Copyright Protecting Using Secure Watermarking

0 downloads 0 Views 491KB Size Report
Jun 17, 2003 - can be seen visually the same, and the visible watermark can only be removed by the user using the correct key and original logo. Experiment ...
Recent Advances in Telecommunications, Signals and Systems

Copyright Protecting Using Secure Watermarking Images in DWT Domain ALI A. ELROWAYATI, ZAKARIA SULIMAN ZUBI, MOHMED A. ABDULALI, Electronic Engineering Department, Computer Science Department, Tele& Networks Department, The College of Industrial Technology, Faculty of Science, Faculty of IT, Misurata-Libya Sirte University Misurata University Sirte, Libya Misurata-Libya Email:[email protected] Email: [email protected] Email: [email protected]

Abstract: - In this paper, a visible and removable watermarking scheme in Discrete Wavelet Transform (DWT) domain is presented. Watermark and original images are decomposed up to 3-levels using discrete wavelet transform. It will be compressed by taken the percentage of largest coefficients from each image depend on compression ratio and quality of reconstructed image. In the scheme, verify the originality of logo image using hash function, with md5 algorithm, and a secure template is generated to modulate the embedding strength of watermark and original image, separately. The template generates different versions of watermarked images which can be seen visually the same, and the visible watermark can only be removed by the user using the correct key and original logo. Experiment results show the visible and removable watermarking algorithm achieves some good performances, such as, robust against the removable attack and explored the HVS model into our algorithm gave better visual quality and better security. Key-Words: - Copyright Protection, Watermarking, Robustness, Discrete Wavelet Transform, Image Compression.

The rise of the internet and computer technology has resulted in many new opportunities for the transmission of multimedia content. With the ease of copy and distribution of content, the copyright protection issue becomes more and more concerned. Digital watermarking protects the copyrights in the way of embedding the watermark into digital multimedia content so that the watermark can later be extracted or detected. Most of the watermarks are invisible [1,2,3] while a few are visible [4,5,6]. Visible watermarking which provides a recognizable identity to the content is used for protecting the publicly available multimedia content. The main advantage of visible watermarking is that it prevents unauthorized using of copyrighted high quality images.

without any watermarking detail at the side. Depending on the scaling factors of HVS the watermarking process is performed by adjusting the pixel value. Based on the difference of image between host image and its approximate version (prediction technique) a reconstruction packet is created for reversibility. Human Visual System (HVS) characteristics are considered to calculate the greater scale factor and lower scale factor. Greater and lower scale factors are assigned to midluminance and textured areas respectively. In [8], Mnajunatha and Shivaprakash developed a robust-invisible watermarking using Haar wavelet transform for copyright protection. Watermark embedding and removal is performed using mask matrix. This matrix is established by MD5 algorithm and random matrix generation using the original image.

In visible watermarking, a secondary image is inserted into the host image such that visible watermark is perceptible to a human observer. Many researches have been done in this area. In [7], Ying et al proposed an invertible recovery of original image

Quality of watermarking scheme is commonly determined by the four factors robustness, imperceptibility, capacity, and blindness. Good quality watermarking scheme should have maximum PSNR (Peak Signal-to-Noise Ratio) and SIMM

1 Introduction

ISBN: 978-1-61804-169-2

89

Recent Advances in Telecommunications, Signals and Systems

different resolution sub images, corresponding to the various frequency bands. This gives better representation of images with localization in both the spatial and frequency domains. This advantage is desirable in image compression, and it is not possible in both Fourier and Discrete Cosine Transforms which give good localization in one domain at the expense of other.

(Structural Similarity). Watermark must be highly robust to distortion introduced during either normal use (unintentional attack), or a deliberate attempt to disable or remove the watermark present (intentional, or malicious attack). SSIM is a widely believed that the statistical properties of the natural visual environment play a fundamental role in the evolution, development and adaptation of the HVS. SSIM is presented as an alternative design philosophy for quantitative image quality assessment methods [14]. In this paper, Assessment and analysis of the reconstruction quality based on an SSIM method used to measure the visual quality of compressed reconstructed images and to be compared with a PSNR method hoping to arrive at different conclusion.

