See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324985989

GRAYSCALE IMAGE WATERMARK DETECTION

Article  in  INTERNATIONAL JOURNAL OF COMPUTERS & TECHNOLOGY · February 2004

DOI: 10.24297/ijct.v3i1b.2748

CITATIONS READS

0 356

2 authors, including:

Randeep Kaur
ct university
4 PUBLICATIONS   0 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

its seperate from project View project

All content following this page was uploaded by Randeep Kaur on 21 July 2018.

The user has requested enhancement of the downloaded file.

Council for Innovative Research International Journal of Computers & Technology
www.ijctonline.com ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

GRAYSCALE IMAGE WATERMARK DETECTION

Randeep Kaur Kamaljit Kaur Dhillon
Assistant Professor Assistant Professor
LCET, Katani kalan GNDEC, Ludhiana

ABSTRACT
A Digital watermarking is a technique that provides a solution 1.1 Applications of watermarking
to the longstanding problems faced with copyrighting digital 1.1.1 Copyright protection
data. Digital watermarks are pieces of information added to The objective is to embed information about the source/owner
digital data (audio, video, or still images) that can be detected or of the digital media in order to prevent other parties from
extracted later to make an assertion about the data. This claiming the ownership of the media.
information can be textual data about the author, its copyright,
etc; or it can be an image itself. Watermarking Based on DCT
Coefficient Modulation technique embeds the watermark in the
1.1.2 Fingerprinting
DCT domain to increase the robustness of the watermarking The objective of fingerprinting is to convey information about
scheme.DCT based watermarking is an example of frequency the recipient of the digital media (rather than the owner) in
domain watermarking. The objective of this research work is to order to identify every single distributed copy of the media.
implement DCT based watermarking technique on gray scale This concept is very similar to serial numbers of software
image. The study focuses on evaluating the robustness of products.
watermarked image after having three different attacks on
watermarked image and extraction of watermark from that 1.1.3 Copy protection
particular image. To compare the DCT based watermarking Watermarking can be used to control data copying devices and
with LSB based watermarking and to validate the proposed prevent them from copying the digital media in case the
work & the comparative results of watermarking using DCT watermark embedded in the media indicates that media is copy-
and LSB are also presented. This paper recommends DCT protected.
based technique for achieving robustness in digital image
watermarking. 1.1.4 Image authentication
The objective is to check the authenticity of the digital media.
Keywords This requires the detection of modifications to the data.
DCT; LSB; Poisson; Speckle
1.2 Classifications of watermarking
1. INTRODUCTION A digital watermark is called robust with respect to
transformations if the embedded information may be detected
A large number of digital watermarking schemes have been
reliably from the marked signal, even if degraded by any
studied to protect the intellectual property rights of the owner.
number of transformations. Typical image degradations are
All these schemes implement either visible or invisible
JPEG compression, rotation, cropping, additive noise, and
watermarks. Digital watermarking is the process by which an
quantization. For video content, temporal modifications and
image is coded with an owner‘s watermark using two general
MPEG compression often are added to this list. A digital
approaches: spatial domain and frequency domain. The spatial
watermark is called imperceptible if the watermarked content is
domain methods are to embed the watermark by directly
modifying the pixel values of the original image. The second is perceptually equivalent to the original, unwatermarked content.
In general, it is easy to create robust watermarks—or—
to transform the host image in to its frequency domain
imperceptible watermarks, but the creation of robust—and—
representation and embed the watermark therein. Regardless of
imperceptible watermarks has proven to be quite challenging.
the embedding method, the embedding technique must satisfy
several requirements such as the watermarked image should There are four factors those are commonly used to determine
quality of watermarking scheme. They are robustness,
retain as closely as possible the quality of the original image.
The watermark should be robust to various types of attacks. imperceptibility, capacity, and blindness.

1.2.1 Robustness
Watermark should be difficult to remove or destroy. Robust is a
measure of immunity of watermark against attempts to image
modification and manipulation like compression, filtering,
rotation, collision attacks, noise etc.

