A Robust Image Watermarking Using Two Level DCT and ... · Watermark embedding is performed in many domains such as spatial, Fourier transform [1 ... A Robust Image Watermarking using

A. H. Taherinia1 and M. Jamzad2, Member, IEEE

1PhD Candidate, Sharif University of Technology, [email protected]

2Associative Professor, Sharif University of Technology, [email protected]

Abstract— In this paper we present a blind low frequency watermarking scheme on gray level images, which is based on DCT transform and spread spectrum communications technique. We compute the DCT of non overlapping 8x8 blocks of the host image, then using the DC coefficients of each block we construct a low-resolution approximation image. We apply block based DCT on this approximation image, then a pseudo random noise sequence is added into its high frequencies. For detection, we extract the approximation image from the watermarked image, then the same pseudo random noise sequence is generated, and its correlation is computed with high frequencies of the watermarked approximation image. In our method, higher robustness is obtained because of embedding the watermark in low frequency. In addition, higher imperceptibility is gained by scattering the watermark's bit in different blocks. We evaluated the robustness of the proposed technique against many common attacks such as JPEG compression, additive Gaussian noise and median filter. Compared with related works, our method proved to be highly resistant in cases of compression and additive noise, while preserving high PSNR for the watermarked images.

Index Terms— Blind Digital Watermarking, DCT, JPEG compression, Spread Spectrum Watermarking

I. INTRODUCTION n recent years, many digital watermarking techniques have been proposed to protect the copyright of digital multimedia

data. Watermark embedding is performed in many domains such as spatial, Fourier transform [1], DCT1 and DWT2 [2].

One of the commonly used domains for embedding a watermark in an image is the DCT [3]~[6]. DCT splits up the image into the frequency bands, so upon the application, the watermark can be embedded in different frequencies. Furthermore, the sensitivity of human visual system to DCT frequencies has been extensively studied; which resulted in the recommended JPEG quantization table [7]. These results can be used for predicting and minimizing the visual impact of distortion caused by embedding the watermark.

If we know the image compression domain, for example DCT, then it is better to embed watermark in those DCT

1 Discrete Cosine Transform 2 Discrete Wavelet Transform

coefficients which are unlikely to be discarded during the compression process. Since we are able to anticipate which DCT coefficients will be quantized by the compression scheme, we can choose not to embed the watermark in those coefficients. This approach can be extended to compression methods in other domains, as well. Furthermore, it is a common practice to apply additive noise for watermark embedding and use the correlation techniques for detection. In [8] a watermarking technique is provided in which a watermark is embedded as pseudo-random noise sequences into middle-frequency range of the image.

The major objective of this paper is to develop a watermarking algorithm based on DCT and spread spectrum communications in such a way that it is highly robust with respect to JPEG compression and also other common attacks. Compared with similar works, our method provided the highest robustness for extracted watermark especially when JPEG compression was applied to the watermarked image. In addition, this high level of robustness did not decrease its PSNR3 (PSNR is defined in section 4).

The rest of paper is structured as follows. In Section 2, some related works are introduced, and then in section 3 a new watermarking method is proposed. In Section 4, the performance of the proposed watermarking method is evaluated by applying various attacks including JPEG compression and adding Gaussian noise to the watermarked images. In section 5, the comparison of proposed method with other methods is reported. Finally in Section 6 we conclude our method.

II. RELATED WORKS Existing literature reveals two techniques for the

watermarking of images: transform domain and spatial domain [1]. Most of the recent watermarking schemes employ mainly the frequency domain approach because it is superior to the spatial domain approach in robustness and stability [1,2].

