Protecting Digital-Image Copyrights: A Robust and Blind Watermarking

Scheme

Dr.M.A. Dorairangaswamy
E-mail: durairangasamyphd@gmail.com

Abstract digital contents. The copyright provisions

established by the owners of the intellectual assets,
Currently, the rapid escalation of the Internet has such as license terms and usage agreements are
made the issue of protecting copyrights of digital honored by DRM [2].
contents very much important. The Digital Rights DRM systems aid in securing valuable digital
Management (DRM) system protects the high-value properties and restricting their distribution and
digital assets and controls the distribution and usage utilization. It is essential that a relentless content
of those digital assets. The protection of copyrights protection against unauthorized access to the digital
of digital images is assured by the watermarking content is offered by a DRM system while only the
technologies. Digital watermarking guards the ones with the appropriate authorization is allowed to
copyrights of digital images by hiding the secret or access. In order to administer usage rights for
personal information in them. In this paper, we have various types of digital content (for instance: music
proposed a robust and blind watermarking scheme files, video streams, digital books, images) across
for protecting the copyrights of digital images. The different platforms (for instance: PCs, laptops,
proposed watermarking scheme protects the PDAs, mobile phones), it necessitates adequate
copyrights of the image by invisibly embedding a robustness [3]. The two components of DRM are 1)
binary watermark image into the host image. The A collection of technologies such as encryption,
proposed scheme embeds binary watermark pixels copy control, digital watermarking, fingerprinting,
into host image pixels to construct the watermarked traitor tracing, authentication, integrity checking,
image. As the proposed watermarking scheme is access control, key management and more 2) Further
blind, we do not require the original image or any of technologies employed to convey copyright
its characteristics for watermark extraction. The permissions in ‘rights expression languages’ and
robustness of the proposed watermarking scheme additional kinds of metadata that make a DRM
has been demonstrated through the attacks presented policy machine-readable[4].
in the experimental results. Besides all the aforementioned technologies, an
important part is played in DRM by copyright
1. Introduction protection. Evading disputes that arise out of
ownership claims through buying and selling digital
The most recent advancements in digital documents is the principle aim of the solution
technologies accompanied by more and more presented by copyright protection mechanisms. To
interconnected high-speed networks and the authenticate the ownership of document prior to its
diminution in costs of high-performance digital sale and a similarly to attest authentic buyers, a fool-
devices at recent times have led to the emergence of proof mechanism is necessitated by the problem [5].
digital content distribution as one of the swiftly The susceptibility of illegitimate replication,
flourishing fields. These progresses in digital content alterations and distribution of digital images is
distribution have created enormous projections for revealed by the emergence of image processing
business content suppliers, which nevertheless poses tools. The ubiquitous internet has made the
threats due to easy possibility of illegal copying and protection of digital images a chief concern. The
distribution of the digital data. Therefore, business media content owners have shown escalating
content suppliers demand technologies supported by concern over probable loss of revenue resulting from
the legislation to secure digital content from digital media piracy [6]. Even as identifying the
illegitimate utilization [1]. The aforementioned issue genuine requirement of the copyright owners, along
can be solved by one of the probable solutions, with the content industry, to protect their copyrights
Digital Rights Management (DRM). DRM, a in spite of technological progression, digital rights
collection of technologies that enforces the management systems will offend the copyright
utilization of digital contents in accordance with the balance amid copyright owners and users.
established privileges aids in the protection of the

