I.J.

Modern Education and Computer Science, 2012, 3, 50-56

Published Online April 2012 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijmecs.2012.03.07

Image Watermarking Using 3-Level Discrete

Wavelet Transform (DWT)
Nikita Kashyap
Department of Electronics & Telecommunication Engineering, Shankaracharya Technical Campus, Bhilai, India
Email: nikita2010k@gmail.com

G. R. SINHA
Associate Director, Faculty of Engg & Tech, Shri Shankaracharya Technical Campus, Bhilai, India
Email: drgrsinha@ieee.org

Abstract—We have implemented a robust image watermarking Spatial Domain Watermarking and Frequency Domain
technique for the copyright protection based on 3-level discrete Watermarking.
wavelet transform (DWT). In this technique a multi-bit Compared to spatial domain techniques [4],
watermark is embedded into the low frequency sub-band of a frequency-domain watermarking techniques proved to
cover image by using alpha blending technique. The insertion
and extraction of the watermark in the grayscale cover image is
be more effective with respect to achieving the
found to be simpler than other transform techniques. The imperceptibility and robustness requirements of digital
proposed method is compared with the 1-level and 2-level watermarking algorithms [5]. Commonly used
DWT based image watermarking methods by using statistical frequency-domain transforms include the Discrete
parameters such as peak-signal-to-noise-ratio (PSNR) and Wavelet Transform (DWT), the Discrete Cosine
mean square error (MSE). The experimental results Transform (DCT) and Discrete Fourier Transform
demonstrate that the watermarks generated with the proposed (DFT). However, DWT has been used in digital image
algorithm are invisible and the quality of watermarked image watermarking more frequently due to its excellent
and the recovered image are improved. spatial localization and multi-resolution characteristics,
Index Terms— Image watermarking, 3-level DWT, wavelet
which are similar to the theoretical models of the human
transform, MSE, PSNR. visual system. Further performance improvements in
DWT-based digital image watermarking algorithms
could be obtained by increasing the level of DWT.
I. INTRODUCTION
II. RELATED WORK
With the widespread distribution of digital
In this section we will look into the review of digital
information over the World Wide Web (www), the
watermarking used for images. It describes the previous
protection of intellectual property rights has become
work which had been done on digital watermarking by
increasingly important. These information, which
using DWT technique and other techniques, including
include still images, video, audio, or text are stored and
the analysis of various watermarking schemes and their
transmitted in a digital format. Information stored in
results.
digital format can be easily copied without loss of
Mistry [6] introduced digital watermarking methods-
quality and efficiently distributed [1]. The digital
Spatial domain (like LSB) and transform domain (like
watermark is then introduced to solve this problem.
DCT, DWT) methods. The spatial domain is the normal
Digital watermarking is a branch of information hiding
image space, in which a change in position in image
which is used to hide proprietary information in digital
directly projects to a change in position in space. Ex.-
media like photographs, digital music, or digital video
Least Significant bit (LSB) method. Transform Domain
[2-3]. The ease with which digital content can be
Method produce high quality watermarked image by
exchanged over the Internet has created copyright
first transforming the original image into the frequency
infringement issues. Copyrighted material can be easily
domain by the use of Fourier Transform, Discrete
exchanged over peer-to-peer networks, and this has
Cosine Transform (DCT) or Discrete Wavelet
caused major concerns to those content providers who
transforms (DWT). Authors found that transform
produce these digital contents.
watermarking is comparatively much better than the
In order for a digital watermarking method to be
spatial domain encoding.
effective it should be imperceptible, and robust to
Van et al. [7] proposed two LSB techniques. First
common image manipulations like compression,
replaces the LSB of the image with a pseudo-noise (PN)
filtering, rotation, scaling cropping, and collusion
sequence, while the second adds a PN sequence to the
attacks among many other digital signal processing
LSB of the data. Another LSB data hiding method called
operations. Current digital image watermarking
Patchwork chooses n pairs (ai ; bi) of the points in an
techniques can be grouped into two major classes:

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56
 Image Watermarking Using 3-Level Discrete Wavelet Transform (DWT) 51

image and increases the brightness of the ai by one unit modifying the singular values of reference image using
while simultaneously decreasing the brightness of bi. the singular values of the watermark.
The M. Barni et al. [13] have developed an improved
problem with this paper is that data is highly sensitive to wavelet-based watermarking through pixel-wise
noise and is easily destroyed. Furthermore, image masking. It is based on masking watermark according to
quality may be degraded by the watermark. characteristics of HVS. The watermark is adaptively
Blossom et al. [8] proposed a DCT based added to the largest detail bands. The watermark
watermarking scheme which provides higher resistance weighing function is calculated as a simple product of
to image processing attacks such as JPEG compression, data extracted from HVS model. The watermark is
noise, rotation, translation etc. In this approach, the detected by correlation. The proposed method is robust
watermark is embedded in the mid frequency band of to various attacks but this method is complex than other
the DCT blocks carrying low frequency components and transform technique.
the high frequency sub band components remain unused. Kundur et al. [14] decomposed binary logo through
Watermark is inserted by adjusting the DCT coefficients DWT. The watermark is scaled by a salience factor,
of the image and by using the private key. Watermark computed on a block by block basis, depending on local
can then be extracted using the same private key without image noise sensitivity. It is then repeatedly added to the
resorting to the original image. Performance analysis subbands of DWT decomposition of host image. Visual
shows that the watermark is robust. masking is thus exploited upto only block resolution. A
W. Hong et al. [9] proposed a robust digital binary code is embedded by suitably quantizing the
watermarking scheme for copyright protection of digital coefficients of detail bands. For watermark recovery, the
images based on sub-sampling. The watermark is a embedded binary code is estimated by analyzing
binary image, which is embedded in discrete transform coefficients quantization. Once the code is estimated, it
coefficient of the host image and not used in the original is correlated and result is compared to a threshold
image. In this scheme, they had used chaotic map in chosen on the basis of a given false positive probability.
watermarked image. However the result of watermark Most methods found in literature are highly complex
image is good and robust to attack. and involve multiple execution stages. In this paper, we
Xia et.al [10] proposed a watermarking scheme based tried to propose a new simple methodology to hide a
on the Discrete Wavelet Transform (DWT). The grayscale image within another grayscale image using 3-
watermark, modeled as Gaussian noise, was added to the level DWT and the alpha-blending technique for the
middle and high frequency bands of the image. The process of security.
decoding process involved taking the DWT of a
potentially marked image. Sections of the watermark III. DISCRETE WAVELET TRANSFORM (DWT)
were extracted and correlated with sections of the
original watermark. If the cross-correlation was above a Discrete Wavelet transform (DWT) is a mathematical
threshold, then the watermark was detected. Otherwise, tool for hierarchically decomposing an image [9]. It is
the image was decomposed into finer and finer bands useful for processing of non-stationary signals. The
until the entire, extracted watermark was correlated with transform is based on small waves, called wavelets, of
the entire, original watermark. Performance analysis varying frequency and limited duration. Wavelet
shows that DWT is more robust to attack than DCT [8]. transform provides both frequency and spatial
The problem with the proposed method is that this description of an image. Unlike conventional Fourier
technique is susceptible to geometric attacks. transform, temporal information is retained in this
Akhil et al. [11] proposed a robust image transformation process. Wavelets are created by
watermarking technique based on 1-level DWT translations and dilations of a fixed function called
(Discrete Wavelet Transform). This method embeds mother wavelet.
invisible watermark into salient features of the original DWT is the multiresolution description of an image
image using alpha blending technique. Experiment result the decoding can be processed sequentially from a low
shows that the embedding and extraction of watermark resolution to the higher resolution [10]. The DWT splits
is depend only on the value of alpha. All the results the signal into high and low frequency parts. The high
obtained for the recovered images and watermark is frequency part contains information about the edge
identical to the original images. components, while the low frequency part is split again
G. Bhatnagar et al [12], presented a semi-blind into high and low frequency parts. The high frequency
reference watermarking scheme based on discrete components are usually used for watermarking since the
wavelet transform (DWT) and singular value human eye is less sensitive to changes in edges [14].
decomposition (SVD) for copyright protection and In two dimensional applications, for each level of
authenticity. Their watermark was a gray scale logo decomposition, we first perform the DWT in the vertical
image. For watermark embedding, their algorithm direction, followed by the DWT in the horizontal
transformed the original image into wavelet domain and direction. After the first level of decomposition, there
a reference sub-image is formed using directive contrast are 4 sub-bands: LL1, LH1, HL1, and HH1. For each
and wavelet coefficients. Then, their algorithm successive level of decomposition, the LL subband of
embedded the watermark into reference image by the previous level is used as the input. To perform
second level decomposition, the DWT is applied to LL1