However, there is a crucial question that should be answered: which frequency band in frequency domain can be robust and imperceptible to various attacks? According to Weber’s rule,

3 Peak Signal-to-Noise Ratio

A Robust Image Watermarking using Two Level DCT and Wavelet Packets Denoising

I

2009 International Conference on Availability, Reliability and Security

978-0-7695-3564-7/09 $25.00 © 2009 IEEE

DOI 10.1109/ARES.2009.132

150

2009 International Conference on Availability, Reliability and Security

978-0-7695-3564-7/09 $25.00 © 2009 IEEE

DOI 10.1109/ARES.2009.132

150

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

the low frequency area is more robust than high and middle frequency areas. There have been various methods to embed the watermark into the low frequency area [22, 23]. It is known that embedded watermark in the prior approaches are robust to various attacks to a certain extent but it is likely to be destroyed if the distortion exceeds a particular level. Despite their robustness, the key concern is that if the low frequency components are changed, the image quality is degraded and the watermark becomes meaningless. Thus, of importance is to make the modifications that made by watermark embedding in low frequency coefficients as low as possible. This problem can be solved by using spread spectrum communication techniques for embedding watermark in low frequencies.

In spread spectrum communications, a narrowband signal is transmitted over a much larger bandwidth such that, the signal energy present in any single frequency is imperceptible. Similarly, in spread spectrum watermarking schemes, the host image is viewed as a communication channel, while the watermark is viewed as a signal to be transmitted. So the watermark is spread over many samples of the host signal by adding a low energy pseudo-random noise sequence to them. The embedded watermark sequence is detected by correlating this specific pseudo random noise sequence with the watermarked signal itself.

In [2], [3], [4] and [14] some techniques are proposed in which the watermark is embedded in the middle and high frequency components. The low frequency components are left unchanged in order to decrease the visibility of the watermark. Embedding the watermark in middle and high frequency components makes these techniques vulnerable to attacks such as compression and noise addition.

It is known that most of the energy of natural images is concentrated in the lower frequency range. Therefore, most lossy compression methods quantize and discard the information hidden in the higher frequency components. However, the human eye is more sensitive to noise in lower frequency components than in higher frequency ones. In order to invisibly embed the watermark that can survive lossy data compressions, a reasonable trade-off is to embed the watermark as low energy pseudo-random noise sequences into the low-frequency range of the image [24].

Also there are other argues that, using the same transforms for both watermarking and compression will result in optimal performance or using complementary transform. For example in [6], Fei et al. proposed using complementary transforms can potentially provide greater robustness.

But in this paper we show that using the same transforms for both watermarking and compression demonstrates the superiority of robustness and performance. That is, by anticipating which coefficients would be modified by the subsequent transform and quantization, we were able to produce a watermarking technique which has the highest resistance to JPEG compression compared with well known recent works. We could extract the watermark even if the watermarked image is compressed by JPEG with quality factor of 1%.

Moreover, watermarking techniques can be divided into two

distinct categories depending on the necessity of original images for the watermark extraction. Although existence of original image may facilitate watermark extraction to a certain extent, two problems can come out: (1) At the risk of insecurity the owners of original images may be compelled to share their work with anyone who wants to check the existence of the watermark and (2) it is time-consuming and cumbersome to search out the originals that correspond to a given watermark within the database. Thus, in order to overcome these problems we need a method for extracting the embedded watermark without requiring the original image.

This method is called a blind watermarking technique. Such techniques appear far more useful since the availability of an original image is usually unwarranted in real-world scenarios.

As described, for extraction of watermark from watermarked image we do not need to have the original host image.

Recently, research efforts have been devoted to security analysis in which successful attacks have been proposed to defeat previously proposed multimedia authentication systems [20]. It is well known that many digital watermarking schemes, especially quantization based schemes, are weak against well-designed sophisticated attacks [23]. Therefore, in the watermark-based authentication systems, security of the overall system including authenticator generation and embedding must be considered. In our development, we assume Kerckhoffs’ principle [28] which requires that the opponent knows the details of all aspects of the authentication system except for the secret key shared between the transmitter and the receiver. We adopt the following stringent definition of security: given that an opponent has full knowledge of the watermarking system details except for the secret key, it must be computationally infeasible for the opponent to alter the watermarked data in an illegitimate manner such that the modified copy is wrongly accepted as legitimate.

