Research Article Video Multiple Watermarking Technique ...downloads.hindawi.com/journals/tswj/2014/634828.pdfa robust video multiple watermarking technique is proposed to solve this

Research ArticleVideo Multiple Watermarking Technique Based onImage Interlacing Using DWT

Mohamed M Ibrahim1 Neamat S Abdel Kader2 and M Zorkany1

1National Telecommunications Institute Cairo 11678 Egypt2Faculty of Engineering Cairo University Cairo 12316 Egypt

Correspondence should be addressed to M Zorkany m zorkanyntiscieg

Received 1 September 2014 Revised 19 November 2014 Accepted 21 November 2014 Published 21 December 2014

Academic Editor Miaoqing Huang

Copyright copy 2014 Mohamed M Ibrahim et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

Digital watermarking is one of the important techniques to secure digital media files in the domains of data authentication andcopyright protection In the nonblind watermarking systems the need of the original host file in the watermark recovery operationmakes an overhead over the system resources doubles memory capacity and doubles communications bandwidth In this papera robust video multiple watermarking technique is proposed to solve this problem This technique is based on image interlacingIn this technique three-level discrete wavelet transform (DWT) is used as a watermark embeddingextracting domain Arnoldtransform is used as a watermark encryptiondecryptionmethod and different types of media (gray image color image and video)are used as watermarksThe robustness of this technique is tested by applying different types of attacks such as geometric noisingformat-compression and image-processing attacks The simulation results show the effectiveness and good performance of theproposed technique in saving system resources memory capacity and communications bandwidth

1 Introduction

Due to the rapid growth of computer and network tech-nologies digital multimedia are becoming more popular andit is very easy to transmit and distribute data So there isa demand for techniques to protect the data and preventunauthorized duplication Digital watermarking protects theillegal copying ofmultimedia Awatermark is secret informa-tion about origin ownership copy control and so forthThisinformation is embedded in multimedia content taking careof robustness and imperceptiblyThewatermark is embeddedand extracted as per requirement to represent the ownershipandor the identity of multimedia [1]

Digital media files may be text images audios or videosSo digital watermarking is classified into four types textwatermarking image watermarking audio watermarkingand video watermarking The watermark may be in a noiseform such as Pseudo-random Gaussian noise or in a dataform such as text image (logo) and video Also digital water-marking is classified according to the number of watermarksper the host media file into two types single and multiplewatermarking

Digital watermarking systems are also categorizedaccording to the watermark extracting process into twotypes nonblind and blind watermarking [2] In the nonblindwatermarking systems as shown in Figure 1 the originalmedia is required in addition to the watermarked mediain the watermark extracting process while in blind water-marking the original media is not required in the watermarkextracting process [3] From this definition the twoadvantages in nonblind watermarking techniques over theblind ones are as follows first being lower in computationalcomplexity and second more robustness against attacks orsignal distortions because the extracted watermark is moresimilar to the original one But also this definition presentsthe main problem of the nonblind watermarking systemsdouble overhead over system resources like memorystorage in both sender and receiver and bandwidth on thecommunications channel between them

Digital video can be defined simply as a collectionof sequential images [3] So video watermarking can beconsidered as an expansion of image watermarking Videowatermarking system has more capacity than image water-marking system so more watermarks that included images

Hindawi Publishing Corporatione Scientific World JournalVolume 2014 Article ID 634828 12 pageshttpdxdoiorg1011552014634828

2 The Scientific World Journal

Original data

Watermark

Secret key

Watermarkembedding

Watermarkeddata

(a) At the sender

Original data

Secret key

WatermarkwatermarkRecoveredextractingWatermarked

data

(b) At the receiver

Figure 1 Classical nonblind watermarking system

and videos can be embedded in the video watermarkingsystem (as proposed in this paper) The previous problemsof the nonblind watermarking have a more effect on videowatermarking systems So a solution for this problem is amore important issue in video watermarking

A proposed image watermarking technique based onimage interlacing has been used to solve thememory capacityand communications bandwidth problems of the nonblindwatermarking systems [4] the target system was imagewatermarking system with a single watermark (color imageas host and gray image as watermark) In this paper videowatermarking and multiple watermarks in different types(gray image color image and video) have been introduced

This paper is organized as follows A brief review ofvideo watermarking techniques is presented in Section 2The based techniques which the proposed method dependson are illustrated in Section 3 The proposed video multiplewatermarking technique is presented in Section 4 Simulationresults and discussions are given in Section 5 and finally theconclusion and the references are included

2 Video Watermarking Techniques

Digital video is a sequence of still images called frames andthese frames are loaded by a constant rate called frame rateSo by reading the frames from the video file frame-by-frameand dealing with each frame as a color image all imagewatermarking techniques can be used as video watermarkingtechniques [5]

As in image watermarking the watermark embeddingand extraction operations can be done in spatial domainor frequency domain In the spatial domain the watermarkis embedded by modifying the pixel values of the hostimagevideo directly In the frequency or transform domainthe watermark is embedded by modifying the frequencycomponents of the host imagevideo [5] this is done bytransforming the host signal to the frequency domain beforeembedding the watermark and retransforming to spatialdomain after that As a result spatial domain watermarkingtechniques are lower in complexity than frequency domaintechniques The characteristics of the human visual system(HVS) are better captured by the frequency coefficients thanspatial coefficients [1] thus frequency domain watermarkingtechniques enjoy better imperceptibility more robustness

against attacks such as noise addition pixel removal rescal-ing rotation and shearing plus more compatibility withcompression standards such as MPEG 1 2 and 4 (MovingPicture Experts Group)

