![ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011
A New Watermarking Algorithm Based on Image
Scrambling and SVD in the Wavelet Domain
U. M. Gokhale1, Y. V. Joshi2
1
U.M.Gokhale is working as Asst.Professor in Electronics and Telecommunication Department in G.H.Raisoni Institute of
Engineering and Technology for Women, Nagpur., Maharashtra, India
(e-mail : umgokhale@gmail.com)
2
Y.V.Joshi is working as Director Walchand College Of Engineering Sangli, Maharashtra, India
(e-mail:yashwant.josh@gmail.com).
Abstract- A new watermarking algorithm which is based on modifying the singular vectors instead of singular values. In
image scrambling and SVD in the wavelet domain is discussed [8] Ghazy et al. Proposed a scheme in which the image is
in this paper. In the proposed algorithm, chaotic signals are divided into blocks and then watermark is embedded in
generated using logistic mapping and are used for scrambling singular values of each block separately. In [9] SVD is used
the original watermark. The initial values of logistic mapping
with a human visual system (HVS) model. In [11] , however, it
are taken as private keys. The covert image is decomposed
into four bands using integer wavelet transform; we apply is demonstrated that a counterfeit attack on SVD watermarked
SVD to each band and embed the scrambled watermark data image is possible and proposes a method to counterattack it.
by modifying the singular values. In [12] and [13] it is pointed out that SVD watermarking suffers
from false watermark detection. In [14] it has been shown
Index words - logistic mapping, singular value decomposition,
discrete wavelet transforms. that SVD based watermarking algorithms are robust to
distortions as long as attacks are not severe, also an attack
method to extract a false watermark from any watermarked
I. INTRODUCTION
image is proposed. Thus SVD based watermarking methods
With the rapid growth of internet and networks techniques, cannot be used for the ownership of an image. In our
Multimedia data transforming and sharing has become proposed scheme watermarking is used for image
common to many people. Multimedia data is easily copied authentication.
and modified, so necessity for copyright protection is
increasing. Digital watermarking has been proposed as the III.SINGULAR VALUE DECOMPOSITION
technique for copyright protection of multimedia data. AND IMAGE ENCRYPTION
Existing watermarking schemes can be divided into two
categories spatial domain and transform domain. Spatial A. Singular Value Decomposition
domain techniques embed data by directly modifying pixel Let A be an image matrix of size N×N. Using SVD the
values of the host image, while transform domain techniques matrix A can be decomposed as:
embed data by modifying transform domain coefficients.
Discrete cosine transform (DCT) and discrete wavelet
transform (DWT), which are used in image compression
standards JPEG and JPEG2000 respectively , are two main
transform methods used in transform domain watermarking.
However, transform methods attempt to decompose images
in terms of a standard basis set. This is not necessarily the
optimum set. Recently Singular value decomposition (SVD)
has been used for implementation of watermarking algorithms
[1-10].
II. THE RELATED WORK
In [1] Gorodetski et al. embed watermark bits by modifying
the quantized singular values of the host image. In [2],
Chandra computed SVD of both the host and watermark Where r is the rank of matrix A(r d” N), UA and VA are or-
images and then singular values of the watermark images are thogonal matrices of size N×N, whose column vectors are ui
minified and added to those of the host image. In [3] Liu and and vi. S is an N×N diagonal matrix containing the singular
Tan applied SVD to only host image and watermark bits are values si assumed to be in decreasing order.
directly added to its singular values. In [4] Ganic et al. Propose
a two layer watermarking scheme. In [5] SVD is used with
DCT and in [6] SVD is used with DWT embedding data in all
frequencies. In [7] Agrawal et al. Propose a scheme of
1
© 2011 ACEEE
DOI: 01.IJNS.02.03.141](https://image.slidesharecdn.com/141-120924224858-phpapp01/85/A-New-Watermarking-Algorithm-Based-on-Image-Scrambling-and-SVD-in-the-Wavelet-Domain-1-320.jpg)
![ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011
TABLE I VARIOUS ATTACKS AND THEIR EFFECT VI. CONCLUSIONS
The proposed watermarking algorithm is non-blind
watermarking technique as the original image is required for
the watermark extraction. The PSNR is 52.46 before the
attacks. The value of NC is close to 1.0000 which shows the
robustness to the attack. In the existing watermarking
algorithms there is always a trade off between higher
robustness and degree of perceptibility. The proposed
algorithm achieves both high robustness and
imperceptibility. The security of the watermark is improved
by its encryption using the chaos sequence generated by
logistic mapping. Thus it can be used for image authentication.