III. PROPOSED METHOD In this paper, we propose a novel watermarking scheme

which is based on low frequencies of DCT transform and spread spectrum watermarking. In this method all the DC components of the block DCT transform of the original image are grouped together to form a pseudo image called DC image. Then each bit of the watermark is scattered through the high frequencies of DCT transform of this DC image. In other word, each bit is scattered in 64 blocks of the original image. Therefore, we obtain the robustness because of embedding the watermark in low frequency and gain the imperceptibility by scattering the watermark's bit in different block.

We then compute the NC4 and PSNR to judge the robustness and the invisibility of the watermarking algorithm (NC and PSNR are defined in section 4).

The PSNR values for watermarked images are all greater than 38 dB, which is the empirical value for the image without any perceivable degradation [20].

4 Normalized Cross-Correlation

151151

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

A. Embedding Algorithm Without loss of generality, we assume the host image is of

size 512x512. BDCT 5 is applied on 8x8 non overlapping blocks. Then to embed the watermark, for each 8x8 transformed block of host image, only its DC coefficient is selected out of the 64 DCT coefficients. In each block, DC coefficient is the most important coefficient which has the largest value. Embedding watermark in DC coefficient makes the watermark robust against many attacks. Those selected coefficients are then mapped into a reduced image which is called low-resolution approximation image (LRAI). Therefore, the size of extracted LRAI is always 1 64 of the host image. For example for 512x512 host images, the size of extracted LRAI is always 64x64 pixels (Fig.1).

After extracting LRAI from host image, the extracted LRAI is divided into 8x8 non-overlapping blocks and BDCT of each block is calculated. Then according to the value of watermark bit which is going to be embedded in each block, a pseudo random noise sequence is added to the high frequencies of DCT transform of each 8x8 block of LRAI using equation (1). There are different definitions for high frequencies in a DCT block, but our experiments shows that the definition which is shown in Fig. 1 has low visual impact on watermarked image and also provides more accurate watermark detection.

Coefficients in the low and middle frequencies that are copied over to the watermarked LRAI remain unaffected.

, ,

,,

( , ) ( , ), ,( , )

( , ), ,

ix y Hw x y

x yx y H

L u v k W u v u v FL u v

L u v u v F

⎧ + × ∈⎪= ⎨∉⎪⎩

(1)

In (1), L denotes DCT transform of LRAI, FH the high band

frequencies, k the gain factor, (x,y) the location of an 8x8 block

5 Block based DCT

of LRAI, (u,v) the DCT coefficient in the corresponding 8x8 block of L, and Wi the pseudo random noise sequence according to the value of i.

We should note here that two separate pseudo random noise sequences are used to represent the bit values of 0 and 1. Furthermore, by choosing these two pseudo random noise sequences to be as un-correlated as possible, we can significantly reduce the rate of false detection.

Then each block is inverse-transformed to give us watermarked LRAI. The final step to construct the watermarked image is to replace the DC coefficients of LRAI with their corresponding watermarked ones, and then compute the IDCT transform of each 8x8 block of watermarked LRAI.

Since in each block of LRAI only one bit of watermark is embedded, so for a host image of size 512x512, watermark size is limited to 8x8 pixels. In other words, assume the size of host image is 2 2n n× , so the size of extracted LRAI will be

3 32 2n n− −× and maximum size of watermark will be 6 62 2n n− −× . Regarding to the watermark size we should note that practically a character string of length 10, or seventy bits, is enough for generation of an identification code and the authentication purpose [9].

The embedding algorithm is summarized as follows:

Embedding algorithm: 1. Compute BDCT of host image, 2. Create LRAI, 3. Compute BDCT of LRAI, 4. Embed each bit of Watermark, as a

pseudo-random noise sequence into FH coefficients of each 8x8 block of transformed LRAI using (1),

5. Compute IBDCT of watermarked LRAI, 6. Replace the watermarked DC coefficients with

the original ones in LRAI, 7. Compute IBDCT of LRAI.