Unlike imagewatermarking there is a third type of water-marking techniques called compressed-domain video water-marking techniques [6] Many digital videos are typicallystored and distributed in compressed form (eg MPEG)Due to the real-time requirements of video broadcastingthere is no time for decompression and recompression so thewatermark in these techniques is embedded directly in thecompressed videoThemain disadvantage of these techniquesis that the watermarked video can be highly susceptible tobe recompressed with different parameters or converted toformats other than MPEG

3 The Based Techniques

The proposed technique in this paper depends on imageinterlacing When deploying this technique to the videomultiple watermarking system three-level discrete wavelettransform (DWT) is used as an example of the frequencydomain watermarking For more security Arnold Transformas one of the popular imagevideo frame encryptionmethodsis used to encrypt the watermarks before embedding them inthe host video The following subsections will discuss thesetechniques before starting with the proposed method

31 Three-Level DWT Discrete fourier transform (DFT)discrete cosine transform (DCT) and discrete wavelet trans-form (DWT) are the most popular reversible transformsused in frequency domain watermarking But DWT as amultiresolution multilevel transform is much preferred inwatermarking thanDCTandDFT because it understands thehuman visual system (HVS) closer than them [7]

At the first level of DWT decomposition the imagevideoframe is filtered horizontally and vertically by low (L) andhigh (H) pass filters to produce four frequency subbands [5]LL1 as the low frequency subband and HL1 LH1 and HH1 asthe high frequency subbandsThe low frequency componentscontain the most significant portions of the imagevideoframe in which any modifications like watermarking candamage it [8] as shown in Figure 2(a) [9]The high frequencycomponents contain the least significant portions of the

The Scientific World Journal 3

LL1 HL1

LH1 HH1

(a) Single-level decomposition

LL3 HL3LH3 HH3

HL2

LH2 HH2

HL1

LH1 HH1

(b) Three-level decomposition

Figure 2 One- and three-level DWT decomposition of Lena image

imagevideo frame which can be eliminated by compressionSo the mid frequency components are the best locations forwatermarking These mid frequency components appearedstrongly at the next levels of DWT decomposition in whichthe LL (n) subband is decomposed into LL (119899+1) LH (119899+1)HL (119899+ 1) and HH (119899+ 1) where ldquonrdquo is the level number LL(119899 + 1) is the lower frequency subband and LH (119899 + 1) HL(119899 + 1) and HH (119899 + 1) are the mid frequency subbands [8]

Three-level DWT as shown in Figure 2(b) [10] is usedin the proposed method This number of decompositionlevels is enough to present the mid frequency componentsstrongly such as the HL3 subband which is considered as thebest location for watermarking operations according to theHVS properties [8] So in our case more than three levelsof decomposition is unnecessary as well as it will be morecomputational complexity

32 Arnold Transform In order to enhance the securityand robustness of the watermarking system the watermarkimagevideo frame is encrypted before embedding ArnoldTransform is one of the popular imagevideo frame encryp-tion methods [4 11] In this transform which is a periodictransform the original organization of the imagevideo framepixels is randomized and after a number of iterations calledArnoldrsquos period the imagevideo frame is returned to itsoriginal state Arnold Transform uses the following formulafor changing the location of each pixel in the imagevideoframe of size119873 times119873

(

1199091015840

1199101015840) = (

1 1

1 2)(

119909

119910) (mod 119873) (1)

where (119909 119910) is the current location and (1199091015840 1199101015840) is the newlocation

To use the Arnold Transform as an encryption methodof type symmetric a number of iterations which must beless than the Arnold period can be used as a symmetricencryption key [12] Figure 3 shows the encryption of ldquoLenardquoimage using Arnold Transform

33 Image Interlacing Image interlacing (also known asinterleaving) is an operation in which any image can be

divided into subimages Deinterlacing is the reverse opera-tion in which the subimages are combined together to gen-erate the original image The interlacing algorithm indicatesthe contents of each subimage and in the proposed methodthe interlacing algorithm is depending on dividing the imagerows into even andodd rows and the image columns into evenand odd columns [4]

In the proposed method there are two levels of inter-lacing one-level interlacing and two-level interlacing One-level Interlacing has two types interlacing by rows only to geteven rows (ER) and odd rows (OR) subimages as shown inFigure 8 and interlacing by columns only to get even columns(EC) and odd columns (OC) subimages as shown in Figure 9

Two-level interlacing is as follows interlacing by rowsfirst and then by columns (or in the opposite order) to getfour subimages even rows even columns (EE) even rows oddcolumns (EO) odd rows even columns (OE) and odd rowsodd columns (OO) as shown in Figure 10

4 Proposed Technique

In the nonblind video watermarking system as shown inFigure 1 two identical copies of the original video are usedAt the sender one of them is watermarked and then is sent tothe receiver The other copy of the original video is sent as itis At the receiver the watermark is recovered using both thewatermarked video and the other copy of the original video

The paper goal is how to prevent sending this originalvideo to the receiver to save the system resources likememory storage and communications bandwidth The ideaof our proposed solution for this problem is as follows ifthere is a technique by which the original video is dividedinto parts or subvideos from these subvideos we can get twoof them that are identical (or at least are very similar to eachother)These two subvideos can play the same role of the twoidentical copies of the original video in such case as a resultthere is no need to another copy of the original video in thewatermark extracting operation at the receiver which is thegoal of this paper

This technique is the image interlacing for image water-marking and the same technique can be used in videosbut in two steps first the video frames are divided intosubframes using image interlacing and then these subframesare collected together to generate the subvideos


(a) Original image (b) Encrypted image

(c) Restored image

Figure 3 Arnold Transform as imagevideo frame encryption method

The proposed watermark embedding and extracting pro-cesses are passing through the following steps

