Content uploaded by Mahdieh Ghazvini
Author content
All content in this area was uploaded by Mahdieh Ghazvini on Apr 08, 2020
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=wasr20
Download by: [Hacettepe University] Date: 19 April 2017, At: 22:50
Journal of Applied Security Research
ISSN: 1936-1610 (Print) 1936-1629 (Online) Journal homepage: http://www.tandfonline.com/loi/wasr20
An Improved Image Watermarking Method in
Frequency Domain
Mahdieh Ghazvini, Elham Mohamadi Hachrood & Mojdeh Mirzadi
To cite this article: Mahdieh Ghazvini, Elham Mohamadi Hachrood & Mojdeh Mirzadi (2017)
An Improved Image Watermarking Method in Frequency Domain, Journal of Applied Security
Research, 12:2, 260-275, DOI: 10.1080/19361610.2017.1277878
To link to this article: http://dx.doi.org/10.1080/19361610.2017.1277878
Published online: 04 Apr 2017.
Submit your article to this journal
Article views: 7
View related articles
View Crossmark data
JOURNAL OF APPLIED SECURITY RESEARCH
,VOL.,NO.,–
http://dx.doi.org/./..
An Improved Image Watermarking Method in Frequency
Domain
Mahdieh Ghazvinia, Elham Mohamadi Hachroodb,andMojdehMirzadi
a
aDepartment of Computer Engineering, Faculty of Engineering, Shahid Bahonar University of Kerman,
Kerman, Iran; bDepartment of Computer Engineering, Islamic Azad University, Kerman Branch,
Kerman, Iran
KEYWORDS
Digital image watermarking;
discrete wavelet transform;
discrete cosine transform;
genetic algorithm
ABSTRACT
Digital watermarking is a technique for resolving copyright law
in E-Commerce. In this article a combined watermarking method
based on Discrete Wavelet Transform (DWT) and Discrete Cosine
Transform (DCT) is proposed. For embedding watermark, a cover
image is decomposed by a 2-level DWT, and the HL2 subband
coecients are divided into 4 ×4 blocks, then the DCT is per-
formed on each of these blocks. The watermark bits are embed-
ded by predened pattern_0 or pattern_1 on the middle band
coecients of DCT. After inserting watermark, inverse DCT is
applied to each 4 ×4 blocks of HL2 subband coecients, and
inverse DWT is applied to obtain the watermarked image. For
watermark extraction, the watermarked image, which may be
attacked by various image attacks, is decomposed with 2-level
DWT and DCT similarly as watermark embedding process, then
correlation between middle band coecients of block DCT and
the predened pattern (pattern_0 and pattern_1) is calculated to
decide whether a 0 bit or a 1 bit is embedded. Genetic algorithm
is used to optimize the performance of embedding and extract-
ing parameters. Simulation results show this technique is robust
against JPEG attacks and many other strong attacks.
Introduction
Today, with fast progress of digital information technology, digital media is easily
available and distributable for all. This media can be copied and modied by users,
resulting in unauthorized replication. This problem has become a critical issue so
that protecting the copyright of digital media is considered an important task. Digi-
tal watermarking is used to solve the problem of copyright violation. It is a technique
by which any watermark is embedded into a cover document using some known
algorithms to identify the original creator and owner of the document. The water-
mark can include information such as copyright (Narang & Vashisth, 2013).
Digital watermarks have three application elds: copyright protection, data
authentication, and data monitoring (Kaushik & Dua, 2014). Digital watermarking
CONTACT Mahdieh Ghazvini mghazvini@uk.ac.ir Computer Engineering Department, Shahid Bahonar Uni-
versity, Pajoohesh Sq., -, P.O. Box -, Kerman, Iran.
Color versions of one or more of the figures in this article can be found online at www.tandfonline.com/wasr.
© Taylor & FrancisGroup, LLC
JOURNAL OF APPLIED SECURITY RESEARCH 261
is invented by Emil Hembrook in 1954. In this invention, an invisible ID code was
attached to digital music le in order to prove copyright (Chandramouli & Memon,
2001).