152152

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

gethgenuth

emwenadtewdiemimimThimacsecotra0 ww

avprpr(fi

fawPSexde

B. Extraction To detect th

enerator algorhat using a denerator willumbers. In thihe watermark p

As it was mbedding the

watermarking tncoded as a pdded to the cchnique, sim

wavelet packeistortion that mbedding stagmage must be mage from wahis new imag

mage by embections as expection, the tonstructed. Tansformed LRand 1 are com

with noise pattwatermark bit o

The presencverage correlare-defined thrresent if the afig. 4). The ext

Extraction 1. Denoise w2. Calculate

image from3. Do steps

for differe4. Compute

transformezero bits,

5. Choose trelated to

6. Compute extracted pre-definebe presengreater tha

In Fig. 2 the

actor k=30 is with this gain fa

SNR of waterxtracted wateetectable, but

n Algorithm he watermarkrithm is seededifferent key produce a is case the expattern [8]. explained in

e watermark wtechnique; in apseudo-randocoefficients o

milar to the prets denoising

was added tge. In order denoised and

atermarked ime indicates thedding the waplained in emtransformed Then the cRAI with both mputed. If thern 0, then a of 1 is detectede of watermaation coefficiereshold. The verage correlatraction algori

algorithm: watermarked im

difference imm watermarke1 through 3 f

ence image, the correlat

ed LRAI, and

the noise pat the extractedthe average cwatermark bi

ed threshold. Tnt if the averan the thresho

e Goldhill imashown. As it actor has no vrmarked imagermark, using

it has 4 bits

k, the same pd with the samas seed of p

different sextraction algor

n the embeddwe benefit froa way that eacm noise sequ

of the host imroposed meth

g method toto the host ito do this, fi

d then by subtmage, a new im

e noise that watermark. By pmbedding algLRAI of d

correlation bnoise patternse higher correwatermark bid in a similar ark is detecteent of detecte

watermark iation is greateithm is given

mage, mage by subed image, from the Emb

tion betweend both noise pa

ttern with hd watermark bicorrelation coits, and then cThe Watermarrage correlati

old.

age that is wais seen, wate

visual impact oge is around 4g our previous of error. Bu

pseudo-randomme key. It is

pseudo-randomequence of rrithm will not

ding algorithom spread sp

ch bit of wateruence and themage. By usihod in [10], extract theimage in watirst the watertracting the demage will be cwas added to tperforming thgorithm in pdifference imbetween blocs related to bitelation was oit of 0 is deteway.

ed by compared watermark is considereder than this thas follows:

tracting deno

bedding algori

n each blockatterns for one

higher correlait,

oefficient ovecompare it wirk is considereion coefficien

atermarked wiermarking thison the host ima43 dB. Althouus method [ut we improv

m noise known

m noise random t detect

hm, for pectrum rmark is en it is ing this we use

e noise termark rmarked enoised created. the host he same previous mage is cks of t values

obtained cted. A

ring the with a

d to be hreshold

oised

ithm

k of e and

ation

r all ith a ed to nt is

ith gain s image age and ugh the [11], is ved our

pre

de[29

exexNC

F

evious work aFirst the watenoising with d9], (fig.3). ThFinally, the tracted by thtracted waterC=1.

Fig. 3- Wat

Fig. 5- The differimage and

Fig. 2. Wawatermark with

as follows. ermarked imagdefault paramen the differenhidden wate

he proposed rmark will b

termarked image

ence image betwd Extracted waterm

atermarked imageerror=4 bits, NC=

ge is denoisedmeters, availabl

nce image is cermark in dmethod. In

be recovered

after wavelet pac

een watermarkedmark with no erro

e k=30, PSNR=42=0.86, denotes

d by wavelet ple in Matlab Rcalculated (fig

difference imthis approacwith no erro

ckets denoising

d image and denoor and NC =1

2.78dB and extracerrors in the extr

packets R2007b g.5). age is

ch, the or and

ised

cted raction

153153

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

IV. EXPERIMENTAL RESULTS AND DISCUSSIONS The proposed technique has been conducted on different

standard test (host) images of size 512x512 with different level of details. Also robustness of the proposed technique against most common attacks is evaluated. For applying the attacks to the watermarked images, Stirmark[12] and Checkmark[13] benchmarks are used. An 8x8 watermark pattern as shown in fig. 5 is embedded in the host images.