1.2.2 Imperceptibility
Means quality of host image should not be destroyed by
presence of watermark.

1.2.3 Capacity

100 | P a g e www.cirworld.com
Council for Innovative Research International Journal of Computers & Technology
www.ijctonline.com ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

It includes techniques that make it possible to embed majority watermarking scheme which provides higher resistance to
of information. image processing attacks mainly JPEG compression. In our
approach the watermark is embedded in the mid frequency band
of the DCT blocks only in the sub band which is carrying low
1.2.4 Blind Watermarking frequency components and the high frequency sub band
Extraction of watermark from watermarked image without components remain untouched.
original image is preferred because sometimes it‘s impossible to
avail original image. In [2] A_New_DCT-Based Watermarking Method for
Copyright Protection of Digital Audio paper proposes a new
1.3 Discrete Cosine Transform audio watermarking method based on Discrete Cosine
A discrete cosine transform (DCT) expresses a sequence of Transformation (DCT) for copyright protection. In our
finitely many data points in terms of a sum of cosine functions proposed watermarking method, the original audio is
oscillating at different frequencies. DCTs are important to transformed into DCT domain. The absolute values of DCT
numerous applications in science and engineering, from lossy coefficients are divided into an arbitrary number of segments
compression of audio and images (where small high-frequency and the energy of each segment is calculated. Watermarks are
components can be discarded), to spectral methods for the then embedded into the selected peaks of the highest energy
numerical solution of partial differential equations. The use of segment. Watermarks are extracted by performing the inverse
cosine rather than sine functions is critical in these applications: operation of watermark embedding process.
for compression, it turns out that cosine functions are much
more efficient (as explained below, fewer are needed to In [3] A robust DCT-based watermarking for copyright
approximate a typical signal), whereas for differential equations protection presents A novel technique for embedding
the cosines express a particular choice of boundary conditions. watermarks into a host image in the frequency domain is
In particular, a DCT is a Fourier-related transform similar to the proposed. Unlike the traditional techniques, the method
DFT, but using only real numbers. DCTs are equivalent to addresses that the watermark is embedded at low frequency.
DFTs of roughly twice the length, operating on real data with The weighted correction is also used to improve the
even symmetry (since the Fourier transform of a real and even imperceptibility of the watermark. Moreover the watermark is
function is real and even), where in some variants the input self-extractable and the algorithm is simple. Experimental
and/or output data are shifted by half a sample. There are eight results demonstrate that the watermark is robust to various
standard DCT variants, of which four are common. The most attacks.
common variant of discrete cosine transform is the type-II
DCT, which is often called simply "the DCT"; its inverse, the In [6] Novel DCT based watermarking scheme for digital
type-III DCT, is correspondingly often called simply "the images‖ presents There is an ever growing interest in copyright
inverse DCT" or "the IDCT". Two related transforms are the protection of multimedia content, thus digital watermarking
discrete sine transforms (DST), which is equivalent to a DFT of techniques are widely practiced. Due to the internet
real and odd functions, and the modified discrete transform connectivity and digital libraries the research interest of
(MDCT), which is based on a DCT of overlapping data. protecting digital content watermarking is extensively
researched. In this paper we present a novel watermark
1.4 Applications of DCT generation scheme based on the histogram of the image and
The DCT, and in particular the DCT-II, is often used in signal apply it to the original image in the transform (DCT) domain.
and image processing, especially for lossy data compression,
most of the signal information tends to be concentrated in a few In [5] Multiple binary images watermarking in spatial and
low-frequency components of the DCT. A related transform, frequency domains proposes a scheme using which more data
the modified discrete cosine transform, or MDCT is used in can be inserted in to an image in different domains using
AAC, WMA, and MP3 audio compression. DCTs are also different techniques. This increases embedding capacity.
widely employed in solving partial differential equations by
spectral methods, where the different variants of the DCT In [4] An overview of transform domain robust digital image
correspond to slightly different even/odd boundary conditions at watermarking algorithms is to provide complete overview of
the two ends of the array. The DCT is used in JPEG image digital image watermarking, performance evaluation metrics
compression, MJPEG, MPEG, DV, and Theora video and possible attacks. The generalized algorithms are presented
compression. There, the two-dimensional DCT-II of NxN for DWT, CDMA based, DCT-DWT combined approach.
blocks are computed and the results are quantized and entropy
coded. In this case, N is typically 8 and the DCT-II formula is 3. PROPOSED WORK
applied to each row and column of the block. The result is an 8 A discrete cosine transform (DCT) expresses a sequence of
× 8 transform coefficient array in which the (0,0) element (top- finitely many data points in terms of a sum of cosine functions
left) is the DC (zero-frequency) component and entries with oscillating at different frequencies. DCTs are important to
increasing vertical and horizontal index values represent higher numerous applications in science and engineering, from lossy
vertical and horizontal. compression of audio and images (where small high-frequency
components can be discarded), to spectral methods for the
numerical solution of partial differential equations. The use of
2. RELATED WORK cosine rather than sine functions is critical in these applications:
Several techniques for watermarking have been purposed in the for compression, it turns out that cosine functions are much
literature some of them are discussed below: more efficient. whereas for differential equations the cosines
express a particular choice of boundary conditions. The discrete
In [1] An Improved and Robust DCT based Digital Image cosine transform (DCT) helps separate the image into parts (or
watermarking scheme‖ presents a new DCT based additive spectral sub-bands) of differing importance (with respect to the