Types of digital watermarking
Digital Watermarking techniques can be classied in a number of ways depend-
ing on dierent parameters like cover medium, resistance, visibility, extraction
method, and embedding domains (Kumar & Malhotra, 2014,asshowninFigure
1). There are four types of watermarking systems in terms of type of the doc-
ument which is watermarked. Watermarking systems of text (Qi & Xin, 2011),
sound (Nagarjuna & Ranjeet, 2013;Yang,Lei,Liu,Zhou,&Luo,2012), image
(Inoue, Miyazaki, & Katsura, 1999; Srivastava, Srivastava, & Srivastava, 2015;Wolf-
gang & Delp, 1996), and video (Checcacci, Barni, Bartolini, & Basagni, 2000;
Wolf g ang , P o di l chuk , & D elp, 1999). Watermarking systems, in terms of their
resistance against dierent attacks, are divided into three groups: robust (Chen,
Chang, & Hwang, 2012;Chen&Zhu,2012;Lee,Chen,Chang,&Tsai,2011;Sri-
vastava et al., 2015;Zhang,Li,&Wei,2012; Zhou, Wang, Xiong, & Yu, 2010),
semifragile (Di Martino & Sessa, 2012;Huo,He,&Chen,2012;Xiao&Shih,
2012), and fragile systems (Hu, Wang, Cao, & Yang, 2011;Qi&Xin,2011;Zhang,
Zhu, Wang, Wang, & Ma, 2012).There are two types of methods in terms of vis-
ibility of watermark within document, methods in which watermark is observ-
able and visible in the watermarked document and methods that include invisi-
ble watermark. There are three methods of watermarking in terms of watermarking
extraction method: blind, semiblind, and nonblind (Hallur, Kuri, Sudi, & Kulkarni,
Figure . Types of digital watermarking.
262 M. GHAZVINI ET AL.
2015). Finally, the most important classication is related to the type of process-
ing methods. In this respect, watermarking methods are classied into two classi-
cations: spatial domain processing and frequency/transform domain processing
(Hallur et al., 2015;Liu&Zhao,2010; Waleed, Jun, Hameed, & Kamil, 2015;Q.
Zhang et al., 2012).
Robust Watermarking: Robust watermarking is a method in which modica-
tion to the watermarked media will not aect the watermark. Fragile watermarking
which is opposed to this, is a method in which the watermark gets destroyed when
the watermarked media is modied or tampered with (Husain, 2012). This type of
watermarking has broad practical applications and many researches have been per-
formed in this area and this type resists intentional and unintentional attacks well.
Supporting copyright is one of the applications of this type of watermarking (Hu,
Wang , Liu , & Gu o, 2011).
Invisible Watermarking: In it, invisible watermark is hidden in the cover media. It
canbedetectedbyanauthorizeduseronly(Kumar&Malhotra,2014). This is a type
ofwatermarkwhichisspecicallyusedtoauthenticateuserorowneranddetecting
unauthorized copier (Kumar & Malhotra, 2014), or even to build the capacity of
adding security information or to confront forging document and securities such
as banknotes and bank checks and so on. The main priority of invisible watermark-
ingisinviolabilityagainstattacksofattackers(Wolfgang&Delp,1996). In Invisible
methods the signal isn’t changed, there are only minor variations in the output sig-
nal. End users don’t know the invisible watermarks, while the addition of watermark
to the signal does not restrict that signal’s use; it provides a mechanism to recognize
the original owner (Kumar & Malhotra, 2014).
Fragile Watermarking: This classication of watermarking does not resist the
broadgroupsofintentionalandunintentionalattacksanddoesnotfeature
extraction.
Nonblind Watermarking: In this method, recipient needs the main image before
embedding for extracting watermark by using intended extraction algorithm.
Semiblind Watermarking: In this method, recipient needs watermark string and
value of some parameters of embedding time for extracting watermark by using
intended extraction algorithm.
Blind Watermarking: In this method, recipient does not need the main image or
watermark string (Husain, 2012) for extracting watermark by using intended extrac-
tion algorithm.
Spatial domain methods
In spatial domain watermarking, during watermark embedding some transforma-
tions are performed directly on the image pixels and no transforms are applied to
the host image. In this method the watermark is embedded by modifying the pixel
value of an image directly (Kumar & Malhotra, 2014).
Although, the information which can be hiding by using spatial domain is
relatively high and distortion caused by this information embedding is very
JOURNAL OF APPLIED SECURITY RESEARCH 263
low, however, existing techniques in this regard are so weak against attacks like
lossy compression and some geometric attacks such as cutting. This watermarking
method is simple but is not robust to common signal processing operations, because
this method does not spread the watermark all over the image and some common
signal processing can easily delete the embedded watermark without damaging
the quality of the watermarked image (Kumar & Malhotra, 2014). So this type of
watermarking is mostly used in fragile watermarking systems. Least Signicant Bit
(LSB) insertion is one of the examples of this category. Although algorithms in this
method have low payload, they can be easily discovered and since pixel strengths
are directly changed in these algorithms, the quality of image after embedding and
extracting the watermark is not acceptable (Sharma & Prashar, 2012).
Frequency domain methods
In frequency (or transform) domain, the watermark is embedded in host image
by modifying the transform coecients, then the image is transformed to the fre-
quency domain, and again transform coecients are modied (Kumar & Malhotra,
2014).