Some experimental results on Goldhill test image which is watermarked with gain factor k=30 are given in Figs. 6~7.

In our experiments, we compute the PSNR to judge the difference between the original image and watermarked image or the attacked watermarked image. The PSNR similarity measure is defined as follows:

1 2

21 2

102

1 1

max( ( , ))10log ( )[ ( , ) ( , )]

M M

i j

M M f i jPSNR dBf i j f i j

= =

× ×=

′−∑∑ (2)

where M1 and M2 are the size of image. ( , )f i j is the original image, ( , )f i j′ is the watermarked image, or the attacked watermarked image.

Also, we compute NC to quantitatively analyze the similarity of the extracted and the original watermark.

1 2

1 2 1 2

1 1

2 2

1 1 1 1

( , ) ( , )

[ ( , )] [ ( , )]

M M

i j

M M M M

i j i j

W i j W i jNC

W i j W i j

= =

= = = =

′×=

′×

∑∑

∑∑ ∑∑ (3)

where M1 and M2 are the size of watermark image, ( , )W i jand ( , )W i j′ are the original and the extracted watermark, respectively.

In the following to analyze the performance of our method, first as a sample, we show its robustness with respect to JPEG compression and Gaussian noise when an image is watermarked with one gain factor. Then to provide a more

detailed analysis we watermark the Lena image with several gain factors and show its behavior when the watermarked image has gone through different attacks such as JPEG and JPEG2000 compression, different kinds of noise and etc. And finally we show the result of analysis when several images are watermarked using different gain factors.

A. A sample of performance analysis for gain factor k=30 In this section we present the robustness with respect to

JPEG and Gaussian noise for only one gain factor (k=30).

JPEG compression Digital images usually are stored and transmitted after image

compression. JPEG is popular among image compression methods for still images. We examined the robustness of the proposed scheme by compressing the watermarked images with JPEG compression with quality factor 8 (Fig. 6). The extracted watermark from this picture is also shown.

As it is seen in Fig. 6 even by setting the quality factor of JPEG compression to 8%, the watermark is extracted with NC=0.96 which is greater than the empirical threshold of NC=0.4. This is one of the most important advantages of our method.

Noise addition We evaluate the robustness by adding Gaussian noise on the

watermarked image. Fig. 7 shows the result of adding 20% Gaussian noise to the watermarked image.

Although the attacked image is absolutely distorted by additive Gaussian noise which drops its PSNR to 9.07 dB, but the watermark is extracted with NC=0.71 (Fig.14). It indicates that the proposed scheme is also robust to noise attack.

B. Performance analysis in presence of different attacks for different gain factors In this section we discuss the robustness of our proposed

method against most common attacks on the standard image of Lena that is watermarked with different gain factors.

154154

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

G

ofcoin

k =

JP

JPtefafaLecoexshcow

ap

Gaussian noiseIn Fig. 8 rob

f different varonclude that increases the ro

FiJP

and ex

Fig 7. Atta= 30, PSNR = 9.0

PEG and JPEFig. 9 and F

PEG2000 comst image Len

actors. As it iactor, robustneena which isompressed usxtracted waterhows the greompression. N

watermarked im

C. Performadifferent imagIn Fig. 11 a

pplying the p

e bustness is testriances. Basedncreasing gai

obustness of w

ig. 6- Attacked wPEG Compressionxtracted watermar

acked watermarke07 dB and extracte

EG2000 ComprFig. 10 show mpressions wina which is is clear from ess of waterms watermarkeing JPEG wi

rmark and origeat robustnesNote that accmage with gain

ance analysis wges and Fig. 12 wproposed meth

ted against add on the resulin factor in em

watermark aga

watermarked imagn QF= 8% , PSNRrk with Error=1 b

ed image plus 20%ed watermark wit

ressions the results ofth different cwatermarked these figures

mark is improved with gainith QF = 1%ginal watermars of our me

cording to Fign factor 25 is

with several g

we show the hod on 8 diff

dding Gaussialts of Fig. 8, mbedding proainst Gaussian

ge, k=30, R = 23.24 bit, NC= 0.96

% Gaussian noiseth Error= 9 bits, N

f applying JPEcompression r

with differens, by increasinved. For examn factor of 2%, the correla

rk is about 0.7ethod againstg. 11, PSNRabout 44 db.