REFERENCES
1. V.Gorodetski, L. Popyack, V. Samoilov and V. Skormin, “ SVD
based approach to transparent embedding data into digital images,”
in proc. International Workshop on mathematical methods, model
and architectures for computer network security (MMM-
ACNS’01), may 2001
2. Chandra D.V.S.; “Digital image watermarking using singular value
decomposition”, Circuits and Systems 2002.MWSCAS-2002, vol.3,
4-7Aug 2002, pp. 264-267.
3. R. Liu, T. Tan, “An SVD –based watermarking scheme for
protecting rightful ownership”, IEEE Transaction on Multimedia
Volume 4, issue 1, March 2002 pp121-128.
4. E.Gagnic, N. Zubair and A.M.Eskicioglu, “An optimal
Watermarking based on singular value decomposition,”in proc
IASTED international Conference on Communication , network
and Information security(CNIS’03),Dec.2003
5. A.Sverdlov, S. Dexter and A.M.Eskicioglu, “Robust DCT-SVD
domain image watermarking for copyright protection : Embedding
data in all frequencies, in proc. the 2004 Multimedia and Security
Workshop, ACM press, sep 2004,pp. 166-174.
6. E. Gagnic and A.M. Eskicioglu, “Robust embedding of visual
watermarks using discrete wavelet transform and singular value
decomposition”, Journal of Electronic Imaging vol. 14, no.4, Dec
2005.
7. R.Agrawal and M.S.Santhanam, “Digital watermarking in the
singular vector domain,”Mar.2006.
8. R.A.Ghazy, N.A El-Fishawy, M. M Hadhoud, M.I.Dessouky
and F.E. Abd El-Samie, “An efficient Block by block SVD based
image watermarking scheme”, Ubiquitous computing and
communication Journal ,2(5),2007,pp. 1-9.
9. Q.Li, C. Yuan and Y.Z. Zhong, “A novel SVD based watermarking
scheme using human visual model,” in Proc. The 2nd International
Symposium on Computational intelligence and Industrial
Applications, Nov 2006.
10. Andrews H, Patterson C., “Singular Value Decomposition (SVD)
Image Coding”, IEEE Transaction on [legacy, pre-1988], Volume
24, Issue 4, April 1976, pp425-432.
11. Y.D.Wu, “On the security of an SVD-Based Ownership
Watermarking, IEEE Transactions on Multimedia, 7 (4), August
2005, pp.624-627
4
© 2011 ACEEE
DOI: 01.IJNS.02.03.141](https://image.slidesharecdn.com/141-120924224858-phpapp01/85/A-New-Watermarking-Algorithm-Based-on-Image-Scrambling-and-SVD-in-the-Wavelet-Domain-4-320.jpg)
A New Watermarking Algorithm Based on Image Scrambling and SVD in the Wavelet Domain
- 1. ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011 A New Watermarking Algorithm Based on Image Scrambling and SVD in the Wavelet Domain U. M. Gokhale1, Y. V. Joshi2 1 U.M.Gokhale is working as Asst.Professor in Electronics and Telecommunication Department in G.H.Raisoni Institute of Engineering and Technology for Women, Nagpur., Maharashtra, India (e-mail : umgokhale@gmail.com) 2 Y.V.Joshi is working as Director Walchand College Of Engineering Sangli, Maharashtra, India (e-mail:yashwant.josh@gmail.com). Abstract- A new watermarking algorithm which is based on modifying the singular vectors instead of singular values. In image scrambling and SVD in the wavelet domain is discussed [8] Ghazy et al. Proposed a scheme in which the image is in this paper. In the proposed algorithm, chaotic signals are divided into blocks and then watermark is embedded in generated using logistic mapping and are used for scrambling singular values of each block separately. In [9] SVD is used the original watermark. The initial values of logistic mapping with a human visual system (HVS) model. In [11] , however, it are taken as private keys. The covert image is decomposed into four bands using integer wavelet transform; we apply is demonstrated that a counterfeit attack on SVD watermarked SVD to each band and embed the scrambled watermark data image is possible and proposes a method to counterattack it. by modifying the singular values. In [12] and [13] it is pointed out that SVD watermarking suffers from false watermark detection. In [14] it has been shown Index words - logistic mapping, singular value decomposition, discrete wavelet transforms. that SVD based watermarking algorithms are robust to distortions as long as attacks are not severe, also an attack method to extract a false watermark from any