On the other hand, rst by appropriate transform function, the host image is
transferred from spatial domain to frequency domain where watermark informa-
tion is embedded in the image and then is returned to spatial domain. In fact, in
transform domain some frequencies are selected and modied from their original
values according to some rules. For example, in frequency transforms, a watermark
is embedded by frequency components, although this change has eect on illumina-
tion of pixels indirectly. In transform domain, transform operation is usually exerted
on image twice that results in increase of computational complexity and insertion
time of watermark. Because the host image changes are often seen by human vision
system (HVS) in some methods of transform domain, certain characteristics can be
used in embedding watermark. As human eyes cannot understand changes in high
frequency components, a suitable transparency can be created in watermarking by
embedding a watermark in this range. The transform domain methods are more
popular because watermark embedding is more robust in this domain as compared
to spatial domain. Although, by this method some resistance are loosed so that
embedding in higher frequencies means decrease of watermark resistance, the opti-
mal algorithm can be obtained through a compromise between transparency and
resistance. Discrete Cosine Transform (DCT; Bi, Zhang, & Li, 2011;Halluretal.,
2015;Q.Zhangetal.,2012), Discrete Fourier Transform (DFT; Liu & Zhao, 2010)
and Discrete Wavelet Transform (DWT; Hallur et al., 2015;Huetal.,2011;Srivas-
tava et al., 2015;Yangetal.,2012), Discrete Hadamard Transform (DHT; Sarker &
Khan, 2013), and wide spectrum-based methods (WST; Nagarjuna & Ranjeet, 2013)
are used in frequency domain watermarking. Figure 2 illustrates frequency domain
methods. In these techniques, a watermark is distributed throughout original data
domain. The DCT-based watermarking systems are robust against attacks such as
lossy compression including JPEG attack and some geometric attacks like cutting. In
264 M. GHAZVINI ET AL.
Figure . Related works in frequency domain watermarking.
these systems, by using DCT, original image is decomposed into dierent frequency
bands. Then a watermark is embedded in middle frequency bands of the image. A
DCT-basedwatermarkingmethodisproposedin(Q.Zhangetal.,2012)thatitis
robust against JPEG attack. Watermarking in DWT domain has many advantages,
its adaptation to HVS is one of these advantages. This technique increases water-
mark resistance, without any damage in quality of original image (Wang & Lin,
2004). Although, in frequency domain watermarking, there are some methods that
use combinations of previous transform methods, and one of optimization methods
like Genetic Algorithm (GA), Singular Value Decomposition (SVD), Bee Colony
(BC), and so forth. Authors of (Anju, 2013;Deb,Al-Seraj,Hoque,&Sarkar,2012;
Huai-bin, Hong-liang, Chun-dong, & Shao-ming, 2010; Kasmani & Naghsh-Nilchi,
2008) used combination of DWT and DCT, (Sikander, Ishtiaq, Jaar, Tariq, &
Mirza, 2010;Wang,Peng,&Shi,2011) used DCT and GA for optimization and in
(Ramanjaneyulu & Rajarajeswari, 2012) DWT and GA were used. Combinations of
DCT, DWT and GA were used (Dubolia, Singh, Bhadoria, & Gupta, 2011;Mingzhi,
Yan , Yaj i an , & Mi n, 2013), too. In addition researchers (Ansari, Devanalamath,
Manikantan, & Ramachandran, 2012;Gunjal&Mali,2015;Halluretal.,2015;
Jane, Ilk, & Elbasi, 2013; Kansal, Singh, & Kranthi, 2012; Loukhaoukha, Refaey,
Zebbiche, & Nabti, 2015;Sharma&Jain,2014) proposed some SVD-based methods
for optimization. In this paper, a watermarking technique based on a combination
JOURNAL OF APPLIED SECURITY RESEARCH 265
of DWT and DCT is presented. First, in embedding step of watermark, the original
image is decomposed by applying two levels of DWT and then DCT is applied on
itandnallythewatermarkisembeddedonmiddlebandcoecientsofDCT.In
this technique, two random sequences are created to embed binary bits by which
watermark bits are embedded and then inverse DCT and DWT are exerted on the
watermarked image. In extraction operation, also, DWT and DCT are applied as
embedding operations and then correlation between middle band coecients of
DCT and predened watermark design is calculated to determine binary bits. We
use genetic algorithm to nd optimal embedding parameters including Gain factor
and frequency bands of DCT coecients that it improves robustness and quality of
watermarked image.
The basic stages in watermarking
Two basic stages in all watermarking methods are embedding and extracting as
shown in Figure 3. In the embedding stage, watermark is embedded within original
image. Embedding data in original image is performed in terms of required appli-
cation in spatial or transform domain, as robust or fragile, visible or invisible, blind
or semiblind or nonblind and reversible or irreversible.
Extracting stage is reverse of embedding stage. In this stage, the recipient by using
intended extraction algorithm extracts watermark that was embedded in original
image.