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56
52 Image Watermarking Using 3-Level Discrete Wavelet Transform (DWT)

band which decomposes the LL1 band into the four sub- bands LL2, LH2, HL2, and HH2.
To perform third level decomposition, the DWT is
applied to LL2 band which decompose this band into the WMI  k * (LL2)  q * (WM2) (1)
four sub-bands – LL3, LH3, HL3, HH3. This results in
10 sub-bands per component. LH1, HL1, and HH1 where WMI = low frequency component of
contain the highest frequency bands present in the image watermarked image, LL3 = low frequency component of
tile, while LL3 contains the lowest frequency band. The the original image obtained by 3-level DWT, WM3 =
three-level DWT decomposition is shown in Fig.1. low frequency component of Watermark image, and k, q
= Scaling factors for the original image and watermark
respectively.
LL1 HL1 After embedding the cover image with watermark
Image image, 3-level Inverse discrete wavelet transform is
applied to the watermarked image coefficient to generate
LH1 HH1 the final secure watermarked image. Fig. 2 shows
watermark embedding process.

Cover Watermark
LL3 HL3 LL2 HL2 Image
Image
LH3 HH3 HL2 HL1 HL1
LH2 HH2 LH2 HH2

LH1 HH1 LH1 HH1

LL3 HL3 LL3 HL3
LH3 HH3 HL2 HL1 LH3 HH3 HL2 HL1
Figure 1. 3-Level discrete wavelet decompositions LH2 HH2 LH2 HH2

DWT is currently used in a wide variety of signal

LH1 HH1 LH1 HH1
processing applications, such as in audio and video
compression, removal of noise in audio, and the
simulation of wireless antenna distribution [8]. Wavelets
have their energy concentrated in time and are well
suited for the analysis of transient, time-varying signals.
α Blending Embedding Technique
Since most of the real life signals encountered are time
varying in nature, the Wavelet Transform suits many
applications very well.
Image
IV. PROPOSED WATERMARKING TECHNIQUE (WA)

Based on DWT technique, we propose a new

watermarking algorithm for remote multimodal Figure 2. Watermark embedding technique
biometric authentication system. The proposed
algorithm is divided into two parts, watermark B. Watermark Extraction
embedding and watermark extraction.
In this process firstly 3-level DWT is applied to
A. Watermark Embedding watermarked image and cover image which decomposed
In this process firstly the gray scale host image is the image in sub-bands. After that the watermark is
taken and 2-D, 3-level DWT (Discrete Wavelet recovered from the watermarked image by using the
Transform) is applied to the image which decomposes formula of the alpha blending.
image into low frequency and high frequency According to the formula of the alpha blending the
components. In the same manner 2-D, 3-level DWT is recovered image is given by
also applied to the watermark image which is to be
embedded in the host image. The wavelet used here is RW  (WMI - k * LL3) (2)
the wavelets of daubecheis. The technique used here for
inserting the watermark is alpha blending [11, 15]. In where RW= Low frequency approximation of
this technique the decomposed components of the host Recovered watermark, LL3= Low frequency
image and the watermark are multiplied by a scaling approximation of the original image, and WMI= Low
factor and are added. Since the watermark embedded in frequency approximation of watermarked image.
this paper is perceptible in nature or visible, it is After extraction process, 3-level Inverse discrete
embedded in the low frequency approximation wavelet transform is applied to the watermark image
component of the host image.According to the formula coefficient to generate the final watermark extracted
of the alpha blending the watermarked image is given by image. Fig. 3 shows the watermark extraction process.

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56
 Image Watermarking using 3-Level Discrete Wavelet Transform (DWT) 53

V. EXPERIMENTAL RESULTS
Image Cover
(WA) Image We compared proposed 3-level 2-D DWT to the 2-
level 2-D DWT technique. For both the technique we
have used grayscale images Lena as original image and
the cameramen’s image as the watermark. Both the
images are of equal size of 256X256. Fig. 4(a) shows the
LL3 HL3 LL3 HL3 original image and Fig. 4(b) shows the watermark image.
LH3 HH3 HL2 HL1 LH3 HH3 HL2 HL1 For embedding of watermark in the original image the
LH2 HH2 LH2 HH2 value of scaling factor k is varied from 1.5 to 0.6 by
keeping q constant and best result is obtained for k=0.98
for both 3 level and 2 level DWT. As the value of k is
LH1 HH1 LH1 HH1 decreased further to 0.2 the watermarked image becomes
darker and finally becomes invisible. Fig. 5 shows the
watermarked image using 3-level discrete wavelet
transform for different value of k.
For the process of recovering the watermark from the
α Blending Extraction Technique watermarked image the value of k is kept constant at
0.009 and q is varied from 1.5 to 0.6. For the higher
values of q the watermark becomes almost invisible and
Extracted as the value of q is reduced best result is obtained, and if
Watermark Image q is further reduced the recovered watermark becomes
darker and PSNR decreases. Fig.6 shows the recovered
image using 3-level discrete wavelet transform for
Figure 3. Watermark extraction technique different value of q.