41 At the Sender (The Watermark Embedding Process) Theproposed watermarking technique main structure at thesender is given in Figure 4

First Step The original video is interlaced into subvideos

Second StepThemost similar two subvideos (ie subvideosAand B as shown in Figures 4 and 5) are selected by calculatingthe similarity factor or normalized correlation (NC) betweenthe two subvideos This calculation of NC is performed firstbetween each of the corresponding two color bands in eachcorresponding two subframes using the following formula[13]

NC = 1119908 times ℎ

119908

sum

119894=1

ℎ

sum

119895=1

(119883119894sdot119895oplus 1198831015840

119894sdot119895) (2)

where 119883119894sdot119895

and 1198831015840119894sdot119895

are the corresponding pixel values intwo subframes and (119908 times ℎ) is the subframe size Then we

calculated the average NC value for all color bands and forall subvideo frames The two subvideos that give the biggestaverage NC value are the most similar two subvideos thatresulted from the interlacing operation on the original hostvideo The names or numbers of these two subvideos will bethe first part of the secret key between the sender and thereceiver

In the proposed method the NC is used in two positionsfirst in the second step to get the most similar two subvideosand second after the watermark extraction process to evalu-ate it by comparing the extracted watermark with the originalone

Third StepMultiplewatermarks in different types (gray imageandor color image andor video) are embedded in the firstone of the selected subvideos (Subvideo A) Each of them isembedded in separate frames whichmeans that the proposedmethod is not affected if one of them is not available andthis is one of the strength points of our proposed methodThe watermarked frame numbers corresponding to eachwatermark will be the second part of the secret key betweenthe sender and the receiver


Originalvideo

Interlacing

Subvideo A

Frame A

Frame B

X frames

Subvideo B(the most

similar to A)

Othersubvideos

Gray imagewatermark

Color imagewatermark

Videowatermark(X frames)

R G B R G B

ArnoldTransform

ArnoldTransform

ArnoldTransform

R G B R G BN

C

R

R

G

G

B

B

Three-level

Three-level

Three-level

DWT

DWT

DWT

IDWT

IDWT

IDWT

Watermarkedvideo

Deinterlacing

Watermarkedsubvideo A

Watermarkedframe A

Watermarkedframe B

WatermarkedX frames

1205721 1205722 1205723

Figure 4 The watermark embedding process in the proposed video multiple watermarking system where RGB stands for red green andblue color bands N is the number of Arnold Transform iterations 120572 is the scaling factor C is the selected color band A and B stand for themost similar subvideos and DWT stands for discrete wavelet transform

Watermarkedvideo

Othersubvideos


Watermarkedframe A

Watermarkedframe B

Watermarked

N

C

RGB

RGB

C

RGB

RGB

Gray imagewatermark



R G B R G B

InverseArnold

InverseArnold

InverseArnold

11205721 11205722 11205723

Three-level Three-level

Three-level

Three-level

Three-level

Three-level

DWT DWT

DWT

DWT

DWT

DWT

Frame A

Frame B

X frames

Subvideo B

Interlacing

X frames

Figure 5 The watermark extracting process in the proposed video multiple-watermarking system


Before the watermark embedding process all watermarksare encrypted using Arnold Transform by a specific numberof iterations (N as shown in Figures 4 and 5) and this numberwill be the third part of the secret key between the sender andthe receiver

The watermark embedding process is performed aftertransforming the selected frames into frequency domainusing three-level DWT and the following equation [4 14]

1198681015840= 119868 + 120572119882 (3)

where 119868 is the HL3 subband of original subvideo frame 1198681015840is the watermarked one119882 is the watermark signal and 120572 isthe scaling factor This scaling factor is a constant value forall subvideo frames and the selection of this value is based onrules as will be presented later Each watermark has its ownscaling factor as shown in Figures 4 and 5 (1205721 1205722 and 1205723 forthe three used watermarks) 120572 for videomade for every framehas been chosen and then the average value to represent the 120572value was calculated These scaling factors will be the fourthand last part of the secret key between the sender and thereceiver

Final Step The watermarked video is generated by deinter-lacing the watermarked subvideo and other subvideos Thiswatermarked video and the secret key (of four parts) are sentto the receiver and no copy of the original video is sent (thepaper goal)

42 At the Receiver (The Watermark Extracting Process)The proposed watermarking technique main structure at thereceiver is given in Figure 5

First the watermarked video is interlaced by the samelevel and type as the original video in the sender SecondThe watermarked subvideo A and the subvideo B wereselected by using the first part of the secret key Finally thewatermark extraction process is performed using the twoselected subvideos as follows first select the frames fromboth selected subvideos using the third part of the secret keySecond transform these frames into frequency domain usingthree-level DWT Third extract the watermarks using thewatermark extraction equation which is expressed as [4 14]

1198821015840=

(11986810158401015840minus 119868lowast)

120572

(4)

where 11986810158401015840 is the HL3 subband of the watermarked subvideo Aframe 119868lowast is the corresponding HL3 subband of the subvideoB frame1198821015840 is the extracted encrypted watermark and 120572 isthe scaling factor from the fourth part of the secret keyThenthe extracted encryptedwatermark is decrypted using inverseArnold Transform by number of iterations from the secondpart of the secret key

Let us talk in detail about the watermark embedding andextracting equations (3) and (4) For the grayscale imagewatermark only one color band from one video frame isneeded for watermark embedding and extracting operationsThis color band (C as shown in Figures 4 and 5) is thecolor band that gives the best NC values when selecting

