Colour Image Watermarking Using CDMA

Color Image Watermarking Robust to JPEG Compression

Al-Hilli Ahmed M. Z. Technical Collage of Najaf

Abstract Digital Watermarking has emerged as a new area of research in an attempt to prevent illegal copying and duplication and false representation. In this paper, a proposed algorithm for digital watermarking for color images was presented. This algorithm make use of the linearity for DCT transform and introduce a watermarking that is robust against JPEG compression, in this algorithm, the linearity of the Discrete Cosine Transform is used which approve a robust watermarking algorithm against JPEG compression for high compression ratios till 1% image quality.

الخلصة ظهرت العلمات المائية كمجال جديد للبحث في محاولة لمنع الستنساخ الغير القانوني و التمثيل الخاطئ. في هذه البحث, تم اقتراح خوارزمية لجعل العلمات المائية مقاومة لعمليات الضغط باستخدام تقنية JPEG حيث تم الستفادة من الخواص الخطية لعملية تحويله الجيب تمام المنفصلة و التي أثبتت مقاومة عالية لعمليات الضغط لنسب ضغط كبيرة وصلت إلى حد %1. Key Words: Watermarking, JPEG, DCT, correlation 1. Introduction The success of the Internet and digital consumer devices has profoundly changed our society and daily lives by making the capture, transmission, and storage of digital data extremely easy and convenient. However, this raises a big concern in how to secure these data and preventing unauthorized use. This issue has become problematic in many areas. For example, there are many studies showing that the music and video industry loses billions of dollars per year due to illegal copying and downloading of copyrighted materials from the Internet (Tsui, et al, 2008). One of the possible and popular solutions is watermarking of digital multimedia contents. Digital watermarking is a technology that creates and detects invisible markings, which can be used to trace the origin, authenticity, and legal usage of digital data. Ideally, they should be hard to notice, difficult to reproduce, and impossible to remove without destroying the medium they protect.(Saha, et al, 2007) An image watermarking scheme should at least meet the following requirements: transparency and robustness. Transparency means that the embedded watermark should be perceptual invisible and robustness means that the embedded watermark shouldn’t be erased by any attack that maintains the host image quality acceptable. Trade off between transparency and robustness is one of the most important issues in image watermarking.(Soheili, 2008) There are two main types of watermarks. A blind (or public) watermark is invisible, and is extracted “blindly” without knowledge of the original host image or the watermark itself. The second is non-blind (also asymmetric marking or private), in which the watermark is embedded in the original host, and is intentionally visible to the human observer. The original data is required for watermark extraction .The blind watermark has many more applications than the visible watermark. The subject of this paper is a blind watermark.( Brannock,2008) Each watermarking application has its own specific requirements, so there is no particular set of requirements to be met by all watermarking techniques, but some general principles are applicable in most cases. • Imperceptibility: an embedded watermark is truly imperceptible if a user cannot distinguish the original from the watermarked version. • Payload: the amount of information that can be stored in a watermark depends on the application. • Robustness: it should not be possible to remove or alter the watermark without sufficient degradation of the perceptual quality of the host data so as to render it unusable. • Security: according to Kerckhoff’s assumption, the security of the encryption techniques must lie in the choice of a key. This assumption is also valid for watermarking techniques (Emek et al., 2005) In the literature, several techniques have been developed for watermarking. In (Brassil et al., 2005), three coding methods for hiding electronic marking in document were proposed. In (Pitas et al.,2005), (Bruyndonckx, et al., 1995), (S. Walton S. , 1995) and (Bender et al., 1995) , the watermarks are applied on the spatial domain. Other than spatial domain watermarking, frequency domain approaches have also been proposed. In (Koch. et al.,1995), a copyright code and its random sequence of locations for embedding are produced, and then superimposed on the image based on a JPEG model. In (Cox et al., 1995), the spread spectrum communication technique is also used in multimedia watermarking. Robustness of the watermark against any attack which try to remove or alter the watermark play a great role in the evaluation of the watermark process. Attacks usually include two types: removing the watermark and rendering the watermark undetectable. Attack categorization may include (but are not limited to) (Kirovski Darko, et al.,2003) (Petitcola., 2000) (Hartung, et al., 2000) : • Adding noise such as Gaussian. • Using linear filtering such as low-pass filtering. • Compressing the image, such as JPEG does. • Applying transforms such as translation, rotation and scaling. The most common method for removing the watermark is by compressing the image with JPEG compression. Since the compression process reduces the redundancy in image, and the watermark adds information to image, the two processes are odds and compression of the image will often remove the watermark and the most common method for compression is JPEG compression, so the need of watermarking algorithm that withstand the JPEG compression is urgent. This paper is organized as follow: section 2 discuss the theoretical concepts of the proposed algorithm, section 3 explain the algorithm for embedding and extraction of the watermark, section 4 reviews the experimental results for the proposed algorithm. 2. The Theory The whole mathematical model are based on the general embedding equation which states that :