101 | P a g e www.cirworld.com
Council for Innovative Research International Journal of Computers & Technology
www.ijctonline.com ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

image's visual quality). The DCT is similar to the f(i,j) is the intensity of the pixel in row i and column
discrete Fourier transform: it transforms a signal or image j;
from the spatial domain to the frequency domain F(u,v) is the DCT coefficient in row k1 and column
k2 of the DCT matrix.
For most images, much of the signal energy lies at
low frequencies; these appear in the upper left corner
of the DCT.
Compression is achieved since the lower right values
represent higher frequencies, and are often small -
small enough to be neglected with little visible
distortion.
The DCT input is an 8 by 8 array of integers. This
array contains each pixel's gray scale level;
8 bit pixels have levels from 0 to 255.
3.1 Discrete Cosine Transform
The discrete cosine transforms is a technique for converting a
signal into elementary frequency components. It represents an
image as a sum of sinusoids of varying magnitudes and
3.2 Embedding Watermark
The information to be embedded in gray scale image is called a
frequencies. With an input image, x, the DCT coefficients for
digital watermark. Watermark is embedded in the coefficients
the transformed output image, y, are computed according to Eq.
of DCT transformed image. The important consideration is,
1 shown below. In the equation, x, is the input image having N
what locations are best for embedding watermarking in the
x M pixels, x (m, n) is the intensity of the pixel in row m and
frequency domain to avoid distortion. It is considered to be
column n of the image, and y (u, v) is the DCT coefficient in
better to insert the data in the middle frequency coefficients of
row u and column v of the DCT matrix.
the host image.

3.2.1 Algorithm
Step1. Compute the 2-D DCT of the image to
be watermarked.
Step2. Locate its K largest coefficients, c1, c2… ck, by
magnitude.
where
Step3. Create a watermark by generating a K-element pseudo-
random sequence of numbers, w1, w2… wk, taken from a
Gaussian distribution with µ = 0 and variance is 1.
Step4. Embed the watermark step 3 in to the K largest DCT
coefficients from step 2 using the following equation
ci‘ = ci * (1 + αwi ) 1< I <K
for a specified constant α > 0. Replace the original ci with the
computed ci ‗from equation as above.
Step5. Compute the inverse DCT of the result from step 4.

The image is reconstructed by applying inverse DCT operation 3.3 Attack on watermarked image
according to Eq. 2: PSNR, RMSE & MSE values are evaluated for different
possible attacks: Following are different attacks applied on the
watermarked image.