Table 1 Video interlacing operation

Interlacinglevel Subvideos NC

Red Green Blue

One ER-OR 09548 09645 09657EC-OC 09713 09760 09755

Two

EE-OE 09384 09516 09516EE-EO 09713 09760 09755EE-OO 09245 09386 09376OE-EO 09226 09371 09365OE-OO 09714 09760 09756EO-OO 09384 09513 09516

Subvideos ER = even rows OR = odd rows EC = even columns OC = oddcolumns EE = even rows even columns EO = even rows odd columns OE= odd rows even columns and OO = odd rows odd columns

subvideos A and B (the second step) For the color imagewatermark all color bands from one video frame is neededand each color band from this watermark is embedded intothe corresponding color band from this video frame [15] Forvideo watermark all color bands from the same number offrames in the video watermark are required which meansthat the number of frames in the video watermark must beless than or equal to the number of frames in host video

5 Simulation Results

The proposed method is tested using multiple testing videosas host videos and multiple testing images and videos aswatermarks Figure 6 presents one example of these hostvideos and examples of each watermark type as followsldquoAirhorseavirdquo as a host video of 50 frames with frame size512 times 704 and frame rate 30 framessecond ldquoEvil Insidejpgrdquoas a grayscale image watermark of size 32 times 32 ldquoLenajpgrdquo asa color image watermark of size 32 times 32 and ldquoCompositeavirdquoas a video watermark of 22 frames with frame size 32 times 32 andframe rate 15 framessecond

There are many operations in the proposed method eachoperation has its own results as follows

51 Interlacing Operation As in the proposed method thereare two levels of interlacing Figures 7 and 8 present theresults of the two types of one-level interlacing and Figure 9presents that of the two-level interlacing In these figures thefirst frames from the original video and from each resultedsubvideo are presented

52 Selection of Subvideos A and B and the Color Band CAfter the interlacing operation we calculate the normalizedcorrelation (NC see (2)) between the two resulted subvideosto get the most similar two subvideos (subvideos A and Bas shown in Figures 4 and 5) in which the watermarkingoperations will be performed later Also the color band thatgives the best NC values will be the color band (C) fromthe specified frame (a) which is used for embedding andextracting the grayscale image watermark

Table 1 shows the NC values between the two resultedsubvideos in each interlacing level From these results we


(a) Original video Frame 1 (b) Grayscale imagewatermark

(c) Color image water-mark

(d) Video watermark Frame 1

Figure 6 Host video and watermarks

(a) Original video (b) Even rows subvideo

(c) Odd rows subvideo

Figure 7 One-level interlacing by rows

note that all NC values are close to one which means thatthe used interlacing algorithmgives subvideos that are similarto each other But if we compare between the two subvideosto get the most similar pair they will be EC and OC in one-level interlacing and OE and OO in two-level interlacing Ifwe compare between the color bands to get the one whichhave the best results it will be the green

53 Watermarking Operations To show the results of ourproposal host video and watermarks in Figure 6 are illus-trated as examples The video watermark is embedded in thefirst 22 frames of the host video (119883 = 22) the grayscaleimage watermark is embedded in the green color band (fromTable 1) of frame number 30 (a = 30) and the color imagewatermark is embedded in frame number 40 (b = 40) Allof these watermarks are encrypted before embedding bynumber of iterations less than theArnold Period as illustratedin Section 32 For 32 times 32 watermark image or video framesthe Arnold Period is equal to 48 [11] so we can encrypt themby a number of iterations such as119873 = 20

Two main tools are used for evaluating the watermarkingsystems NC (normalized correlation) for evaluating thewatermark extracting process by comparing the extractedwatermark with the original one and PSNR (peak signal tonoise ratio) [13] for evaluating the watermark embeddingprocess by comparing the watermarked frame with theoriginal one For bothNC and PSNRwe calculate the averagevalue for all frames In Table 1 NC values are the averagevalues for all frames In Table 2 both NC and PSNR valuesare the average values for all frames In Table 3 both NC andPSNR values are for individualsrsquo frames

PSNR is defined in terms of mean square error (MSE) asfollows

PSNR = 10 times log10

2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)


(a) Original video (b) Even columnssubvideo

(c) Odd columns subvideo

Figure 8 One-level interlacing by columns

(a) Original video (b) Even rows even columns (EE) sub-video

(c) Even rows odd columns (EO) sub-video

(d) Odd rows Even columns (OE) sub-video

(e) Odd rows odd columns (OO) sub-video

Figure 9 Two-level interlacing

where119867119894sdot119895and1198671015840

119894sdot119895are the corresponding pixel values in the

original andwatermarked frames respectively and the size ofeach frame is 119903 times 119888

Tables 2 and 3 show the effectiveness and good perfor-mance of the proposed multiple-video watermarking tech-nique

54 Selection Scaling Factor Value 120572 The scaling factor 120572is the value that gives the best results in the evaluation ofthe watermarking system There are two main parametersfor evaluating such systems PSNR and NC As shown inFigure 10 there is a linear relationship between PSNR andscaling factor the PSNR is reverse proportional to the scalingfactor the peak value of PSNR is when the scaling factor isequal to zero (No watermarking) so PSNR cannot be usedas a reference for the best scaling factor value alone But forNC there is a nonlinear orGaussian relationship betweenNC

PSNR NC

120572 Scaling factor

Figure 10 Scaling factor versus PSNR and NC

and scaling factor with a peak value so the scaling factor 120572can be selected as the scaling factor value that gives the bestNC value when evaluating the watermark extraction processIn this proposed paper choice of 120572 for video was made for


Table 2 Multiple Watermarks (Gray Image Color Image and Video)

Grayscale image watermark Color image watermark Video watermarkInterlacing level No One Two No One Two One TwoSubvideos EC-OC OE-OO EC-OC OE-OO EC-OC OE-OOScaling factor (120572) 05 18 18 05 18 18 4 4PSNR 367755 352069 339840 367755 352069 339840 282955 282710NC