The proposed watermarking technique using genetic algorithm
In this paper, an image watermarking technique has been presented that a mul-
tipurpose genetic algorithm is used. In this optimal technique, the values of two
important criteria in watermarking, PSNR and NCC, are considered and by using
genetic algorithm steps and Pareto function, the most desirable and optimal water-
marking place and coecients will be chosen. This will balance between resistance
and imperceptibility of a watermarking system eectively. In genetic algorithm, all
steps and usual procedures have been observed and Find_Pareto_Front has been
Figure . Embedding and extracting watermark in digital watermarking, (A): watermark embedding
stage, (B): watermark extracting stage.
266 M. GHAZVINI ET AL.
used as metric function to nd the best member in the society. In the proposed tech-
nique, DWT and DCT are applied on image. For DWT, there are three frequency
band sets: low band frequency, middle band frequency, and high band frequency.
In this paper, middle band frequency has been used. If, watermark is embedded in
low band frequency, the ability of imperceptibility is very weak. Also, in high band
frequency, because of compression, JPEG attacks and noise, watermark may be
eliminated, so, the best place for watermark is middle band where there is a balance
between parameters of resistance and imperceptibility (Chu, 2003;Deng&Wang,
2003;Lin&Chen,2000;Wu&Hsieh,2000). In this technique, DWT divides image
into four areas: LL1, LH1, HL1, and HH1. LL1 is approximate coecient. HL1,
LH1, and HH1 are detail coecients (vertical, horizontal, and diagonal, respec-
tively). To obtain the second level coecients of DWT, subband LL1 is decom-
posed more and among its coecients, HL2 is selected as a subband to embed
watermark.
Watermark embedding steps
Figure 4 shows watermark embedding steps and explains details step by step.
Figure . Watermark embedding steps.
JOURNAL OF APPLIED SECURITY RESEARCH 267
Figure . Applying the second level of DWT.
Step 1: Applying DWT on the host image and dividing it in four nonoverlapped
subbands LL1, LH1, HL1, and HH1.
Step 2: Applying the second level of DWT on subband LL1 and dividing it into
four smaller subbands LL2, HL2, LH2, and HH2 according to Figure 5.
Step 3: Dividing subband HL2 in 4 ×4blocks.
Step 4: Applying DCT on each block in subband HL2.
Step 5: Producing two sequences of random numbers called pattern_0 and pat-
tern_1. Inverse pattern_0 bits are used to obtain pattern_1 bit. Lengths
of pattern_0 and pattern_1 equal selected middle band coecients of
DCT, where pattern_0 has been used to embed zero watermark bit and
pattern_1 has been used to embed one watermark bit. It must be noted
that since zero value has not a good result, in this paper, regarding to
the results of previous similar papers, −1 has been used instead of zero,
meaning, in pattern_0 operation, zero has been replaced by −1inorder
to obtaining more desirable results at the time of embedding. In this step,
in fact, watermark bits, rst, are encoded by created sequences which play
the role of keys and then will embed in the suitable coecients of the host
image.
Step 6: We embed pattern_0 and pattern_1 values with some value of gain factor
and middle band coecients of DCT according to the following formula:
X=X+α∗patt ern_0if wat er mark bit =−1
X+α∗patt ern_1 if wat er mark bit =1
Where αis Gain factor value, X is selected middle band coecients of
DCT before embedding, and Xis selected middle band coecients after
embedding.
Step 7: Applying inverse DCT on each 4 ×4 block after adding pattern_0 and
pattern_1 in the selected middle band coecients.
Step 8: Applying inverse DWT on the watermarked image.
Watermark extraction steps
The scheme of watermark extraction operation is shown in Figure 6 and we explain
its details step by step.
268 M. GHAZVINI ET AL.
Figure . Watermark extraction steps.
Step 1: Applying DWT on cover image and dividing it into four nonoverlapped
subbands: LL1, HL1, LH1, and HH1.
Step 2: Applying the second level of DWT on subband LL1 and dividing it into
smaller subbands: LL2, HL2, LH2, and HH2 according to Figure 5.
Step 3: Dividing subband HL2 in 4 ×4blocks.
Step 4: Applying DCT ON each block of subband HL2.
Step 5: In this step, sequences which were created in embedding step, as keys to
watermark bits, are reused. In fact, this kind of cryptography which has
been used in this technique is symmetric, meaning the same key is used
in the embedding and extraction operation.
Step 6: For each block in subband HL2, Correlation between middle band coef-
cients and the produced random number sequences is calculated. If the
Correlation with pattern_0 is more, then the extracted watermark bit is
considered zero and otherwise, the extracted watermark bit is considered
one.
Step 7: The extracted watermark image is reconstructed according to watermark
bits.
Optimize watermarking using genetic algorithm
One of the problems in watermarking is optimization. In the main concepts
of watermarking, there are three opposite attributes which make watermarking
JOURNAL OF APPLIED SECURITY RESEARCH 269
eective (Swami, 2013): Robustness, imperceptibility, and capacity. By NCC factor,
the level of similarity between the original watermark image and the extracted water-
mark image is calculated. This value indicates watermarking technique robustness.