978-1-4244-4615-5/09/$25.00 ©2009 IEEE 423

Authorized licensed use limited to: K J Somaiya Inst of Engg and Information Tech. Downloaded on March 25,2010 at 07:23:02 EDT from IEEE Xplore. Restrictions apply.
 Numerous techniques are incorporated in DRM to computer program can however extract the
evade copying or to restrict utilization of a digital embedded data [13]. In general, tolerance from un-
file. The fact that the standard use of which is malicious or malicious attacks like scaling, cropping,
typically authorized is also restricted causing a lossy compression, and so forth is the function of
disagreement in such technology. Alleged forensic robust watermarking. Copyright protection is the
techniques are less controversial than DRM and they principle aim of robust watermarking. In contrast,
aid the copyright holder when unauthorized copies identification of any small change to the original
emerge, to restrict the pirates and prosecute due to digital content is the aim of the fragile watermarking
the fact that the forensic techniques are brought into [14]. There is more than one restriction for a visible
action only when a crime is evident; however the watermark. It signifies the image fidelity and when
alleged forensic techniques are not capable of attacked through direct image processing it is
preventing duplication [7]. Although there are susceptible. There are several researches on
several technological methods to tackle copyright copyright protection of digital images through
piracy, the unavailability of perfect or generally watermarking schemes [16 – 20].
recognized solution has left profuse scope for This research work presents a robust and blind
innovative research. There is an escalation in the watermarking scheme for copyright protection
interest towards digital watermarking, suggested as a against piracy of digital images. This scheme makes
technique for copyright protection or ownership use of a binary image as watermark data for
identification of digital images. protecting the copyrights of the digital image. Here,
The digital watermarking technique can be for every pixel of the host image, a binary watermark
employed to protect the digital images from illegal image pixel is embedded on the basis of the gain
copying and manipulation. A data is embedded into a factor and the generated random matrix. The
multimedia element like image, audio or video embedded binary watermark is extracted from the
through the process of watermarking [8]. For diverse watermarked image using watermark image size and
intentions comprising copyright protection, access the correlation coefficient. Also, as our proposed
control, and broadcast monitoring, the embedded scheme follows blind watermarking, it does not
data can be extracted in the future or distinguish in necessitate the original host image or its desirable
the multimedia element. Image watermarking, video characteristics for extraction. The experimental
watermarking and audio watermarking are the types results portray the robustness of the presented
of the digital watermarking sorted on the basis of the watermarking algorithm.
application. The existing digital watermarking The rest of the paper is organized as follows:
methods primarily aim at image and video copyright Section 2 presents a brief review of some of the
protection [9]. It presents the image with a sense of recent works that employ digital watermarking for
ownership or authenticity by operating as a digital copyright protection of digital images. The proposed
signature. The chief merit of watermarking is the methodology for robust and blind watermarking
indivisibility of the watermark from the content. scheme is presented in Section 3. The experimental
There are several essential traits in a watermark. results are given in Section 4 and conclusions are
Therefore, the watermark is tough to perceive, defies summed up in Section 5.
ordinary distortions, tolerates malevolent attacks,
holds several bits of information, is able to coexist 2. Review of Related Works
with other watermarks, and requires modest
computation for insertion or identification [10]. For copyright protection of digital images our
There are numerous ways to classify the work has moved by a number of earlier works
watermarks and watermarking techniques. The available in the literature that make use of digital
watermarking is categorized into Non-blind, Semi- image watermarking. A few of the latest
Blind and Blind schemes based on the requirements considerable researches are briefly illustrated as
for watermark extraction or detection [11], [12]. follows:
Non-blind watermarking schemes detect the An asymmetrical watermarking method for
watermark by using the original image and secret copyright protection that satisfies the zero
keys. Semi-Blind schemes require the secret key(s) knowledge principle with the objective to defeat the
and the watermark bit sequence. However, the blind weaknesses of contemporary symmetric
schemes employ only the secret key(s) for extraction. watermarking methods is offered by Jengnan Tzeng
There are visible embedded data (watermark) and et al. [16]. The improvement of the watermark space
invisible embedded data (watermark). In visible concept of their foregoing symmetric watermarking
watermarking of images a secondary image (the method in their method made their asymmetric
watermark) is embedded in a primary image in such design a robust one. Due to the important trust of
a manner that it is deliberately perceptible to a their watermark on the unique image it is doubtful to
human observer, whereas the embedded data is not get rid of the watermark devoid of visibly deforming
detectable in case of invisible watermarking where a the watermarked image.

