinpaintinginpainting,, SIGGRAPH 2000SIGGRAPH 2000, Aug. 2000, Aug. 2000

nn ImageImage inpaintinginpaintingnn is anis an iterativeiterative image recovery technique.image recovery technique.

nn prolongs the approachingprolongs the approachingisophotesisophotes into damaged areas.into damaged areas.nn successfully reconstruct the edges of damaged area.successfully reconstruct the edges of damaged area.

),(),,(

),,(

),(),( jiInjiN

njiN

jiLjiI

nnn

t

= d


17/22

BasicBasic InpaintingInpaintingAttacksAttacks

nn Attacks against visible watermarking are regarded as common imagAttacks against visible watermarking are regarded as common image recoverye recoveryproblems.problems.

nn Good results can be obtained for areas composed ofGood results can be obtained for areas composed ofthinthin copyright patterns,copyright patterns,but areas composed ofbut areas composed ofthickthickpatterns cannot be successfully recovered.patterns cannot be successfully recovered.


18/22

General AttacksGeneral Attacks

nn Classifying flat areas within watermarked area byClassifying flat areas within watermarked area by

analyzing remaining edge information of host imagesanalyzing remaining edge information of host images

nn Directly extend colors of surrounding flat areas intoDirectly extend colors of surrounding flat areas intowatermarked areaswatermarked areas

Edge Area

Watermark Area


19/22

Further ImprovementFurther Improvement


20/22

Experimental Results (I)Experimental Results (I)


21/22

Experimental Results (II)Experimental Results (II)


22/22

Experimental Results (III)Experimental Results (III)

Introduction to Visible Watermarking

Documents