Iw  I   W...... (1) Where:

Iw :is the watermarked image. I : the original image. W : the watermark.  : is a constant. Here, for explaining, we define the compression procedure as a function called JPEG: JPEG(x)  The compressed image……………………………(2) Where x :represent any two dimensional matrix. This function represents the tiling, the DCT transformation and the quantization process since the rest represent the lossless compression where no loss of information (no attack against the watermark) occurs. Since the tiling, DCT and quantization are linear processes, applying JPEG function to the equation (1) gets:

JPEG(Iw )  JPEG(I)    JPEG(W )...... (3) Equation (3) states that if we create a watermark that is survive the JPEG compression, that is : JPEG(W )  W...... (4) Equation (3) changes to:

JPEG(Iw )  JPEG(I)   W...... (5) And by using the correlation, the watermark is easily extracted from the compressed image. So, the main task here is to create a watermark that achieves equation (4). Here, we propose a watermark that is constant for 8x8 blocks as shown in fig (1). This pattern can be created as follow: M 1  i  w  1,1 8 ...... (6) ij N 1  j  8 Where : M : is the number of rows in image. N : is the number of columns in image. and create W by: M 1  i  W  w 8 ...... (7) 8i7,8 j 7 i, j N 1  j  8

One last thing, since W has homogenous blocks of size 8x8, it is expected that the correlation between W and the original image is high; and if we want to detect the watermark, the correlation matrix must be stored and subtracted from the correlation result as follow: cor(JPEG(Iw ),W )  cor(JPEG(I)  W ,W ) ...... (8) And since the correlation is linear, we can write equation (8) as: cor(JPEG(Iw ),W )  cor(JPEG(I),W )  cor(W ,W ) ...... (9) Subtracting cor(JPEG(I),W ) from equation (9), we get: cor(JPEG(Iw ),W )  cor(JPEG(I),W )  cor(W ,W ) ...... (10)

And here from equation 10, it is expected that the watermark will probably extracted for all compression ratios of JPEG compression

Fig (1) The Proposed Watermark 3. The Algorithm: The algorithm consists of two steps: embedding of the digital watermark, and recovery of the watermark from image that possibly attacked. Figure (2) represents the general embedding and extraction procedures. The embedding algorithm is summarized by assigning each bit of the watermark a Pseudo-Noise (PN) pattern which satisfy equations (6) and (7), that is every bit is act as a random noise to each other and that the correlation of each two bits pattern are small. First, the image is converted from RGB space to YCrCb color space since this transformation splits the luminance which represents the grayscale signal used to display pictures on monochrome (black and white) televisions ( Y) and two other components carry the hue and saturation information which represent the color information (Cb and Cr). The embedded watermark is added to the Y matrix which represent the most perceptual component in the image. Then for each bit corresponding to the high level in the watermark image, using equation 6 and 7 to generate a PN sequence using an independent key, the summation of all these PN sequence is scaled and added to the cover image as in the equation below:

Fig 2 The generic embedding and extracting model Iw (x, y)  I(x, y)   *W (x, y)...... 11 Where:

Iw (x, y) represent the watermarked image. I(x, y) represent the cover image. W (x, y) represent the watermark  is a constant represent the gain factor. Increasing the gain factor  increases the robustness of the watermark at the expense of the quality of the watermarked image. In the extraction stage, the same PN sequence is generated using the same independent key according to equation 6 and 7, and this sequence is correlated to the watermarked image and if the correlation is higher than a pre-specified threshold, then the watermark exists and this process is repeated for each bit in the watermark.

4. Experimental Results Simulation are performed to evaluate the proposed algorithm above. The simulation is performed using two images. Figure 3 shows the watermark, figure 4 and 5 represent the original images. In this simulation, PSNR is used to evaluate the distortion of the watermarked image with respect to the original image where PSNR is (Umbaugh S. E., 1998):

   2   (L 1)  PSNR  10log10 N 1 N 1 ...... 12  1 2   I (r,c)  I(r,c)   N  M r 0 c0  Where : L : the number of gray level I : is the watermarked image I : is the original image N,M : the image dimensions. And in order to measure the quantitative similarity between the embedded watermark and the extracted one, the normalized correlation coefficient is used in this paper:

Amn  ABmn  B  r  m n ...... 13  2  2  Amn  A Bmn  B    m n  m n 

Fig (3) The watermark

For the Peppers image, after embedding the watermark, the PSNR is 34.16 for   1 and the watermarked image is shown in figure 6, the image was applied to JPEG compression with different compression quality as listed in table 1 For the Baboon image, the PSNR is 34.3711 for   1 and the watermarked image is shown in figure 7, the image was applied to JPEG compression with different compression quality as listed in table 2 It can be concluded from the results that the proposed algorithm is robust to JPEG compression and the correlation coefficient is constant for any compression quality even for 1% image quality which distort the image completely as shown in figure 8 and 9 and the watermark for 1% image quality is shown in figure 10 and 11. Table 1 The results obtained for Baboon image