No attacks 09824 09578 09725 09824 09578 09725 09358 09486Cropping 18 intermediate 04251 06129 06413 04251 06129 06413 07682 07800Cropping 18 left corner 07250 07252 07303 07250 07252 07303 08029 08174Gaussian noise 0001 08981 09438 09606 08981 09438 09606 08751 08911Gaussian noise 001 06231 08794 08962 06231 08794 08962 08724 08868Gaussian noise 005 03181 07290 07462 03181 07290 07462 08579 08746Salt and pepper noise 0001 09603 09562 09706 09603 09562 09706 09343 09471Salt and pepper noise 001 08100 09382 09556 08100 09382 09556 09199 09342Salt and pepper noise 005 05351 08697 08783 05351 08697 08783 08630 08805Median filtering 08677 09507 09672 08677 09507 09672 09211 09341JPEG compression 70 08902 09338 09473 08902 09338 09473 09522 09612JPEG compression 50 08734 09174 09299 08734 09174 09299 09462 09423JPEG compression 30 08098 09070 09122 08098 09070 09122 09415 09448Brightening 05780 08869 08514 05780 08869 08514 08621 08477Darkening 06584 09101 08870 06584 09101 08870 08838 08762Sharpening 09002 08591 08708 09002 08591 08708 08903 08942

Table 3 Video Watermark (without interlacing) for all video frames

Watermark type Video frames numberFrame number 1 2 3 4 5 6 7 8 9 10 20 21 AveragePSNR 291022 297661 302200 302013 304375 304683 305608 305818 305984 sdot sdot sdot 287894 301212NC

No attacks 09575 09537 09542 09587 09544 09499 09486 09524 09536 sdot sdot sdot 09614 09543Cropping 18 intermediate 09428 09312 09365 09378 09327 09236 09277 09247 0925 sdot sdot sdot 09469 09314Cropping 18 left corner 09262 09111 0913 09294 09357 09152 09103 09152 09224 sdot sdot sdot 09278 09210Gaussian noise 0001 08613 08395 08383 08479 0826 0815 08075 08115 08085 sdot sdot sdot 08824 08304Gaussian noise 001 0869 08416 08328 08454 08267 08169 08071 08079 08062 sdot sdot sdot 08766 08309Gaussian noise 005 08499 08125 08146 08358 08147 08003 07797 07939 07851 sdot sdot sdot 08601 08115Salt and pepper noise 0001 09559 09499 09498 0956 09502 09455 09439 09465 09492 sdot sdot sdot 09594 09502Salt and pepper noise 001 09326 09206 09168 09274 09185 09081 09052 091 09071 sdot sdot sdot 09405 09175Salt and pepper noise 005 08515 08204 08108 08288 0803 0785 07781 07779 07751 sdot sdot sdot 08684 08062Median filtering 09173 09108 09015 0914 09011 08867 08771 08685 08818 sdot sdot sdot 0928 08976JPEG compression 70 0956 09523 09522 0957 09532 0948 09467 09504 09516 sdot sdot sdot 09603 09526JPEG compression 50 09511 0945 0947 09521 09475 09431 09397 09443 09456 sdot sdot sdot 09551 09467JPEG compression 30 09486 09408 09424 09487 09435 09383 09354 09384 09406 sdot sdot sdot 09531 09424Brightening 08696 08423 08404 0871 08489 08226 08081 07949 08031 sdot sdot sdot 0895 08344Darkening 08927 08721 0869 08954 08767 0855 08421 08341 0842 sdot sdot sdot 09138 08652Sharpening 09344 0931 09339 09376 09313 09223 09243 09246 09228 sdot sdot sdot 09397 09294

every frame and then the average value to represent 120572 valuewas calculated

55 Robustness against Attacks Evaluation During the com-munications channel between the sender and the receiver the

watermarked video is attacked these attacks are categorizedinto five categoriesThese categories are as follows geometricattacks noising attacks denoising (filtering) attacks format-compression attacks and image-processing attacks In orderto ensure that our proposed solution is not affected against


(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 11 Comparison between the classical nonblind video multiple watermarking system (grayscale image watermark) and the proposedmethod where (a) and (b) show original video frame and original watermark (c) and (d) show watermarked video frame and recoveredwatermark without interlacing (e) and (f) show watermarked video frame and recovered watermark with one-level interlacing and (g) and(h) show watermarked video frame and recovered watermark with two-level interlacing

the robustness of the video multiple watermarking systemsagainst attacks one or more examples from each category areapplied on these systems before and after using the proposedsolution as shown in the next subsection

56 Comparison between the Classical Video Multiple Water-marking System and the Proposed System A comparisonis made between a classical video multiple watermarkingsystem (without interlacing) and the proposed system (withinterlacing) Figures 11 12 and 13 present this comparisonwithout attacks one watermark in each figure and in thesefigures the comparison is illustrated by showing the water-marked video frame and the extracted watermark beforeand after applying the proposed method From these figuresit is obviously that there is no degradation in the qualityof the original video frame and the original watermarkafter watermarking using the nonblind watermarking systemwithout interlacing and the same system with interlacing

Also Table 2 presents the same comparison for eachwatermark with the attacks For image the watermarkingand attacking operations are performed on only one framebut in the video the watermarking and attacking operationsare performed on a number of frames equal to the numberof watermark frames (21 frames) and the values of NC and

PSNR are the average values for all color bands and for allwatermarked frames as shown in Table 3

From the results of Table 2 there are three notes asfollows

(i) First the results of the watermarking system withand without interlacing are close to each other whichindicates that the goal of this paper is achievedwithout any degradation in its imperceptibly or itsrobustness against attacks