(a) Original Image (b) Watermark

Figure 4. Original and Watermarked image

(a) For k=1.5, q=0.009 (b) For k=.98, q=0.009 (c) For k=.78, q=0.009 (d) For k=0.6, q=0.009

Figure 5. Watermarked images using 3-level DWT for various values of scaling factors k

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56
54 Image Watermarking using 3-Level Discrete Wavelet Transform (DWT)

(a) For q=1.5, k=0.009 (b) For q=.98, k=0.009 (c) For q=.78, k=0.009 (d) For q=0.6, k=0.009

Figure 6. Recovered images using 3-level DWT for various values of scaling factors q

The values of the MSE (Mean square error) and PSNR values of the scaling factors k and q for 1-level, 2-level
(Peak signal to noise ratio) are calculated for various and 3-level DWT as shown in Table I & Table II.

TABLE I.
COMPARISON OF 1-LEVEL, 2-LEVEL AND 3-LEVEL DWT FOR WATERMARKED IMAGE IN TERMS OF MSE & PSNR
q k MSE PSNR MSE PSNR MSE PSNR
Level1 Level1 Level2 Level2 Level3 Level3 Observation

1.5 0.009 3200 13.08 3156 13.14 3100 13.24

1.0 0.009 1.48 46.55 1.42 46.60 1.397 46.68

0.98 0.009 1.98 45.16 1.88 45.37 1.74 45.72 Best Result

0.95 0.009 21.33 34.84 20.85 34.94 20.08 35.10

0.90 0.009 21.3 28.00 100 28.08 98.16 28.21

0.85 0.009 247 24.21 243 24.29 240 24.41

0.78 0.009 551 20.71 540 20.79 530 20.91

0.6 0.009 1890 15.36 1900 15.43 1800 15.54

TABLE II.
COMPARISON OF 1-LEVEL, 2-LEVEL AND 3-LEVEL DWT FOR RECOVERED IMAGE IN TERMS OF MSE & PSNR
q k MSE PSNR MSE PSNR MSE PSNR
Level1 Level1 Level2 Level2 Level3 Level3 Observation

1.5 0.009 4450 11.65 4385 11.7 4300 11.79

1.0 0.009 6.4e-26 310 2.5e-26 304 5.3e-26 301

0.98 0.009 7.11 39.6 7.016 39.67 6.90 39.74 Best Result

0.95 0.009 44.46 31.65 43.85 31.71 43.11 31.79

0.90 0.009 178 25.6 175 25.690 170 25.76

0.85 0.009 400 22.1 395 22.169 390 22.24

0.78 0.009 861 18.78 849 18.84 840 18.92

0.6 0.009 2850 13.58 2810 13.649 1800 15.54

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56
 Image Watermarking using 3-Level Discrete Wavelet Transform (DWT) 55