gain factors on

results obtaifferent standar

an noise we can

ocedure, n noise.

e

NC = 0.71

EG and atios to nt gain ng gain

mple for 25, and ation of 73. This t JPEG

of the

n

ined by rd gray

scaanunwaorihigwa

Figb

c

ale images (Lnd Zelda). All nderstood thatatermarked imiginal watermgher gain factatermark robu

Fig. 8 – The relwatermarked imaextracted waterm

g. 9 – The relatiobetween original

Fig. 10 – The relcorrelation betwe

Lena, Boat, Maof them were t higher gain

mages, but as smark and the ex

tors. The tradustness is abou

lation between thage and the normmark using differe

on between JPEG and extracted wa

waterm

lation between JPeen original and e

factors for w

andrill, Barbasized 512x51

n factors makshown in Fig.xtracted waterde-off betweenut gain factors

e amounts of addmalized correlation

ent gain factors fo

quality factor anatermark using di

marking Lena

PEG2000 quality extracted watermawatermarking Len

ara, Goldhill, 12. From Fig. ke lower PSN12 the correlarmark gets betn visual quali 25 to 30.

ded Gaussian noisn between originaor watermarking

nd normalized corifferent gain facto

factor and normaark using differenna

Pepper 11 it is

NR for ation of tter for ity and

se to al and Lena

rrelation ors for

alized nt gain

155155

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

V. COMPARING PROPOSED METHOD WITH OTHER METHODS In this section, we compare the average robustness of the

proposed method with 3 other recently published methods which will be described later in this section. Also we compare the results with our previous method which did not have the denoising scheme [15]. The proposed scheme in [25] is based on embedding a pseudorandom sequence of real numbers in DCT coefficients of each segment of the host image. It relies on some of the ideas proposed by Cox et al. [8] and outperforms the Cox algorithm but still it is not robust against most of attacks. In their scheme, rather than embedding the watermark globally in the host image as the Cox algorithm suggests, the host image is first segmented in different segments based on Voronoi diagram and the feature extraction points. Then, a pseudorandom sequence of real numbers is embedded in the DCT domain of each image segment. This method is referred as Seg-DCT.

In [26], an approach is proposed which is based on the method given by Dugad et al. [21]. In [26] the watermark is embedded in the discrete multi-wavelet transform (DMT) coefficients that are larger than some threshold values. They use the GA6 techniques to search for optimal values for these parameters in order to achieve optimum performance. They

Fig. 11 – The relation between gain factor and PSNR of watermarked images

for 8 different test images

Fig. 12 – The relation of gain factor and normalized correlation between

original and extracted watermark for 8 different test images

6 Genetic Algorithm

compared the experimental results before and after optimization using GA and also compared them with the results of previous works. We choose GA-Dugad method which has the best performance from [26] to compare with our results.

In [27], two novel transforms are defined, which are respectively called DWT-IDCT transform and DCT-IDWT transform. Then two novel watermarking schemes based on these two transforms are proposed. The authors claim that the scheme based on DCT-IDWT has better robustness against the attacks than the other, so we choose it to compare with our method.

In Fig. 13 the average robustness of our proposed method and other mentioned methods against JPEG compression with different quality factors are shown. As described before, and it is clear from this figure, our proposed method has the highest robustness against JPEG compression.

In Fig.14, robustness of the mentioned methods against median filtering of different size is reported. This figure shows that even against median filtering of size 9x9 our proposed method has higher robustness compared with other methods.