(ii) Second no enhancement in simulation results intwo-level interlacing over one level interlacing whichindicates that there is no need for more levels ofinterlacing to save processing time and speed upwatermarking process Also in the view of codecomplexity the one-level interlacing is more preferredthan the two-level interlacing

(iii) Third about the attacks in general the results areclose to each other between the classical system with-out interlacing and the same system with interlacing

These notes are the same notes as [4] which indicatesthat our proposed solution (image interlacing) is workingin all types of nonblind watermarking systems (image singlewatermarking and video multiple watermarking)


(a) (b) (c) (d)

(e) (f) (g) (h)Figure 12 Comparison between the classical nonblind video multiple watermarking system (color image watermark) and the proposedmethod where (a) and (b) show original video frame and original watermark (c) and (d) show watermarked video frame and recoveredwatermark without interlacing (e) and (f) show watermarked video frame and recovered watermark with one-level interlacing and (g) and(h) show watermarked video frame and recovered watermark with two-Level interlacing

(a) (b) (c) (d)

(e) (f) (g) (h)Figure 13 Comparison between the classical non-blindVideoMultipleWatermarking System (VideoWatermark) and the ProposedMethodwhere (a) and (b) Original Video Frame and Original Watermark (c) and (d)Watermarked Video Frame and RecoveredWatermark withoutInterlacing (e) and (f) Watermarked Video Frame and Recovered Watermark with One Level Interlacing (g) and (h) Watermarked VideoFrame and Recovered Watermark with Two Level Interlacing


6 Conclusion

In this paper a robust video multiple watermarking tech-nique was proposed to solve nonblind watermarking sys-tem problems This technique was based on image inter-lacing technique In this technique three-level discretewavelet transform (DWT) was used as watermark embed-dingextracting domain and Arnold transform as watermarkencryptiondecryption method In this paper different typesof media as gray image color image and video were usedas watermarks The robustness of this technique was testedby applying different types of attacks Simulation resultsshowed the effectiveness of the proposed method with alltypes of nonblind watermarking systems there is no need tothe original host signal in the watermark extraction processand no overhead over system resources Also there is nodegradation in performance of these systems after applyingthis solution and no degradation in their robustness againstattacks

Also simulation results showed the effectiveness andgood performance of this proposed technique with savingsystem resources memory capacity and communicationsbandwidthThe percentage of savingmemory and bandwidthis 50 due to prevent sending the original video in theproposed video watermarking system

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

References

[1] P Singh and R S Chadha ldquoA survey of digital watermarkingtechniques applications and attacksrdquo International Journal ofEngineering and Innovative Technology vol 2 no 9 pp 165ndash1752013

[2] A Kumar Gupta and M S Raval ldquoA robust and securewatermarking scheme based on singular values replacementrdquoSadhana vol 37 no 4 pp 425ndash440 2012

[3] F Hartung and M Kutter ldquoMultimedia watermarking tech-niquesrdquo Proceedings of the IEEE vol 87 no 7 pp 1079ndash11071999

[4] M M Ibrahim N S Abdel Kader and M Zorkany ldquoA robustimage watermarking technique based on image interlacingrdquo inProceedings of the 31st National Radio Science Conference (NRSCrsquo14) pp 92ndash98 Cairo Egypt April 2014

[5] G V Mane and G G Chiddarwar ldquoReview paper on videowatermarking techniquesrdquo International Journal of Scientificand Research Publications vol 3 no 4 2013

[6] J K Su F Hartung and B Girod ldquoDigital watermarking of textimage and video documentsrdquo Computers amp Graphics vol 22no 6 pp 687ndash695 1998

[7] M Kaur S Jindal and S Behal ldquoA study of digital imagewatermarkingrdquo International Journal of Research in Engineeringamp Applied Sciences vol 2 no 2 pp 126ndash136 2012

[8] B L Gunjal and S NMali ldquoComparative performance analysisof DWT-SVD based color image watermarking technique inYUV RGB and YIQ color spacesrdquo International Journal ofComputer Theory and Engineering vol 3 no 6 2011

[9] M K Khorasani and M M Sheikholeslami ldquoAn DWT-SVDbased digital imagewatermarking using a novel wavelet analysisfunctionrdquo in Proceedings of the 4th International Conferenceon Computational Intelligence Communication Systems andNetworks (CICSyN rsquo12) pp 254ndash256 July 2012

[10] M-S Hsieh ldquoPerceptual copyright protection using multires-olution wavelet-based watermarking and fuzzy logicrdquo Interna-tional Journal of Artificial Intelligence amp Applications (IJAIA)vol 1 no 3 2010

[11] C Pradhan V Saxena and A K Bisoi ldquoNon blind digitalwatermarking technique using DCT and cross chaos maprdquo inProceedings of the International Conference on CommunicationsDevices and Intelligent Systems (CODIS rsquo12) pp 274ndash277 IEEEKolkata India December 2012

[12] N Chaturvedi ldquoVarious digital imagewatermarking techniquesand wavelet transformsrdquo International Journal of EmergingTechnology and Advanced Engineering vol 2 no 5 2012

[13] B Surekha and G N Swamy ldquoA semi-blind image watermark-ing based on discrete wavelet transform and secret sharingrdquo inProceedings of the International Conference on CommunicationInformation amp Computing Technology (ICCICT rsquo12) pp 1ndash5Mumbai India October 2012

[14] E Hussein and M A Belal ldquoDigital watermarking techniquesapplications and attacks applied to digital media a surveyrdquoInternational Journal of Engineering Research amp Technology vol1 no 7 2012