Best result for watermarked image is obtained at k= [9] W. Hong and M. Hang, Robust Digital
0.98 & q=0.009 and best result for recovered image is Watermarking Scheme for Copy Right Protection,
obtained at k= 0.009 & q= 0.98. IEEE Trans. Signal Process, vo.l2, pp. 1- 8, 2006.
From Table 1 and Table 2, it can be observed that [10] X. Xia, C. Boncelet, and G. Arce, A
the value of PSNR (peak signal to noise ratio) is higher Multiresolution Watermark for Digital Images,
in case of 3-level Discrete wavelet transform for both Proc. IEEE Int. Conf. on Image Processing, Oct.
the watermarked and recovered image which means 1997.
that the proposed 3-level DWT algorithm provide the [11] Akhil Pratap Shing, Agya Mishra, Wavelet Based
better performance than 1-level & 2-level DWT Watermarking on Digital Image, Indian Journal of
algorithm. computer Science and Engineering, 2011.
[12] Bhatnagar, G. and Raman, B., A new robust
VI. CONCLUSION reference watermarking scheme based on DWT-
SVD, Elsevier B.V. All rights reserved, 2008.
In this paper, an image watermarking technique [13] Barni M, Bartolini F, Piva, An Improved Wavelet
based on a 3-level discrete wavelet transform has been Based Watermarking Through Pixelwise Masking,
implemented. This technique can embed the invisible IEEE transactions on image processing, 2001.
watermark into salient features of the image using
alpha blending technique. Experiment results shows [14] D. Kundur and D. Hatzinakos, Digital
that the quality of the watermarked image and the Watermarking using Multiresolution Wavelet
recovered watermark are dependent only on the scaling Decomposition, Proceedings, IEEE International
factors k and q and also indicate that the three level Conference Acoustic, Speech, Signal Processing,
DWT provide better performance than 1-level and 2- 1998.
level DWT. All the results obtained for the recovered [15] Nilanjan Dey, Anamitra Bardhan Roy, Sayantan
images and the watermark are identical to the original Dey, A novel approach of color image hiding
images. using RGB color planes and DWT, International
Journal of Computer Applications, 2011.
REFERENCES
Nikita Kashyap received the BE degree in
[1] Ibrahim, R. and Kuan, T. S., Steganography Electronic & Telecommunication Engineering from
Imaging (SIS): Hiding Secret Message inside an Institute of technology, Guru Ghasidas Central
Image. Proceedings of the World Congress on
University, Bilaspur in 2010. She is currently working
Engineering and Computer Science, 2010, San
towards the M.E. degree in Communication
Francisco, USA.
Engineering from Shankaracharya college of
[2] Er-Hsien Fu, Literature Survey on Digital Image engineering and technology, Bhilai. Her areas of
Watermarking, EE381K Multidimensional Signal interests are image processing, data hiding and image
Processing, 1998.
watermarking.
[3] L. Robert, T. Shanmugapriya, A Study on Digital
Watermarking Techniques, International Journal
G.R.Sinha is Professor of Electronics &
of Recent Trends in Engineering, 2009.
Telecommunication and Associate Director of Faculty
[4] G. Rosline Nesa Kumari, B. Vijaya Kumar, L. of Engineering and Technology of Shri
Sumalatha, and Dr V. V. Krishna, Secure and Shankaracharya Group of Institutions Bhilai, India. He
Robust Digital Watermarking on Grey Level obtained his B.E. (Electronics) and M.Tech. (Computer
Images, International Journal of Advanced Science Technology) from Government Engineering College
and Technology, 2009. (Now National Institute of Technology, Raipur). He
[5] Baisa L. Gunjal, R.R. Manthalkar, An overview of received Gold Medal for obtaining first position in the
transform domain robust digital image University. He received his Ph.D. in Electronics &
watermarking algorithms, Journal of Emerging Telecommunication from Chhattisgarh Swami
Trends in Computing and Information Sciences, Vivekanand Technical University, Bhilai.
2010. His research interest includes Digital Image
[6] Darshana Mistry, Comparison of Digital Processing and it applications in biometric security,
Watermarking methods, 21st Computer Science forensic science, pattern recognition, early detection of
Seminar SA1-T1-7, IJCSE, 2010.
breast cancer, content retrieval of underwater imaging,
[7] R. G. van Schyndel, A. Z. Tirkel, and C. F.
nano imaging, neuro‐fuzzy based Vehicle license
Osborne, A digital watermark, in Proc. IEEE Int.
Conf. Image Processing (ICIP), 1994. plate recognition, multimodal biometrics etc. He has
[8] Blossom Kaur, Amandeep Kaur, Jasdeep Singh, published 101 research papers in various international
Steganographic Approach for hiding Image in and national journals and conferences. He is active
DCT Domain, International Journal of Advances reviewer and editorial member of more than 12
in Engineering & Technology, July 2011. international journals such as Applied Physics research
of Canada, IBSU SCIENTIFIC JOURNAL of Georgia,

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56
56 Image Watermarking using 3-Level Discrete Wavelet Transform (DWT)

Scientific Journals International (SJI) of USA, Computer research degree committee of MATS University in
Science Journals of Malaysia etc. computer science & engineering. Dr. G.R. Sinha is Vice
He is recipient of many awards like Engineer of the President of Computer Society of India for Bhilai Chapter.
Year Award 2011, Young Engineer Award 2008, Young He has delivered many keynote speeches and chaired
Scientist Award 2005, IEI Expert Engineer Award 2007, many technical sessions in international conferences in
nominated for ISCA Young Scientist Award 2006 and SINGAPORE, Mumbai, Nagpur and across the country.
awarded Deshbandhu Merit Scholarship for 05 years. He He is member of signal processing society of IEEE,
has been selected as Distinguished IEEE Lecturer in IACSIT and also of many national professional bodies
IEEE India council for Bombay section. He is member of like IETE, ISTE, CSI, ISCA, and IEI.
various committees of the University also the member of

Copyright © 2012 MECS I.J. Modern Education and Computer Science, 2012, 3, 50-56