424

Authorized licensed use limited to: K J Somaiya Inst of Engg and Information Tech. Downloaded on March 25,2010 at 07:23:02 EDT from IEEE Xplore. Restrictions apply.
 Xu Zhou et al. [18] presented a method that Owing to the copyright protection in opposition to
embeds watermark into the angle parameter of piracy of digital images an invisible and blind
polygonal lines or curves in a contour map after its watermarking scheme is offered by
parameterization. Their scheme guards in opposition M.A.Dorairangaswamy [21]. A binary watermark
to general geometric transform (attacks) due to its image is embedded in a watermarking scheme
geometric nature. The lack of ability of their invisibly into a host image in order to preserve its
algorithm to go up against vertex removal (polygonal copyrights. By means of this approach for every 2x2
line simplification) or addition operation in a robust block of the host image, a watermark pixel was
manner is the weakness of their scheme. This embedded. Since the planned watermarking scheme
problem can be resolved with no difficulty when was blind, the extraction of watermark wants only
original data are concerned; on the other hand the the watermarked image and it doesn’t demand the
safeguard of the original data should be very much original image or any of its characteristics. The
damaged. efficiency of the watermarking scheme was
A two-phase watermarking scheme which illustrated from the outcomes of the experiment.
extracts together the grayscale watermark and the
binary one from the protected images was presented 3. Proposed Methodology for Robust and
by Ming-Chiang Hu et al. [19] so as to accomplish Blind Watermarking Scheme
the copyright protection. To begin with, their scheme
engaged the pixel values of the original image in The proposed robust and blind watermarking
order to build a grayscale watermark image. Then, scheme for copyright protection of digital images is
their scheme aims to get back a binary watermark explained in this section. The proposed digital
image by make use of the just-procured-permuted watermarking scheme is blind and hence it doesn’t
grayscale watermark from the first phase. The result require the original image or any of its characteristics
of their scheme is the lossless embedding i.e. the for the extraction. A binary image is used as
protected images and the original ones are alike watermark data and its pixels are embedded into the
when we look at. It is illustrated by the obtained host image by means of invisible watermarking for
results that the system accomplishes the widespread protecting the copyrights of the host image. The
requirements of image watermarking and is better in steps involved in the watermark embedding and
contrast with the existing system in terms of extraction processes are described in the following
transparency and robustness. subsections.
Shang-Lin Hsieh et al. [20] presented a
watermarking scheme for copyright protection of
3.1 Watermark Embedding Steps
color images. By the scheme the requirement of
imperceptibility and robustness for a sensible
The embedding of the binary watermark image
watermarking scheme. By the experiment it has been
pixels into the host image is presented in this sub-
demonstrated that the resistance of their scheme in
section. The host image’s size should be dyadic
opposition to many attacks for example cropping,
(2nx2n) and a binary image is used as watermark.
scaling, and JPEG compression. Additionally, the
Initially, the binary watermark image is transformed
capability of the scheme to extract unique features
into a vector. Then, a random matrix is generated
from different images, which is a very important
identical to the size of the host image. As the
prerequisite for feature extraction, was established
watermark image chosen is binary, the pixel values
by the only one of its kind identification experiment.
are either 0 or 1. Here, each ‘0’ pixel in the
A digital watermarking technique and proposed to
watermark image is embedded into the host image to
solve some significant problems in the digital world,
obtain the watermarked image. Figure 1 shows the
such as copyright protection, copy protection, and
block diagram of the watermark embedding process.
content authentication via their extensive research
was offered by Lu et al. [15]. They offered a
proficient multipurpose watermarking algorithm
based on the mean-removed vector quantization
(VQ). Ming-Shi Wang and Wei-Che Chen [17]
proposed a digital image copyright protection
scheme based on visual cryptography (VC) and
singular value decomposition (SVD) techniques. The
SVD is applied at first in their image to a host image
to build a master share. In the scheme, the secret
image is embedded with no alteration of the host
image. It is not essential to employ the original host Figure 1. Watermark embedding process
image and the help of computers to extract the out of
sight secret image in addition.

425

Authorized licensed use limited to: K J Somaiya Inst of Engg and Information Tech. Downloaded on March 25,2010 at 07:23:02 EDT from IEEE Xplore. Restrictions apply.
 The steps involved in the watermark embedding
are as follows:
Input: Host image ( H I ), Binary watermark image
( WI ), gain factor ( α ).
Output: Watermarked image ( WM I )
1. The watermark image is represented as a matrix
of size n × n , which is converted into a vector. The
pixels in the vector are represented as follows:
WI = [ P1 P2 P3 Pn ]
2. A random matrix RM is generated with the
Figure 2. Watermark extraction process
size of host image with the aid of a seed value as
follows:
The steps involved in the extraction process are as
RM = round (2 ∗ rand ( H I − 0.5)) follows:
3. The binary watermark image is embedded into Input: Watermarked image ( WM I ), Size of
the host image. In this, all the ‘0’ pixels in the
watermark image ( W I )
watermark image are embedded into the host image
based on the gain factor and the generated random Output: Watermark image ( WI )
matrix. 1. Construct a unit vector with the size of the
if [ Pi ]WI == 0 then watermark image.
H I = H I + α ∗ RM WI = [ P1 P2 P3 Pn ] ; Pi = 1
end if 2. A random matrix RM is generated with the
4. Then the watermark embedded pixels are size of the watermarked image. The above process is
mapped back to its original position to obtain the represented as follows.
watermarked image ( WM I ). RM = round (2 ∗ rand (WM I − 0.5))
3. Calculate the correlation coefficient CCF
3.2 Watermark Extraction steps
between the generated random matrix and the
The extraction of binary watermark image from watermarked image matrix. The equation to calculate
the watermarked image is explained in this sub- the CCF is as follows:
section. As the proposed scheme is blind, the & & ( Amn − A)( B mn − B )
extraction process only requires the watermarked m n
r=
image, size of watermark image and correlation % "% "
coefficient. Initially, a vector is constructed for the # & & ( Amn − A) 2 # & & ( B mn − B ) 2
# #
size of watermark image with all the values $ m n !$ m n !
initialized to ‘1’. Then a random matrix is generated Where,
with the size of the watermarked image. r → correlatio n coefficien t
Subsequently, we find the correlation coefficient
Amn → watermarked image matrix WM I
between the generated random matrix and the
watermarked image. The above process is iterated Bmn → Random matrix RM
for the size of the watermark image. Afterwards, find
A → Mean value of A
the mean value of the correlation coefficients and the
resultant is considered as a threshold value. The B → Mean value of B
threshold and the coefficients are utilized in the 4. Repeat step 3 for the size of watermark image.
extraction of the watermark. Finally, the extracted 5. Calculate the mean value of all the correlation
watermark pixel values are placed in the initialized coefficient values and the mean value is considered
vector in order to obtain the watermark image. as a threshold value.
Figure 2 shows the block diagram of the watermark n CCF
(i )
extraction process. CCF = &
i =1
n
6. If the value of the correlation coefficient is
greater than the threshold value, the values of the
watermark image vector is changed to zero.