[15] Y Zhang J Wang and X Chen ldquoWatermarking techniquebased on wavelet transform for color imagesrdquo in Proceedings ofthe 24th Chinese Control and Decision Conference (CCDC rsquo12)pp 1909ndash1913 University of Science and Technology LiaoningAnshan China May 2012

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



Original data

Watermark

Secret key

Watermarkembedding

Watermarkeddata

(a) At the sender

Original data

Secret key

WatermarkwatermarkRecoveredextractingWatermarked

data

(b) At the receiver

Figure 1 Classical nonblind watermarking system

and videos can be embedded in the video watermarkingsystem (as proposed in this paper) The previous problemsof the nonblind watermarking have a more effect on videowatermarking systems So a solution for this problem is amore important issue in video watermarking

A proposed image watermarking technique based onimage interlacing has been used to solve thememory capacityand communications bandwidth problems of the nonblindwatermarking systems [4] the target system was imagewatermarking system with a single watermark (color imageas host and gray image as watermark) In this paper videowatermarking and multiple watermarks in different types(gray image color image and video) have been introduced

This paper is organized as follows A brief review ofvideo watermarking techniques is presented in Section 2The based techniques which the proposed method dependson are illustrated in Section 3 The proposed video multiplewatermarking technique is presented in Section 4 Simulationresults and discussions are given in Section 5 and finally theconclusion and the references are included

2 Video Watermarking Techniques

Digital video is a sequence of still images called frames andthese frames are loaded by a constant rate called frame rateSo by reading the frames from the video file frame-by-frameand dealing with each frame as a color image all imagewatermarking techniques can be used as video watermarkingtechniques [5]

As in image watermarking the watermark embeddingand extraction operations can be done in spatial domainor frequency domain In the spatial domain the watermarkis embedded by modifying the pixel values of the hostimagevideo directly In the frequency or transform domainthe watermark is embedded by modifying the frequencycomponents of the host imagevideo [5] this is done bytransforming the host signal to the frequency domain beforeembedding the watermark and retransforming to spatialdomain after that As a result spatial domain watermarkingtechniques are lower in complexity than frequency domaintechniques The characteristics of the human visual system(HVS) are better captured by the frequency coefficients thanspatial coefficients [1] thus frequency domain watermarkingtechniques enjoy better imperceptibility more robustness

against attacks such as noise addition pixel removal rescal-ing rotation and shearing plus more compatibility withcompression standards such as MPEG 1 2 and 4 (MovingPicture Experts Group)

Unlike imagewatermarking there is a third type of water-marking techniques called compressed-domain video water-marking techniques [6] Many digital videos are typicallystored and distributed in compressed form (eg MPEG)Due to the real-time requirements of video broadcastingthere is no time for decompression and recompression so thewatermark in these techniques is embedded directly in thecompressed videoThemain disadvantage of these techniquesis that the watermarked video can be highly susceptible tobe recompressed with different parameters or converted toformats other than MPEG

3 The Based Techniques

The proposed technique in this paper depends on imageinterlacing When deploying this technique to the videomultiple watermarking system three-level discrete wavelettransform (DWT) is used as an example of the frequencydomain watermarking For more security Arnold Transformas one of the popular imagevideo frame encryptionmethodsis used to encrypt the watermarks before embedding them inthe host video The following subsections will discuss thesetechniques before starting with the proposed method

31 Three-Level DWT Discrete fourier transform (DFT)discrete cosine transform (DCT) and discrete wavelet trans-form (DWT) are the most popular reversible transformsused in frequency domain watermarking But DWT as amultiresolution multilevel transform is much preferred inwatermarking thanDCTandDFT because it understands thehuman visual system (HVS) closer than them [7]

At the first level of DWT decomposition the imagevideoframe is filtered horizontally and vertically by low (L) andhigh (H) pass filters to produce four frequency subbands [5]LL1 as the low frequency subband and HL1 LH1 and HH1 asthe high frequency subbandsThe low frequency componentscontain the most significant portions of the imagevideoframe in which any modifications like watermarking candamage it [8] as shown in Figure 2(a) [9]The high frequencycomponents contain the least significant portions of the


LL1 HL1

LH1 HH1


LL3 HL3LH3 HH3

HL2

LH2 HH2

HL1

LH1 HH1






(

1199091015840

1199101015840) = (

1 1

1 2)(

119909

119910) (mod 119873) (1)













(c) Restored image







119908

sum

119894=1

ℎ

sum

119895=1

(119883119894sdot119895oplus 1198831015840

119894sdot119895) (2)

where 119883119894sdot119895

and 1198831015840119894sdot119895






Originalvideo

Interlacing

Subvideo A

Frame A

Frame B

X frames

Subvideo B(the most

similar to A)

Othersubvideos

Gray imagewatermark



R G B R G B

ArnoldTransform

ArnoldTransform

ArnoldTransform

R G B R G BN

C

R

R

G

G

B

B

Three-level

Three-level

Three-level

DWT

DWT

DWT

IDWT

IDWT

IDWT

Watermarkedvideo

Deinterlacing


Watermarkedframe A

Watermarkedframe B

WatermarkedX frames

1205721 1205722 1205723


Watermarkedvideo

Othersubvideos


Watermarkedframe A

Watermarkedframe B

Watermarked

N

C

RGB

RGB

C

RGB

RGB

Gray imagewatermark



R G B R G B

InverseArnold

InverseArnold

InverseArnold

11205721 11205722 11205723


Three-level

Three-level

Three-level

Three-level

DWT DWT

DWT

DWT

DWT

DWT

Frame A

Frame B

X frames

Subvideo B

Interlacing

X frames





1198681015840= 119868 + 120572119882 (3)