Imperceptibility is measured by PSNR factor in a watermarking technique. PSNR,
in fact, equals inverse value of the host image distortion. And capacity is value of
bits which can be embedded in the cover image. A good watermarking technique is
a technique in which, all of these attributes together have the best possible states and
results. On the other hand, maximum value of PSNR decreases two other attributes.
Here, to resolve and optimize this technique and have the best state, multipurpose
genetic algorithm is used. In this paper, 1024-bit capacity is considered and 42-db
for PSNR and one for NCC are the best values that are considered, meaning these
two factors should be the best together, we don’t intend one of these parameters.
PSNR is obtained by the following formula (Mingzhi et al., 2013):
PSNR =10∗log 10 ∗255∗255
1/(M∗N)M
x=1N
y=1fi,j−gi,j2(1)
Where M and N are height and width, f(i,j) and g(i,j) are pixels value of cover
imageandtheattackedwatermarkedimage.
NCC value is calculated after extracting the attacked watermark. NCC is calcu-
lated by the following formula (Mingzhi et al., 2013):
NCC =m
i=1n
j=1wi,j−wmeanvi,j−vmean
m
i=1n
j=1wi,j−wmean2m
i=1n
j=1vi,j−vmean2
(2)
Where m and n are height and width, W(i,j) and v(i,j) are pixels value of the ori-
gin and extracted watermark. In this technique, multipurpose genetic algorithm and
Find-Pareto-Front function have been used. The use characteristic of this technique
is that two or more purposes which were dened as main genetic purposes should
have all purposed values of the best state and one of the two purposes is bigger and
better than the other, meaning we will seek the best members in the dened popula-
tion. Applying program for nding such members might take longer, but the result
and its simulation results are much better than the other techniques, even ones using
single or multipurpose genetic algorithms. After the embedding operation, PSNR
of watermarked image is calculated and this image may be attacked by one or more
attacks including JEPG compression, Gaussian lter, cutting, rotation, and so forth.
The watermarked image, after applying attacks is extracted and NCC is calculated
from the origin and extracted watermarks.
Experimental results
The original image of Lena and watermark image are shown in Figure 7. The size
of Lena image is 512 ×512. We can use any other picture with this size, like Pep-
pers image, Barbara image, and Cameraman image. And the size of the watermark
image is 32 ×32. Matlab software version 2013 was used for implementation. The
parameters are: population size of 20, the number of repeat 100 times, contact rates
270 M. GHAZVINI ET AL.
Figure . Watermark image—The origin Image of Lena.
0.8, mutation rates 0.0056. Values 42 dB for PSNR and 1 for NCC are parts of pur-
poses of genetic algorithm.
In this paper, the implementation results of four important attacks with more per-
centageofdamageanddestructionwerecalculated.TheattackerinJPEGattackcan
easily convert the format of watermarked image to JPEG and using a lower quality
factor of JPEG compression (J. C. Lee, 2006).Thisattackisverysimpleandanyone
is able to save JPEG image using lower quality factors, so resistance to JPEG com-
pression is very important in this assessment of robustness. Each algorithm tries
to nd the quality factor that the watermark can still be extracted. Image scaling
is commonly available, for example, by a paint program in Microsoft Windows. In
this attack, attacker downscaled the watermarked image and the watermark will be
extracted. When the algorithm requires same size as the original, a resizing process
may be needed. The size which the algorithm can work well will be the grading point
forthatalgorithm.Meanandmedianltersaresimplefunctionsinimageprocessing
that the resistance of a watermarked algorithm against them depends on where the
watermark information is embedded. Embedding the watermark in low frequency
Figure . The results of applying attacks on the lena, cameraman, barbara and peppers images.
JOURNAL OF APPLIED SECURITY RESEARCH 271
will remain relatively resistant to such lter attacks. The results of applying attacks
on the Lena, Cameraman, Barbara and Peppers images are shown in Figure 8.The
mentioned technique were very robust against these attacks, it is surely very robust
against other attacks and give the better results.
Conclusion
In this article, a robust watermarking technique based on combining DWT and
DCT is presented and to optimize it, generic algorithm is used. In this technique,
watermark embedding operation is performed on middle band HL2 which divided
it into 4 ×4 blocks and DWT is applied on it. To embed watermark bits, rst, its bits
are encoded by using two random sequences pattern_0 and pattern_1 and then are
embedded. In extraction operation, also, the same things are done and nally, cor-
relation between predetermined design and middle band coecient in DCT blocks
is calculated; According to this correlation, if the correlation between pattern_0 and
middle band coecients of DCT block is more, the extracted watermark bit is con-
sidered zero and otherwise it is considered one. In this technique, to optimize Gain
Factorvalueandtheselectedmiddleband,geneticalgorithmisusedbywhichthe
best and most suitable place for embedding watermark is determined. Optimized
values of gain factor and selected-middle-band in comparison with the other tech-
niques are given in Tab l e 1. Multipurpose genetic algorithm and Find_ Pareto_ Front
function are used to obtain the best results. PSNR and NCC values are measured in
the represented technique on Lena image and binary watermark image 32 ×32 and
eventually were compared with the other techniques and its superiority to them is
approved. Comparison of NCC value after applying attacks in the mentioned tech-
niques in comparison with the other techniques are shown in Tab l e 2 and Table 3 .