426

Authorized licensed use limited to: K J Somaiya Inst of Engg and Information Tech. Downloaded on March 25,2010 at 07:23:02 EDT from IEEE Xplore. Restrictions apply.
 if CCFi > Thresh then
[WI ( i ) ] == 0
else
[W I (i ) ] == 1
end if
7. Finally, the vector WI formed is converted into
matrix with watermark image’s size to obtain the
watermark image.

4. Experimental Results
The experimental results of the presented
watermarking scheme are presented in this section.
The proposed watermarking scheme is programmed Figure 4. Attacked images and the parameters of the
in Matlab (Matlab7.4) and tested with attacked corresponding attacks
images. The binary watermark images are embedded
into the host images effectively. Subsequently, the
embedded watermarks are extracted from the
watermarked images efficiently. The watermarked
images have good Peak Signal to Noise Ratio
(PSNR) and good visual quality. Figure 3 shows the
host image, watermark and watermarked image
along with the PSNR values.

Figure 3. (a) Host Image (b) Watermark image (c)

Watermarked Image with PSNR value (dB) Figure 5. Extracted watermarked images and the
correlation coefficient values
The robustness of the presented watermarking
scheme is demonstrated by means of various attacks. 5. Conclusion
The chosen attacks were Gaussian blur, Intensity Recently, there has been an unbelievable
adjustment, median filtering and Gaussian noise. The development of electronic commerce applications
watermark images have been extracted from the and online services nevertheless, the service
attacked watermarked images effectively. The providers are anxious about the unhindered
correlation coefficient is calculated between the duplication and distribution of copyrighted material.
extracted watermark image and the original The importance of the security of images has risen
watermark image. The experimental results owing to the ubiquitous internet. In this paper, we
demonstrate that the correlation coefficient value for have presented a robust and blind digital
all the attacks is above 0.5. The correlation values watermarking scheme for copyright protection of
assure that we retrieve a reliable watermark and the images. A binary image is utilized as the watermark.
experimental results illustrate that the proposed The binary watermark image’s pixels are embedded
watermarking scheme is robust. Figure 4 and 5 into the host image. Subsequently, the scheme
shows the attacked watermarked images and the discussed is used to extract the watermark image
extracted watermark images along with their from the watermarked image. The visual quality of
the watermarked images has been good and also
correlation coefficient values respectively.
good PSNR values are obtained. The robustness of
the presented watermarking algorithm has been
demonstrated with the help of various attacks.