1198821015840=

(11986810158401015840minus 119868lowast)

120572

(4)





Red Green Blue

One ER-OR 09548 09645 09657EC-OC 09713 09760 09755

Two























2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)










where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




LL1 HL1

LH1 HH1


LL3 HL3LH3 HH3

HL2

LH2 HH2

HL1

LH1 HH1






(

1199091015840

1199101015840) = (

1 1

1 2)(

119909

119910) (mod 119873) (1)













(c) Restored image







119908

sum

119894=1

ℎ

sum

119895=1

(119883119894sdot119895oplus 1198831015840

119894sdot119895) (2)

where 119883119894sdot119895

and 1198831015840119894sdot119895






Originalvideo

Interlacing

Subvideo A

Frame A

Frame B

X frames

Subvideo B(the most

similar to A)

Othersubvideos

Gray imagewatermark



R G B R G B

ArnoldTransform

ArnoldTransform

ArnoldTransform

R G B R G BN

C

R

R

G

G

B

B

Three-level

Three-level

Three-level

DWT

DWT

DWT

IDWT

IDWT

IDWT

Watermarkedvideo

Deinterlacing


Watermarkedframe A

Watermarkedframe B

WatermarkedX frames

1205721 1205722 1205723


Watermarkedvideo

Othersubvideos


Watermarkedframe A

Watermarkedframe B

Watermarked

N

C

RGB

RGB

C

RGB

RGB

Gray imagewatermark



R G B R G B

InverseArnold

InverseArnold

InverseArnold

11205721 11205722 11205723


Three-level

Three-level

Three-level

Three-level

DWT DWT

DWT

DWT

DWT

DWT

Frame A

Frame B

X frames

Subvideo B

Interlacing

X frames





1198681015840= 119868 + 120572119882 (3)





1198821015840=

(11986810158401015840minus 119868lowast)

120572

(4)





Red Green Blue

One ER-OR 09548 09645 09657EC-OC 09713 09760 09755

Two























2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)










where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





(c) Restored image







119908

sum

119894=1

ℎ

sum

119895=1

(119883119894sdot119895oplus 1198831015840

119894sdot119895) (2)

where 119883119894sdot119895

and 1198831015840119894sdot119895






Originalvideo

Interlacing

Subvideo A

Frame A

Frame B

X frames

Subvideo B(the most

similar to A)

Othersubvideos

Gray imagewatermark



R G B R G B

ArnoldTransform

ArnoldTransform

ArnoldTransform

R G B R G BN

C

R

R

G

G

B

B

Three-level

Three-level

Three-level

DWT

DWT

DWT

IDWT

IDWT

IDWT

Watermarkedvideo

Deinterlacing


Watermarkedframe A

Watermarkedframe B

WatermarkedX frames

1205721 1205722 1205723


Watermarkedvideo

Othersubvideos


Watermarkedframe A

Watermarkedframe B

Watermarked

N

C

RGB

RGB

C

RGB

RGB

Gray imagewatermark



R G B R G B

InverseArnold

InverseArnold

InverseArnold

11205721 11205722 11205723


Three-level

Three-level

Three-level

Three-level

DWT DWT

DWT

DWT

DWT

DWT

Frame A

Frame B

X frames

Subvideo B

Interlacing

X frames





1198681015840= 119868 + 120572119882 (3)





1198821015840=

(11986810158401015840minus 119868lowast)

120572

(4)





Red Green Blue

One ER-OR 09548 09645 09657EC-OC 09713 09760 09755

Two























2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)










where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Originalvideo

Interlacing

Subvideo A

Frame A

Frame B

X frames

Subvideo B(the most

similar to A)

Othersubvideos

Gray imagewatermark



R G B R G B

ArnoldTransform

ArnoldTransform

ArnoldTransform

R G B R G BN

C

R

R

G

G

B

B

Three-level

Three-level

Three-level

DWT

DWT

DWT

IDWT

IDWT

IDWT

Watermarkedvideo

Deinterlacing


Watermarkedframe A

Watermarkedframe B

WatermarkedX frames

1205721 1205722 1205723


Watermarkedvideo

Othersubvideos


Watermarkedframe A

Watermarkedframe B

Watermarked

N

C

RGB

RGB

C

RGB

RGB

Gray imagewatermark



R G B R G B

InverseArnold

InverseArnold

InverseArnold

11205721 11205722 11205723


Three-level

Three-level

Three-level

Three-level

DWT DWT

DWT

DWT

DWT

DWT

Frame A

Frame B

X frames

Subvideo B

Interlacing

X frames





1198681015840= 119868 + 120572119882 (3)





1198821015840=

(11986810158401015840minus 119868lowast)

120572

(4)





Red Green Blue

One ER-OR 09548 09645 09657EC-OC 09713 09760 09755

Two























2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)










where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






1198681015840= 119868 + 120572119882 (3)





1198821015840=

(11986810158401015840minus 119868lowast)

120572

(4)





Red Green Blue

One ER-OR 09548 09645 09657EC-OC 09713 09760 09755

Two























2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)










where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in
















2552

MSE

MSE = 1119903 times 119888

119903

sum

119894=1

119888

sum

119895=1

(119867119894sdot119895oplus 1198671015840

119894sdot119895)

2

(5)










where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in












where119867119894sdot119895and1198671015840





PSNR NC















(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in














(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




(a) (b) (c) (d)

(e) (f) (g) (h)












(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




(a) (b) (c) (d)


(a) (b) (c) (d)



6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




6 Conclusion





References























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



Research Article Video Multiple Watermarking Technique ...downloads.hindawi.com/journals/tswj/2014/634828.pdfa robust video multiple watermarking technique is proposed to solve this

Documents