watermarked I. INTRODUCTION image is proposed. Thus SVD based watermarking methods With the rapid growth of internet and networks techniques, cannot be used for the ownership of an image. In our Multimedia data transforming and sharing has become proposed scheme watermarking is used for image common to many people. Multimedia data is easily copied authentication. and modified, so necessity for copyright protection is increasing. Digital watermarking has been proposed as the III.SINGULAR VALUE DECOMPOSITION technique for copyright protection of multimedia data. AND IMAGE ENCRYPTION Existing watermarking schemes can be divided into two categories spatial domain and transform domain. Spatial A. Singular Value Decomposition domain techniques embed data by directly modifying pixel Let A be an image matrix of size N×N. Using SVD the values of the host image, while transform domain techniques matrix A can be decomposed as: embed data by modifying transform domain coefficients. Discrete cosine transform (DCT) and discrete wavelet transform (DWT), which are used in image compression standards JPEG and JPEG2000 respectively , are two main transform methods used in transform domain watermarking. However, transform methods attempt to decompose images in terms of a standard basis set. This is not necessarily the optimum set. Recently Singular value decomposition (SVD) has been used for implementation of watermarking algorithms [1-10]. II. THE RELATED WORK In [1] Gorodetski et al. embed watermark bits by modifying the quantized singular values of the host image. In [2], Chandra computed SVD of both the host and watermark Where r is the rank of matrix A(r d” N), UA and VA are or- images and then singular values of the watermark images are thogonal matrices of size N×N, whose column vectors are ui minified and added to those of the host image. In [3] Liu and and vi. S is an N×N diagonal matrix containing the singular Tan applied SVD to only host image and watermark bits are values si assumed to be in decreasing order. directly added to its singular values. In [4] Ganic et al. Propose a two layer watermarking scheme. In [5] SVD is used with DCT and in [6] SVD is used with DWT embedding data in all frequencies. In [7] Agrawal et al. Propose a scheme of 1 © 2011 ACEEE DOI: 01.IJNS.02.03.141
- 2. ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011 In watermarking applications, SVD has following properties: A. Watermark embedding: 1) SVD is able to efficiently represent the intrinsic algebraic The watermark embedding algorithm is as follows: properties of an image, where singular values correspond to 1) Using the integer wavelet transform(IWT), cover image A the luminance of the image and singular vectors reflect is first decomposed into four sub bands LL,HL,LH,HH as geometry characteristics of the image. shown in Fig.2. 2) Singular values have good stability, which means small perturbation added to image will not significantly change the 2) Apply SVD to each sub band image : corresponding singular values. 3) An image matrix has many small singular values compared with the first value. If these values are ignored it will have 3) Obtain the scrambled or encrypted image from the original much effect on the quality of reconstructed image. image by using logistic mapping as described in section 2. B. Image Encryption Chaos signal are a kind of pseudorandom, irreversible 4) Apply SVD to the encrypted image. and dynamical signals generated by deterministic non linear equations, which possess good characteristics of pseudorandom sequences. There are many ways to generate chaos sequence. We apply logistic mapping chaos sequence. 