3.3.1 Salt & Pepper Noise

Salt and pepper noise is a form of noise typically seen on
The popular block-based DCT transform segments an image images. It represents itself as randomly occurring white and
non-overlapping block and applies DCT to each block. These black pixels. An effective noise reduction method for this type
results in giving three frequency sub-bands: low frequency sub- of noise involves the usage of a median filter, morphological
band, mid-frequency sub-ban and high frequency sub-band. filter or a contra harmonic mean filter. Salt and pepper noise
DCT based watermarking is based on two facts. The first fact is creeps into images in situations where quick transients, such as
that much of the signal energy lies at low-frequencies sub-band faulty switching, take place.
which contains the most important visual parts of the image.
The second fact is that high frequency components of the image
are usually removed through compression and noise attacks.
3.3.2 Speckle noise
The watermark is therefore embedded by modifying the Speckle noise is a granular noise that inherently exists in and
coefficients of the middle frequency sub-band so that the degrades the quality of the active radar and synthetic aperture
visibility of the image will not be affected and the watermark radar (SAR) images. Speckle noise in conventional radar results
will not be removed by compression. The basic operation of the from random fluctuations in the return signal from an object
that is no bigger than a single image-processing element. It
DCT is as follows:
increases the mean grey level of a local area.
The input image is N by M;

102 | P a g e www.cirworld.com
Council for Innovative Research International Journal of Computers & Technology
www.ijctonline.com ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

3.3.3 Poisson noise

Poisson noise or shot noise is a type of electronic noise that
occurs when the finite number of particles that carry energy,
such as electrons in an electronic circuit or photons in an optical
device, is small enough to give rise to detectable statistical
fluctuations in a measurement. It is important in electronics,
telecommunications, and fundamental physics. The parameter λ
is not only the mean number of occurrences E [K], but also its
variance σk2 = E [K2]-E[k] 2. Thus, the number of observed
occurrences fluctuates about its mean λ with a standard
deviation σk is root of mean. These fluctuations are denoted as
poisson noise or (particularly in electronics) as shot noise.

3.4 Extraction of watermark

Step1. Compute the 2-D DCT of the attacked watermarked Fig 4.2: Above two images “flower.bmp” are input and
image. watermarked images & below three images are images after
Step2. Extract the K DCT coefficients and denote the three different attacks.
coefficients c1˶, c2˶,……… ck.
Step3.Compute the watermark using wi‘ = ci˶-ci for 1<i<K.
Step4 .Measure the similarity of w i‘ and wi using a metric such
as the correlation coefficient.
Step5.Compare the measured similarity, λ, to a predefined
threshold, T, and make a binary detection decision D=1
indicates that watermark is present with respect to the specified
threshold; D=0 indicates it was not present.

4. Experimental Results
To verify the effectiveness of the proposed method, a series of
experiments were conducted. In these experiments, a set of
original images of size 512 x 512 are used. It has been validated
by applying algorithm on synthetic image. The following
figures show the input image, DCT based watermarked image
and watermarked images after applying Salt&pepper attack,
Speckle attack and the poisson attack.
Fig 4.3: Above two images “Lady.bmp” are input and
watermarked images & below three images are images after
three different attacks.

Fig 4.1: Above two images “boat.bmp” are input and

watermarked images & below three images are images after
three different attacks.
Fig 4.4: Above two images “Lena.bmp” are input and
watermarked images & below three images are images after
three different attacks.

103 | P a g e www.cirworld.com
Council for Innovative Research International Journal of Computers & Technology
www.ijctonline.com ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

Table 4.2 Computational Result and Analysis of results

based on RMSE metric.

Type of No Salt &

Speckle Poisson
Images attack pepper

Boat 3.9 225.8 225.8 225.8

Flower 3.3 225.8 225.7 225.7

Lady 3.4 225.9 225.8 225.8

Fig 4.5: Above two images “Grass.bmp” are input and
watermarked images & below three images are images after
three different attacks. Lena 3.4 225.7 225.7 225.9

Grass 3.9 225.8 225.9 225.7

Fig 4.6: Original Watermark Image

Table 4.3 Computational Result and Analysis of results
based on MSE metric.

Type of No Salt &

Fig 4.7: DCT based Extracted watermarks after applying Images attack pepper Speckle Poisson
Salt & pepper, Speckle & poisson attack respectively.
Boat 13.4 50981.9 50995.1 50968.7
Extracted watermarks shown above represents that Salt &
pepper and poisson noise does not affect the watermark applied Flower 10.9 51002.3 50972.7 50967.6
on the image .So, the DCT based watermarking scheme is
robust and even after applying three different attacks best
Lady 11.5 50986 50967.6 50967.6
quality watermarks extracted in DCT based watermarking
technology. These results have been analyzed based on PSNR,
RMSE & MSE metric. Lena 12.1 50977.8 50981.4 50968.2

Table 4.1 Computational Result and Analysis of results Grass 15.3 50985.5 50991.1 50968.7
based on PSNR metric.