Fig. 13 – The relation between JPEG quality factor and normalized

correlation between original and extracted watermark using different methods for watermarking Lena

Fig. 14 – The relation between the median filtered watermarked image and the

normalized correlation between original and extracted watermark using different methods for watermarking Lena

156156

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.

VI. CONCLUSION In this paper, we proposed a DCT-based blind watermarking

scheme based on spread spectrum communications. The low frequency nature of the proposed algorithm makes the embedded watermark very robust to common image manipulations such as filtering, scaling, compression and malicious attacking. By anticipating which coefficients would be modified by the subsequent transform and quantization, we were able to produce a watermarking technique which to the best of our knowledge has the highest resistance to JPEG compression compared with well known similar works. We could extract the watermark even if the watermarked image is compressed with JPEG with quality factor of 1%. In addition, our method is also robust with respect to additive Gaussian noise, median filtering and other attacks that were mentioned in this paper.

In our future works, we will try to generalize the proposed method for color images. Also we will consider other compression techniques like EPIC [16], SPIHT [17], EZW [18] and JPEG2000 [19] for watermarking. Since for each compression technique, it is better to embed watermark in those coefficients which are unlikely to be discarded during the compression process. For a known compression technique we are able to anticipate which coefficients will be quantized by the compression scheme; thus we can choose not to embed the watermark in those coefficients. So the watermark will be robust against that compression technique.

REFERENCES Lu C.-S., "Multimedia Security: Steganography and Digital Watermarking Techniques for Protection of Intellectual Property", Idea Group Publishing, 2005, ISBN: 1-59140-192-5

[1]

Katzenbeisser S. and Petitcolas A.-P., "Information Hiding Techniques for Steganography and Digital Watermarking," Artech House, January 2000, ISBN: 1580530354.

[2]

Voloshynovskiy S. and Deguillaume F., "Information-Theoretic Data-Hiding: Recent Achievements And Open Problems", International Journal of Image and Graphics, Vol. 5, No. 1, p. 5–36, 2005

[3]

P. Meerwald, A. Uhl, “A Survey of Wavelet-Domain Watermarking Algorithms”, EI San Jose, CA, USA, 2001

[4]

Cox I. J., Miller M. L., "The First 50 years of Electronic Watermarking", Journal of Applied Signal Processing, vol. 2, Issue 2, p. 126, 2002.

[5]

Fei C., Kundur D. and Kwong R., "Analysis and Design of Watermarking Algorithms for Improved Resistance to Compression", IEEE Trans. on Image Processing, Vol. 13, No.2, p.760-771, 2004.

[6]

Langelaar G., Setyawan I. and Lagendijk R., "Watermarking digital image and video data: A state-of-the-art overview," IEEE Signal Processing Magazine, vol. 17, no. 5, p. 20-46, September 2000

[7]

Cox I.J., Kilian J., Leighton T., Shamoon T., "Secure spread spectrum watermarking for multimedia", IEEE Trans. on image processing, vol. 6, no. 12, p. 1673, 1997.

[8]

Ashourian, M., Enteshari, R., Jeon , J., "Digital Watermarking of Three-dimensional Polygonal Models in the Spherical Coordinate System", Proc. of the IEEE Computer Graphics International (CGI'04), 2004

[9]

Pan Z., Li L., Zhang M. and Zhang D., "Watermark Extraction by Magnifying Noise and Applying Global Minimum Decoder", IEEE Proceedings of the 3rd International Conference on Image and Graphics (ICIG’04), p. 349 - 352, 2004

[10]

Taherinia A.H., Jamzad M., "Two level DCT based digital watermarking", Proc. of 13th International Conference on Systems, Signals & Image Processing (IWSSIP'06), September 2006, Budapest, Hungary

[11]

Peticolas A.-P., Anderson R.-J., and Kuhn M.-G., "Attacks on copyright marking systems", In 2nd Workshop on Information Hiding, Portland, Oregon, April 1998

[12]

Pereira S., Voloshynovsky S., Madueo M., Marchand-Maillet S. and Pun T., “Second generation benchmarking and application oriented evaluation,” Information Hiding Workshop, April 2001, PA, USA.