The wavelet transform, in general, produces floating point coefficients. Although these coefficients can be used to reconstruct an original image perfectly in theory, the use of finite precision arithmetic and quantization results in a lossy scheme [12]. Recently, reversible integer wavelet transforms that transform integers to integers and allow perfect reconstruction of the original signal have been introduced [9]. In [10], Calderbank et al. introduced how to use the lifting scheme presented in [11], where Sweldens showed that the computational complexity of convolution based biorthogonal wavelet transforms can be reduced by implementing a lifting based scheme as shown in Fig. 1. Note that only the composition part of wavelet transform is depicted in Fig. 1 because the reconstruction process is just the reverse version of the one in Fig. 1. The lifting based wavelet transform consists of splitting, lifting, and scaling modules and the wavelet transform is treated as prediction error decomposition. It provides a complete spatial interpretation of DWT. In Fig. 1, let X denotes the input signal and X LI and X HI be the decomposed

Unintentional attacks involve transforms that are commonly applied to images during normal use, such as addition of noise, cropping, resizing, contrast enhancement, filtering.etc. In order to be successful, the watermark should be invisible and Robust to premeditated or spontaneous modification of the image. From the literature, most of the watermarks are binary logos, and few methods of color visible removable watermarking are proposed in DWT domain. DWT is used frequently in digital image watermarking due to its Multi-resolution property i.e. time (space)/frequency decomposition characteristics, which resemble to the theoretical models of the human visual system [9]. In this paper, we aim to invent a secure visible and removable watermarking scheme based on DWT Domain. The organization of this paper is as follow. Section 2 focus on the DWT coding. Sections 3 gives an overview of the watermarking algorithms proposed. Section 4 discussed the experimental results are shown. Finally the paper concludes in Section 5.

output signals, where they are obtained through the following three modules of lifting based D-DWT: (1) Splitting: In this module, the original signal X is divided into two disjoint parts, i.e., xe (n) = x(2n) and x0 = x (2n + 1) that denotes all even-indexed and odd indexed samples of X , respectively.

2 Discrete Wavelet Transform Wavelet theory [9] has provided a promising hope for image processing applications because of its flexibility in representing images and its ability in taking into account human visual system characteristics. It is mainly used to decorrelate the image data, so the resulting coefficients can be efficiently coded. It also has good energy compaction capability which results in a high compression ratio. A wavelet transform decomposes an image into a set of

ISBN: 978-1-61804-169-2

(2) Lifting: In this module, the prediction operation P is used to estimate x0 ( n) from

xe (n) and results in an error signal d (n ) which represents the detail part of the original signal. Then we update d (n ) by applying it to the

90

Recent Advances in Telecommunications, Signals and Systems

decomposition, most of the coefficients with high frequency (low scale) region are either zero or in close proximity to zero. Hence, most of significant coefficients can be extracted and coded by applying strategies such as applying thresholding. It will be compressed by taken the percentage (R%) of largest coefficients from each image depend on compression ration and quality of reconstructed image.

update operation U and the resulting signal is combined with xe (n ) to estimate s (n ) which represents the smooth part of the original signal. (3) Scaling: A normalization factor is applied to d (n ) and s (n ) , respectively. In the even indexed part is multiplied by a normalization factor K e to produce the wavelet sub band X LI . Similarly in the odd index part, the error signal d (n ) is multiplied by K o to obtain the wavelet

3.1 Watermark Embedding Algorithm

sub band X HI .

Fig.2 Block Diagram of Watermark Embedding Algorithm

The watermark Embedding procedure is shown in Fig 2 and described in details in the following steps. Step 1:- Apply DWT to the original and water mark images to get sub bands coefficients at third level decomposition..

Fig. 1. The lifting-based wavelets transform.

DWT is applied on an image in order to reduce the inter pixel redundancy [13]. As a result of decomposition, most of the coefficients with high frequency low scale region are either zero or in close proximity to zero. Hence, most of significant coefficients can be extracted and coded by applying strategies such as designing a JPEG like quantization table or applying threshold. Threshold parameter value is chosen intuitively based on experimentation and satisfactory visual effect of reconstructed image as reported in [9]. The significance of coefficients is directly related to its magnitude as well as their sub bands after wavelet decomposition.