Tab le . Optimized values of gain factor and selected-middle-band on Lena image.
Technique Gain Factor Selected-Middle-Band
(Mingzhi et al., ) [,,,]
The proposed method . [,,,]
Tab le . NCC comparison on the Lena image.
Attacks (Mingzhi et al., ) The proposed method
JPEG Q = .
JPEG Q = .
JPEG Q = . .
JPEG Q = . .
JPEG Q = . .
NOISE ATTACK . .
MEAN FILTERING∗ . .
272 M. GHAZVINI ET AL.
Tab le . Comparison of NCC value after applying attacks in the mentioned techniques.
Type of attacks
(Yuan,Huang, & Liu,
)(PSNR=. dB)
(S.-H. Wang & Lin, )
(PSNR =. dB)
(Mingzhi et al., )
(PSNR =.dB)
The proposed
method
(PSNR = dB)
Median filter ( ×) . . . .
Median filter ( ×) . . . .
JPEG, QF = . . . .
JPEG, QF = . . . .
JPEG, QF = . . . .
JPEG, QF = . . .
JPEG, QF = . .
Tab le . Comparison of improvement levelof NCC value in the mentioned images in comparison with
technique (Mingzhi et al., ).
Image Lena Barbara Peppers Cameraman
Improvement percentage . . . .
References
Anju, R. (2013). Modied Algorithm for Digital Image Watermarking Using Combined DCT and
DWT. International Journal of Information and Computation Technology,3(7), 691–700.
Ansari, R., Devanalamath, M. M., Manikantan, K., & Ramachandran, S. (2012). Robust digital
image watermarking algorithm in DWT-DFT-SVD domain for color images. Paper presented
at the Communication, Information & Computing Technology (ICCICT), 2012 International
Conference on Communication, Information & Computing Technology (ICCICT), Mumbai,
India.
Bi, H., Zhang, Y., & Li, X. (2011). Video watermarking robust against spatio-temporal attacks.
Journal of Networks,6(6), 932–936.
Chandramouli, R., & Memon, N. (2001). Analysis of LSB based image steganography techniques.
Paper presented at the Image Processing, 2001. Proceedings. 2001 International Conference
on Image Processing, ICIP 2001, Thessaloniki, Greece, 7–10 Oct. 2001, Vol. 3, pp. 1019–1022.
Checcacci, N., Barni, M., Bartolini, F., & Basagni, S. (2000). Robust video watermarking for wireless
multimedia communications. Paper presented at the Wireless Communications and Network-
ing Confernce, 2000. WCNC. (Vol. 3, pp. 1530–1535). IEEE. Chicago, IL, USA, September
23–28.
Chen, H.-C., Chang, Y.-W., & Hwang, R.-C. (2012). A watermarking technique based on the fre-
quency domain. Journal of Multimedia,7(1), 82–89.
Chen, H.-Y., & Zhu, Y.-S. (2012). A robust watermarking algorithm based on QR factorization
and DCT using quantization index modulation technique. Journal of Zhejiang University SCI-
ENCE C,13(8), 573–584.
Chu, W. C. (2003). DCT-based image watermarking using subsampling. IEEE transactions on
multimedia,5(1), 34–38.
Deb, K., Al-Seraj, M. S., Hoque, M. M., & Sarkar, M. I. H. (2012). Combined DWT-DCT based dig-
ital image watermarking technique for copyright protection. Paper presented at the Electrical &
Computer Engineering (ICECE), 2012 7th International Conference on Electrical and Com-
puter Engineering 20–22 Dec. 2012, Pan Pacic Sonargaon Dhaka, Dhaka-1215, Bangladesh.
Deng, F., & Wang, B. (2003). AnoveltechniqueforrobustimagewatermarkingintheDCTdomain.
Paper presented at the Neural Networks and Signal Processing, 2003. Proceedings of the 2003
International Conference on Neural Networks and Signal Processing, 2003, vol. 2, pp. 1525–
1528. Nanjing. China, December 14–17.
JOURNAL OF APPLIED SECURITY RESEARCH 273
Di Martino, F., & Sessa, S. (2012). Fragile watermarking tamper detection with images compressed
by fuzzy transform. Information Sciences,195, 62–90.
Dubolia, R., Singh, R., Bhadoria, S. S., & Gupta, R. (2011). Digital image watermarking by using
discrete wavelet transform and discrete cosine transform and comparison based on PSNR.Paper
presented at the Communication Systems and Network Technologies (CSNT), 2011 Interna-
tional Conference on Communication Systems and Network Technologies, 3–5 June 2011,
(pp. 593–596). IEEE.