427

Authorized licensed use limited to: K J Somaiya Inst of Engg and Information Tech. Downloaded on March 25,2010 at 07:23:02 EDT from IEEE Xplore. Restrictions apply.
 6. References
[13] Yeung, M. & Minzter, F., “An Invisible
Watermarking technique for image verification,”
[1] Claudine Conrado, Milan Petkovic, Michiel van der
Veen and Wytse van der Velde "Controlled Sharing of Proceeding on the IEEE International Conference on
Personal Content Using Digital Rights Management", Image Processing, pp: 680-683, 1997.
Journal of Research and Practice in Information
[14] Shaowei Weng, Yao Zhao and Jeng-Shyang Pan, “A
Technology, Vol. 38, No. 1, February 2006.
Novel Reversible Data Hiding Scheme,” International
[2] L. C. Anderson, J. B. Lotspiech, “Rights Management Journal of Innovative Computing, Information and
Control, Vol. 4, No. 2, pp. 351-358, 2008.
and Security in the Electronic Library,” Bulletin of the
American Society for Information Science, Vol. 22, No.1,
pp.21-3, October-November 1995. [15] Zhe-Ming Lu, Wei-Min Zheng, Jeng-Shyang Pan and
Zhen Sun, "Multipurpose Image Watermarking Method
[3] Qiong Liu, Reihaneh Safavi-Naini and Nicholas Paul Based on Mean-removed Vector Quantization", Journal of
Information Assurance and Security, Vol. 1, pp. 33-42,
Sheppard, "Digital Rights Management for Content
Distribution", Proceedings of the Australasian information 2006.
security workshop conference on ACSW frontiers,
Adelaide, Australia, Vol. 21, pp. 49 - 58, 2003 [16] Jengnan Tzeng, Wen-Liang Hwang, and I-Liang
Chern, "An Asymmetric Subspace Watermarking Method
[4] Ian Kerr, "Hacking@privacy: Why We Need for Copyright Protection", IEEE Transactions on Signal
Protection from the Technologies That Protect Copyright", Processing, Vol. 53, No. 2, February 2005
In proc. of Conference on privacy and identity, 2007.
[17] Ming-Shi Wang and Wei-Che Chen, "Digital image
[5] A. Piva, F. Bartolini, and M. Barni, “Managing copyright protection scheme based on visual cryptography
copyright in open networks,” IEEE Trans. Internet and singular value decomposition", Optical Engineering,
Computing, vol. 6, no. 3, pp. 18–26, May–Jun. 2002. Vol. 46, No. 6, 2007.

[6] Shang-Lin Hsieh, Lung-Yao Hsu, and I-Ju Tsai, “A [18] Xu Zhou, Yu Ren, and Xuezeng Pan, "Watermark
Copyright Protection Scheme for Color Images using Embedded in Polygonal Line for Copyright Protection of
Secret Sharing and Wavelet Transform”, proceedings of Contour Map", IJCSNS International Journal of Computer
Science and Network Security, Vol.6 No.7B, pp. 202-205,
World Academy of Science, Engineering And Technology,
Vol. 10, December 2005. July 2006.

[7] Hans Georg Schaathun, "On [19] Ming-Chiang Hu, Der-Chyuan Lou and Ming-Chang
Chang, "Dual-wrapped digital watermarking scheme for
watermarking/fingerprinting for copyright protection",
image copyright protection," Computers & Security, Vol.
First International Conference on Innovative Computing,
Information and Control, ICICIC '06, Beijing, Vol. 3, pp. 26, No. 4, pp. 319-330,2007.
50-53, Aug. 30 2006-Sept. 1 2006.
[20] Shang-Lin Hsieh, I-Ju Tsai, Bin-Yuan Huang and Jh-
Jie Jian, "Protecting Copyrights of Color Images using a
[8] Authors Emir Ganic, Ahmet M. Eskicioglu, "Robust
DWT-SVD domain image watermarking: embedding data Watermarking Scheme Based on Secret Sharing and
in all frequencies", International Multimedia Conference, Wavelet Transform", Journal Of Multimedia, vol. 3, no. 4,
October 2008.
Magdeburg, Germany, pp. 166 - 174, 2004.

[9] Xiang-Yang Wang and Hong Zhao, "A Novel [21] M.A.Dorairangaswamy, "A Novel Invisible and Blind
Synchronization Invariant Audio Watermarking Scheme Watermarking Scheme For Copyright Protection of Digital
Based on DWT and DCT", IEEE Transactions On Signal Images ", Vol.:9, No. 4, pp. 71-78, 2009.
Processing, Vol. 54, No. 12, December 2006.

[10] Miller, M.; Cox, I.J.; Linnartz, J.P.M.G.; Kalker, T.,

“A review of watermarking principles and practices,” In
Digital Signal Processing in Multimedia Systems, Edit. K.
K. Parhi and T. Nishitani, Marcell Dekker Inc., pp. 461-
485, 1999.

[11] P. Tao and A. M. Eskicioglu, “A Robust Multiple

Watermarking Scheme in the DWT Domain,” Optics East
2004 Symposium, Internet Multimedia Management
Systems V Conference, Philadelphia, PA, pp. 133-144,
October 25-28, 2004.

[12] Ersin Elbasi and Ahmet M. Eskicioglu, “A Semi-

Blind Watermarking Scheme for Color Images Using a
Tree Structure," in proc. of IEEE Sarnoff Symposium,
March, 2006.

428

Authorized licensed use limited to: K J Somaiya Inst of Engg and Information Tech. Downloaded on March 25,2010 at 07:23:02 EDT from IEEE Xplore. Restrictions apply.