Type of No Salt & Speckle Poisson Table 4.4 Comparative values of three metrics based on
DCT & LSB for Synthetic Image “Grass.png”.
Images attack pepper

Type of Algorithm
Boat 36.9 1.04 1.05 1.058 PSNR RMSE MSE
attack used

Flower 37.6 1.05 1.05 1.06 DCT 36.3 3.9 15.3

NO Attack
Lady 37.5 1.05 1.06 1.07 LSB 57.9 0.3 0.10

DCT 1.05 225.8 50985.5

Lena 37.3 1.1 1.06 1.05 Salt &
pepper
LSB 5.9 129.2 16711.4
Grass 36.3 1.05 1.05 1.07
DCT 1.05 225.9 50991.1
Speckle
LSB 2.8 184.02 33866.5

DCT 1.07 225.7 50968.7

Poisson
LSB 2.9 182.2 33218.3

104 | P a g e www.cirworld.com
 Council for Innovative Research International Journal of Computers & Technology
www.ijctonline.com ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

The performance of proposed algorithm is evaluated based on 6 ACKNOWLEDMENT

three different metrics. In this work results which have been I am extremely grateful to Assistant Professor Akshay
achieved using DCT based watermarking are compared with the Girdhar, Head of the Department I.T.,GNDEC,
LSB based watermarking technique. The following figure LUDHIANA helping in carrying the present work and is
shows the watermarks extracted from the LSB based acknowledged with reverential thanks. I am extremely grateful
watermarking after passing through salt & pepper, speckle & to Ms. Kamaljeet Kaur Assistant Professor GNDEC,
poisson attack. LUDHIANA. Without the wise counsel and able guidance, it
would have been impossible to complete this work in this
manner.

7 REFERENCES
Fig 4.8: LSB based Extracted watermarks after applying [1] Tribhuwan Kumar tiwari and vikas saxena (2010) in ―An
Salt & pepper, Speckle & poisson attack respectively. Improved and Robust DCT based Digital Image watermarking
scheme‖ International Journal of Computer Applications (0975
From the results shown above, the performance of DCT based – 8887) Volume 3 – No.1, June 2010
watermarking can be compared with the LSB based
watermarking technique. Watermarks extracted from the [2] Pranab Kumar Dhar, Mohammad Ibrahim Khan and Saif
proposed algorithm are comparatively better than the Ahmad (2010) in ―A_New_DCT-Based Watermarking Method
watermarks extracted from the LSB based watermarking for Copyright Protection of Digital Audio‖ International journal
technique. of computer science & information Technology (IJCSIT) Vol.2,
No.5, October 2010
5. CONCLUSIONS AND FUTURE WORK [3] Lin, S.D.; Chin-Feng Chen; in ―A robust DCT-based
In this paper we have presented DCT based watermarking watermarking for copyright protection‖
technique for authentication of grayscale image. Watermarks
are also extracted after passing through salt & pepper, speckle [4] Baisa L.Gunjal, R.R. Manthalkar (2011) in ―An overview of
& poisson attack. Based on the results in the previous chapter, I transform domain robust digital image watermarking
conclude that DCT based watermarking give best quality algorithms‖.
extracted watermarks after passing through salt & pepper,
speckle & poisson attack. It shows that DCT based [5] K.Ganesan, Tarun Kumar Gupta (2010) in ―Multiple binary
watermarking is robust technique. Proposed algorithm is also images watermarking in spatial and frequency domains.
compared with the existing watermarking technique. Proposed
algorithm can be applied on color images & the watermarking [6] Neminath Hubballi, Kanyakumari D P, ―Novel DCT based
image can also color image. watermarking scheme for digital Images‖ International Journal
of Recent Trends in Engineering, Vol. 1, No. 1, May 2009

[7] R. Liu and T. Tan, ― A SVD-Based Watermarking Scheme

for Protecting Rightful Ownership,‖ IEEE Transactions on
Multimedia, 4(1), March 2002, pp.121-128.

105 | P a g e www.cirworld.com

View publication stats