5) Modify the singular values of the cover image in each sub The equation for logistic mapping chaos is given by equation band with singular values of the encrypted watermark; (5). Where 0 d” µ d” 4, is called as branch parameter, x 6) Obtain the four sets of modified IWT coefficients. õ(0,1).Logistic map is chaotic when 3.569945d” µ d” 4,chaotic systems are highly sensitive to initial parameters. In order to 7) Apply the inverse IWT using the four sets of modified get chaotic sequence, the chaotic signal x (n+1) must be IWT coefficients to produce the watermarked cover image. transformed into binary sequences. We use the logistic map to generate sequence W ( i ). Then, we set a threshold T. If element of sequence is larger than the threshold, we replace that element by 1; otherwise, replace by 0, as described by equation ( 6 ). Make the xor operation between the sequence and the matrix of the original watermark to obtain the scrambled watermark or encrypted watermark. Fig 1 shows the original and the encrypted watermark. Figure 2 Wavelet decomposition B. Watermark detection The watermark detection algorithm is as follows 1)Using DWT, decompose the watermarked (and possibly attacked) cover image  into four sub bands LL, HL, LH, HH as shown in Fig 2. IV. PROPOSED METHOD Proposed method is explained in the following section. The scrambled watermark is obtained from the original 2) Apply SVD to each sub band image : watermark and is embedded into the cover image. The watermarked image is distributed. When required the test image is checked for the presence of the watermark by the 3) Extract the singular values from each sub band watermark detection algorithm. As the watermark is semi fragile it allows to alter the image by specific image processing operations. 2 © 2011 ACEEE DOI: 01.IJNS.02.03.141
- 3. ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011 1) Construct four watermark images from four sub bands. 2) The original watermark can be obtained by xor operation with the chaotic sequence W (i). V. EXPERIMENTAL RESULTS Figure 4 Watermarked image after adding salt pepper noise The experimental simulation is carried out using MATLAB. The standard test images of 512×512×8 greyscale were used for studying the effects of perceptibility and robustness of the watermarking algorithm on a 256 × 256 binary watermark image. In order to evaluate the difference between cover image and watermarked image, we used Mean square error (MSE) and Peak Signal to Noise Ratio (PSNR) to estimate the watermark imperceptibility. a) Variance =0.001 b) variance = 0.002 Figure 5 after adding Gaussian noise Where, MSE is the Mean Square Error between the original and watermarked image. Where x (i, j) and y (i, j) represent the pixel value of the original and the watermarked image respectively. A higher PSNR indicates that the quality of the watermarked image is a) 300 b) 45 0 closer to the original image. Fig 2 shows the original and Figure 6 Wwatermarked image after Rotation watermarked image. We estimate the similarity between the TABLE II PSNR AND NC FOR G AUSSIAN NOISE ATTACK original watermark and the extracted watermark using normalized correlation (NC): The NC shows the robustness of the algorithm. Its value TABLE III PSNR AND NC FOR SALT AND PEPPER NOISE ATTACK is 1.0000 before the watermark image is attacked. The results for different attacks are shown in table I. In order to investigate robustness watermarked image was attacked by various attacks. The original image is shown in Figure 3(a), and the watermarked image is shown in Figure 3(b). Fig 4 shows the salt and pepper noise attack. Fig 5 shows Gaussian TABLE IV PSNR AND NC FOR R OTATION ATTACK noise attack and Fig 6 show the rotation attack. Table I-IV shows the results for the various attacks and their effects on PSNR, NC and extracted watermark. Figure 3 a) Original image b) Watermarked image 3 © 2011 ACEEE DOI: 01.IJNS.02.03.141
- 4. ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011 TABLE I VARIOUS ATTACKS AND THEIR EFFECT VI. CONCLUSIONS The proposed watermarking algorithm is non-blind watermarking technique as the original image is required for the watermark extraction. The PSNR is 52.46 before the attacks. The value of NC is close to 1.0000 which shows the robustness to the attack. In the existing watermarking algorithms there is always a trade off between higher robustness and degree of perceptibility. The proposed algorithm achieves both high robustness and imperceptibility. The security of the watermark is improved by its encryption using the chaos sequence generated by logistic mapping. Thus it can be used for image authentication. REFERENCES 1. V.Gorodetski, L. Popyack, V. Samoilov and V. Skormin, “ SVD based approach to transparent embedding data into digital images,” in proc. International Workshop on mathematical methods, model and architectures for computer network security (MMM- ACNS’01), may 2001 2. Chandra D.V.S.; “Digital image watermarking using singular value decomposition”, Circuits and Systems 2002.MWSCAS-2002, vol.3, 4-7Aug 2002, pp. 264-267. 