[13]

Meerwald P., "Digital image watermarking in the wavelet transform domain", Master's Thesis, Salzburg University, Salzburg, Austria, January 2001

[14]

Taherinia A.H., "Improving Watermarking Robustness against Different Types of Noise and Compression", Master's Thesis (in Persian), Sharif University of Technology, Tehran, Iran, September 2006.

[15]

Simoncelli E. and Adelson E. H., "Subband image coding with three-tap pyramids [EPIC]", Picture Coding Symposium, p. 143-192,Cambridge, MA. Mar, 1990.

[16]

Said A. and Pearlman W. A., "A new, fast, and efficient image codec based on set partitioing in hierarchical trees", In IEEE Transactions on Circuits and Systems for Video Technology, vol. 6, p. 243 - 250, June 1996

[17]

Shapiro J. M., "Embedded image coding using zerotrees of wavelet coefficients", In IEEE Transactions on Signal Processing, volume 41, p. 3445 - 3462, December 1993

[18]

Charrier M., Santa Cruz D., Larsson M., "JPEG2000, the Next Millennium Compression Standard for Still Images", In Processing of IEEE International Conference on Multimedia and Computing Systems, ICMCS '99, vol. 1, p. 131-132, Florence, Italy, June 1999.

[19]

M.S. Hsieh, D.C. Tseng, Perceptual digital watermarking for image authentication in electronic commerce, Electron. Commerce Res. 4 (2004) 157–170.

[20]

R. Dugad, K. Ratakonda, and N. Ahuja, “A new wavelet-based scheme for watermarking images,” in Proc. IEEE Int. Conf. Image Processing (ICIP 1998), vol. 2, Oct. 1998, p. 419–423.

[21]

Ho Seok Moon, Taewoo You, Myung Ho Sohn, Hye Soo Kim and Dong Sik Jang, "Expert system for low frequency adaptive image watermarking: Using psychological experiments on human image perception", Expert Systems with Applications 32 (2007) p. 674–686

[22]

Dimitar Taskovski, Sofija Bogdanova and Momcilo Bogdanov, "Blind Watermarking via Low Frequency Component Modification",4th EURASIP Conference focused on Video and lmaqe Processing and Multimedia Communications, 2-5 July 2003, Zagreb, Croatia

[23]

Xingliang Huang and Bo Zhang, "Statistically Robust Detection of Multiplicative Spread-Spectrum Watermarks", IEEE Transactions on Information Forensics and Security, Vol. 2, NO. 1, March 2007

[24]

Mohamed A. Suhail and Mohammad S. Obaidat, "Digital Watermarking-Based DCT and JPEG Model", IEEE Transactions On Instrumentation and Measurement, Vol. 52, NO. 5, p.1640-1647, October 2003

[25]

Prayoth Kumsawat, Kitti Attakitmongcol and Arthit Srikaew, "A New Approach for Optimization in Image Watermarking by Using Genetic Algorithms", IEEE Transactions on Signal Processing, Vol. 53, NO. 12, p.4707-4719, December 2005

[26]

Jinwei Wang, Guangjie Liu, Yuewei Dai, Zhiquan Wang and Shiguo Lian, "Robustness Analysis of Watermarking Schemes Based on Two Novel Transforms," First International Conference on Innovative Computing, Information and Control, ICICIC '06, Vol. 3, 30-01 Aug. 2006 Page(s):471 – 474

[27]

Kerckhoffs, A. "La Cryptographie Militaire", Journal des sciences militaires, vol. 4, p. 5–38, January 1883

[28]

Matlab Image Processing Toolbox Version 6.0 (R2007b), “www.mathworks.com” , 2007

[29]

157157

Authorized licensed use limited to: Jawaharlal Nehru Technological University. Downloaded on July 20, 2009 at 22:55 from IEEE Xplore. Restrictions apply.