Compression Quality % Normalized Correlation Coefficient 100 0.9739 90 0.9739 80 0.9739 70 0.9739 60 1 50 0.9739 40 0.9739 30 1 20 0.9739 10 0.9739 5 0.9035 1 0.8238

Fig (4) The Original peppers image

Fig (5) The original Baboons image Table 2 The result for Peppers image ompression Quality % Normalized Correlation Coefficient 100 1 90 0.9739 80 0.9739 70 1 60 1 50 0.9739 40 1 30 1 20 1 10 0.9493 5 0.9258 1 0.8424

Figure (6) The Peppers Watermarked Image with PSNR=34.4946

Figure 7 The Baboon Watermarked Image with PSNR=33.6448 Figure 8 The Peppers Watermarked Image with Compression Quality 1%

Figure 9 The Baboon Watermarked image compressed to Compression Quality to 1%

Figure 10 Extracted Watermark for Peppers with 1% JPEG Compression Quality Figure 11 Extracted Watermark for Baboon with 1% JPEG Compression Quality References Bender, W. Gruhl, D. and Morimoto, N., Feb. 1995 ,“Techniques for data hiding,” Proc. SPIE, vol. 2420, p. 40. Brassil, J. T. Low, S. Maxemchuk, N. F.and O’Gorman, L.Oct. 1995, “Electronic marking and identification techniques to discourage document copying,” IEEE J. Select. Areas Commum., vol. 13, pp. 1495–1504. Bruyndonckx, O. Quisquater, J. J. and Macq, B. June 1995, “Spatial method for copyright labeling of digital images,” in Proc. IEEE Nonlinear Signal and Image Processing, pp. 456–459. Cox, I. J. Kilian, J. Leighton, T. and Shammoon, T. 1995, “Secure spread spectrum watermarking for multimedia,” Tech. Rep. 95-10, NEC Res. Inst., Princeton, NJ. Darko Kirovski and Fabian A. P. Petitcolas, 2003, “Blind Pattern Matching Attack on Watermarking Systems,” IEEE Transactions on Signal Processing, Volume 51, Number 4, pages 1045-1053. Emek Serkan, Pazarci Melih, (2005),"A Cascade Dwt-Dct Based Digital Watermarking Scheme", European Signal Processing Conference (Eusipco-2005), Ieee-Eurasip Workshop On Nonlinear Signal and Image Processing, NSIP'99 and Balkan Conference on Communications and Circuits and Signal Processing. Evelyn Brannock, Michael Weeks, Robert Harrison, 2008," Watermarking withWavelets: Simplicity Leads to Robustness", Southeastcon, 2008. IEEE, pp. 587-592 Hartung F. and Kutter, M.Stefan Katzenbeisser and Fabien Petitcolas, A. P., 2000, "Information Hiding Techniques for Steganography and Digital Watermarking", Artech House. Koch E. and Zhao, J.June 1995, “Toward robust and hidden image copyright labeling,” in Proc. IEEE Nonlinear Signal and Image Processing, pp. 452–455. Mohammad Reza Soheili, 2008," Redundant Watermarking Using Wavelet Packets", Computer Systems and Applications, 2008. AICCSA 2008. IEEE/ACS International Conference on Volume , Issue , March 31 2008-April 4 2008 Page(s):591 – 598 Moumita Saha, Monu Kedia, and Indranil Sen Gupta, 2007, " A Robust Digital Watermarking Scheme for Media Files", TENCON 2007 - 2007 IEEE Region 10 Conference Volume , Issue , Oct. 30 2007- Nov. 2 2007 Page(s):1 – 4 Petitcolas, F.A.P., September 2000, “Watermarking Schemes Evaluation” IEEE Signal Processing Magazine, Volume 17, pages 58-64. Pitas and I. Kaskalis, T.H.June 1995, “Applying signatures on digital images,” in Proc. IEEE Nonlinear Signal and Image Processing, pp. 460–463. Scott E. Umbaugh, 1998, "Computer Vision and Image Processing: A Practical Approach Using CVIPtool", Prentice Hall PTR. Tsz Kin Tsui, Xiao-Ping Zhang ,Dimitrios Androutsos, March 2008,"Color Image Watermarking Using Multidimensional Fourier Transforms", IEEE Transactions on Information Forensics and Security, Vol. 3, No. 1. Walton, S.Apr. 1995, “Image authentication for a slippery new age,” Dr. Dobb‘s J., pp. 18–26.