Gunjal, B. L., & Mali, S. N. (2015). MEO based secured, robust, high capacity and perceptual
quality image watermarking in DWT-SVD domain. SpringerPlus,4(1), 126.
Hallur, S. R., Kuri, S., Sudi, G. S., & Kulkarni, D. G. (2015). A robust digital watermarking for
gray scale image. International Journal For Technological Research In Engineering,2(10), 2440–
2443.
Hu, Y., Wang, G., Cao, X., & Yang, L. (2011). A robust paper public-key watermarking based on
contourlet algorithm transform and its application. Journal of Software,6(11), 2247–2254.
Hu, Y., Wang, Z., Liu, H., & Guo, G. (2011). A geometric distortion resilient image watermark
algorithm based on DWT-DFT. Journal of Software,6(9), 1805–1812.
Huai-bin, W., Hong-liang, Y., Chun-dong, W., & Shao-ming, W. (2010). Anewwatermarkingalgo-
rithmbasedonDCTandDWTfusion. Paper presented at the Electrical and Control Engineer-
ing (ICECE), 2010 International Conference on Electrical and Control Engineering 25–27
June 2010, Wuhan, China, pp. 2614–2617.
Huo, Y., He, H., & Chen, F. (2012). Alterable-capacity fragile watermarking scheme with restora-
tion capability. Optics Communications,285(7), 1759–1766.
Husain, F. (2012). A survey of digital watermarking techniques for multimedia data.International
Journal of Electronics and Communication Engineering,2(1), 37–43.
Inoue, H., Miyazaki, A., & Katsura, T. (1999). An image watermarking method based on the
wavelet transform. Paper presented at the Image Processing, 1999. ICIP 99. Proceedings. 1999
International Conference on Image Processing, ICIP ’99, Kobe, Japan, October 24–28, 1999,
pp. 296–300.
Jane, O., Ilk, H. G., & Elbasi, E. (2013). A robust transform domain watermarking technique by
triangular and diagonal factorization. Paper presented at the Telecommunications and Signal
Processing (TSP), 2013 36th International Conference on Telecommunications and Signal
Processing, TSP 2013, Rome, Italy, 2–4 July, 2013, pp. 867–871.
Kansal, M., Singh, G., & Kranthi, B. (2012). DWT, DCT and SVD based digital image watermark-
ing. Paper presented at the Computing Sciences (ICCS), 2012 International Conference on
Computing Sciences, 14–15 September 2012, Phagwara, Punjab, India, pp. 77–81.
Kasmani, S. A., & Naghsh-Nilchi, A. (2008). A new robust digital image watermarking technique
based on joint DWT-DCT transformation. Paper presented at the Convergence and Hybrid
Information Technology, 2008. ICCIT’08. Third International Conference on Convergence
and Hybrid Information Technology, 11–13 November 2008, Busan, South Korea, vol. 2,
pp. 539–544.
Kaushik, P., & Dua, E. S. (2014). Digital image watermarking using BFO optimized DWT and
DCT & comparison between DWT, DWT+DCT, DWT+DCT+BFO. International Journal
of Recent Research Aspects,1(3), 13–16.
Kumar, M., & Malhotra, M. P. (2014). Digital image watermarking: a review. International Journal
of Recent Research Aspects,2(2), 137–142.
Lee, C. F., Chen, K. N., Chang, C. C., & Tsai, M. C. (2011). A Hierarchical Fragile Watermarking
with VQ Index Recovery. Journal of Multimedia,6(3), 277–284.
Lee, J. C. Analysis of attacks on common watermarking techniques.
Lin, S. D., & Chen, C.-F. (2000). A robust DCT-based watermarking for copyright protection.
IEEE Transactions on Consumer Electronics,46(3), 415–421.
274 M. GHAZVINI ET AL.
Liu, Y., & Zhao, J. (2010). A new video watermarking algorithm based on 1D DFT and Radon
transform. Signal Processing,90(2), 626–639.
Loukhaoukha, K., Refaey, A., Zebbiche, K., & Nabti, M. (2015). On the Security of Robust Image
Watermarking Algorithm based on Discrete Wavelet Transform, Discrete Cosine Transform
and Singular Value Decomposition. International Journal of Applied Mathematics and Infor-
mation Sciences,9(3), 1159–1166.
Mingzhi, C., Yan, L., Yajian, Z., & Min, L. (2013). A combined DWT and DCT watermarking
scheme optimized using genetic algorithm. Journal of Multimedia,8(3), 299–305.
Nagarjuna, P., & Ranjeet, K. (2013). Robust blind digital image watermarking scheme based on
stationary wavelet transform. Paper presented at the Contemporary Computing (IC3), 2013
Sixth International Conference on Contemporary Computing (IC3), 8–10 Aug, Noida, India,
pp. 451–454.