3. R. Liu, T. Tan, “An SVD –based watermarking scheme for protecting rightful ownership”, IEEE Transaction on Multimedia Volume 4, issue 1, March 2002 pp121-128. 4. E.Gagnic, N. Zubair and A.M.Eskicioglu, “An optimal Watermarking based on singular value decomposition,”in proc IASTED international Conference on Communication , network and Information security(CNIS’03),Dec.2003 5. A.Sverdlov, S. Dexter and A.M.Eskicioglu, “Robust DCT-SVD domain image watermarking for copyright protection : Embedding data in all frequencies, in proc. the 2004 Multimedia and Security Workshop, ACM press, sep 2004,pp. 166-174. 6. E. Gagnic and A.M. Eskicioglu, “Robust embedding of visual watermarks using discrete wavelet transform and singular value decomposition”, Journal of Electronic Imaging vol. 14, no.4, Dec 2005. 7. R.Agrawal and M.S.Santhanam, “Digital watermarking in the singular vector domain,”Mar.2006. 8. R.A.Ghazy, N.A El-Fishawy, M. M Hadhoud, M.I.Dessouky and F.E. Abd El-Samie, “An efficient Block by block SVD based image watermarking scheme”, Ubiquitous computing and communication Journal ,2(5),2007,pp. 1-9. 9. Q.Li, C. Yuan and Y.Z. Zhong, “A novel SVD based watermarking scheme using human visual model,” in Proc. The 2nd International Symposium on Computational intelligence and Industrial Applications, Nov 2006. 10. Andrews H, Patterson C., “Singular Value Decomposition (SVD) Image Coding”, IEEE Transaction on [legacy, pre-1988], Volume 24, Issue 4, April 1976, pp425-432. 11. Y.D.Wu, “On the security of an SVD-Based Ownership Watermarking, IEEE Transactions on Multimedia, 7 (4), August 2005, pp.624-627 4 © 2011 ACEEE DOI: 01.IJNS.02.03.141
- 5. ACEEE Int. J. on Network Security , Vol. 02, No. 03, July 2011 12. X.P. Zhang, K.Li comments on “An SVD-Based Watermarking Dr. Y. V. Joshi is presently working as Director scheme for Protecting Rightful Ownership”, IEEE Transaction on of Walchand College of Engineering, Sangli since multimedia Vol.7,no.2,2005, pp.593-594. May 2009. Earlier he was at SGGS Institute of 13. R.Rykaczewski, comments on “An SVD-Based Watermarking Engineering and Technology, Vishnupuri, scheme for Protecting Rightful Ownership”, IEEE Transaction on Nanded since 1986 in various capacities starting multimedia Vol.9,no.2,2007,pp.421-423. with Lecturer (1986-1993), Assistant Professor 14. Xiong Changzhen, Guo Fenhong,Li Zhengxi, “Weakness (1993-2001), Professor (2001 onwards). He Analysis of Singular Value based Watermarking”, in proceedings of also served as Head of Electronics and the 2009 IEEE international Conference on Mechtronics and Telecommunication Engineering department (2002-04), First Dean Automation August 9-12, Changchun, China. of Academics (2004-06), Dean (Finance and Resource Mobilization (2007-08). He did his graduation B. E. Electronics in 1986 and post U.M Gokhale is presently working as graduation M. E. Electronics 1991 from SGGS Institute of Asst.Professor and Head in Department of Engineering and Technology, Vishnupuri, Nanded. He completed Electronics and Telecommunication in Ph. D. (1998) from IIT, Delhi. He has 15 international Journal G.H.Raisoni Institute of Engineering and publications and 25 national and international conference Technology for women, Nagpur (MS), India. He publications to his credit. He is Life Member of ISTE. He conducts is Life member of Indian Society for Technical and supervises research in the areas of Signal and Image Education (ISTE) and also Associate member of processing.He has so far supervised more than 25 M.E./M. Tech Institution of Engineers (IE).He has 22 years teaching experience in dissertations and 3 Ph. D. students. Engineering College. 5 © 2011 ACEEE DOI: 01.IJNS.02.03.141