Narang, M., & Vashisth, S. (2013). Digital watermarking using discrete wavelet transform. Inter-
national Journal of Computer Applications,74(20), 34–38.
Qi, X., & Xin, X. (2011). A quantization-based semi-fragile watermarking scheme for image con-
tent authentication. Journal of Visual Communication and Image Representation,22(2), 187–
200.
Ramanjaneyulu, K., & Rajarajeswari, K. (2012). Wavelet-based oblivious image watermarking
scheme using genetic algorithm. IET Image Processing,6(4), 364–373.
Sarker, M. I. H., & Khan, M. I. (2013). An improved blind watermarking method in frequency
domain for image authentication. Paper presented at the Informatics, Electronics & Vision
(ICIEV), 2013 2nd International Conference on Informatics, Electronics & Vision (ICIEV),
17–18 May 2013, Dhaka, Bangladesh, pp. 1–5.
Sharma, C., & Prashar, D. (2012). DWT based robust technique of watermarking applied on digi-
tal images. International Journal of Soft Computing and Engineering (IJSCE) ISSN, 2231-2307.
Sharma, P., & Jain, T. (2014). Robust digital watermarking for coloured images using SVD and
DWT technique. Paper presented at the Advance Computing Conference (IACC), 2014 IEEE
International Advance Computing Conference (IACC), 21–22 Feb. 2014, Gurgaon, India,
pp. 1024–1027.
Sikander, B., Ishtiaq, M., Jaar, M. A., Tariq, M., & Mirza, A. M. (2010). Adaptive digital water-
marking of images using genetic algorithm. Paper presented at the Information Science and
Applications (ICISA), 2010 International Conference on Information Science and Applica-
tions 21–23 April 2010, Seoul, Korea (South), pp. 1–8.
Srivastava, M., Srivastava, H., & Srivastava, M. (2015). A robust watermarking using DWT. Inter-
national Journal of Electrical and Electronic Engineering & Telecommunications,1(2), 229–240.
Swami, S. (2013). Digital image watermarking using 3 level discrete wavelet transform.Conference
on Advances in Communication and Control Systems 2013 (CAC2S 2013), Dehradun, India,
pp. 129–133.
Waleed, J., Jun, H. D., Hameed, S., & Kamil, M. (2015). Optimal positions selection for watermark
inclusion based on a nature inspired algorithm. International Journal of Signal Processing,
Image Processing and Pattern Recognition,8(1), 147–160.
Wang, J., Peng, H., & Shi, P. (2011). An optimal image watermarking approach based on a multi-
objective genetic algorithm. Information Sciences,181(24), 5501–5514.
Wang, S. H., & Lin, Y. P. (2004). Wavelet tree quantization for copyright protection watermarking.
IEEE Transactions on Image Processing,13(2), 154–165.
Wolfgang, R. B., & Delp, E. J. (1996). Awatermarkfordigitalimages. Paper presented at the Image
Processing, 3rd IEEE International Conference on Image Processing, Lausanne, Switzerland,
19–19 Sept. 1996, Vol. 3, pp. 219–222.
Wolfgang, R. B., Podilchuk, C., & Delp, E. J. (1999). Perceptual watermarks for digital images and
video. Proceedings of the IEEE,87(7), 1108–1126.
JOURNAL OF APPLIED SECURITY RESEARCH 275
Wu, C. F., & Hsieh, W. S. (2000). Digital watermarking using zerotree of DCT. IEEE Transactions
on Consumer Electronics,46(1), 87–94.
Xiao, D., & Shih, F. Y. (2012). An improved hierarchical fragile watermarking scheme using
chaotic sequence sorting and subblock post-processing. Optics Communications,285(10),
2596–2606.
Yang, Y., Lei, M., Liu, H., Zhou, Y., & Luo, Q. (2012). A novel robust zero-watermarking scheme
based on discrete wavelet transform. Journal of Multimedia,7(4), 303–308.
Yuan, Y., Huang, D., & Liu, D. (2006). An integer wavelet based multiple logo-watermarking scheme.
Paper presented at the Computer and Computational Sciences, 2006. IMSCCS’06. First Inter-
national Multi-Symposiums on Computer and Computational Sciences, 2006. IMSCCS ’06.
In 20–24 June, Hangzhou, Zhejiang, China, Vol. 2, pp. 175–179.
Zhang, Q., Li, Y., & Wei, X. (2012). An improved robust and adaptive watermarking algorithm
based on DCT. Journal of Applied Research and Technology,10(3), 405–415.
Zhang, X., Zhu, G., Wang, W., Wang, M., & Ma, S. (2012). New public-key cryptosystem based
on two-dimension DLP. Journal of Computers,7(1), 169–178.
Zhou, X., Wang, S., Xiong, S., & Yu, J. (2010). Attack model and performance evaluation of text
digital watermarking. Journal of Computers,5(12), 1933–1941.