Step 2:- compress them by taken the percentage (R%) of largest coefficients from each image depend on compression ration and quality of reconstructed image. Step3 :- Generate a secure template with two parts, a public key and hash function. Step4:-Add the output from step2 and step3 to produce the watermarked image

3.2 Watermark Extraction Algorithm

3 Watermarking Algorithms

The watermark extraction procedure is shown in Fig 3, and described in details in the following steps:

The host and original images are NxN color images. We Apply DWT to them to get sub bands Coefficients at third level decomposition, as a result of

ISBN: 978-1-61804-169-2

91

Recent Advances in Telecommunications, Signals and Systems

However, in case of unauthorized access with an incorrectly removed with the (wrong key or Unoriginal logo) is also indicted in this paper.

Fig.3 Block Diagram of Watermark Extracting Algorithm Lena(a)

Step 1: Perform DWT on the watermarked image and original logo at third level decomposition.

Logo(c)

Fig 4 Watermark original images.

Step2: thresholding, compress original logo by taken the percentage (R%) of largest coefficients, using the same R% which used in the watermark embedding procedure. Step3: verify the originality of logo image using hash function with md5 algorithm. Step4: If the logo is original, then using the same key which used in the watermark embedding procedure. Otherwise it will be refused. Step5: Add the output from step4 and step1 to produce the reconstructed image. Step6:- Perform the inverse DWT (IDWT) on reconstructed image and display it.

Fig 5 and Table 1 illustrates some results of watermarking removing using the correct key and original logo. Table1 showns the PSNR and SIMM increase, when users with the right key and original logo can remove the watermark, recover the original image. Tab1 : watermarking removing using the correct key and original logo with different Compression Ratio. Reconstructed Compression PSNR SIMM Image Ratio (db) Lena 41.70 0.9860 %90 Pepper %90 41.54 0.9852 46.88 0.9960 Lena %80 Pepper 46.42 %80 0.9958

4 Experimental Results Our experiment includes testing the performance of the proposed algorithm with a number of original images shown in Fig 4. We used the PSNR and SSIM to measure the security performance of the original images. The watermark image is 24-bit images and the original image is a 512×512 24-bit image. In the proposed scheme, the security of watermark will be considered as a robust against illegally removable attack. In the proposed scheme also, the visible watermark removing depends on the template under the control of the secret key and the original image of the logo. The visible logo should be correctly removed from the watermarked image with the correct key and original logo in case of an authorized access with a correct key access.

ISBN: 978-1-61804-169-2

Peppers(b)

Correct Result (vegetables) Correct Result (Lena) Fig 5. Watermark removing with a correct access with a correct key and correct original logo at Different Compression Ratio.

Fig 6 and Table 2 give some results of watermarking removing while using wrong key or Unoriginal logo. Table2 shown the PSNR and SIMM decrease, when

92

Recent Advances in Telecommunications, Signals and Systems

visible watermark for the purposed that the embedding visible logo can be removed from the watermarked image after verify the originality of logo image using hash function, with md5 algorithm. Experiments show that proposed scheme is robustness against removable attack. Users with the right key and correct original logo can remove the watermark, and recover the original image. However, with the wrong key and/or unoriginal logo, the remaining pixels of the logo are left on the host image of which affects the visual quality. In the addition, we explored the HVS model or SIMM into our algorithm gave a better visual quality and better security.

users with wrong key and/or unoriginal logo cannot remove the watermark, and recover the original image. Tab 2 : watermarking removing using the wrong key and/or unoriginal logo with different Compression Ratio. Reconstructed Image Lena Pepper

Lena Pepper

Compression Ratio %90 %90 %80 %80

PSNR (db) 18.36 18.84 19.93 19.81

SIMM 0.7864 0.7791 0.7981 0.8012

References: [1] I. Cox, J. Kilian, F. Leighton, and F. Shamoon, “Secure spread spectrum watermarking for multimedia”, IEEE Transactions on Image Processing, Vol. 6, No. 12, Dec. 1997, pp.1673-1687. Wrong Result (Lena)

Wrong Result (vegetables)

Fig 6. watermarking removing using the wrong key and/or unoriginal logo with different Compression Ratio.

By using the correct key and original logo, the image is well recovered and the watermark is removed successfully. However in case of an illegally removing, the much energy residue of the visible watermark still exists and makes the image less valuable. The correct access of the PSNR and SIMM will drive to an authorized access to the original images. Any unauthorized access to the images with a non match number of our PSNR and SIMM number will drive to a wrong access to the images. Using a correct PSNR and SIMM scheme obtains a high security for the images. If more perceptible pixel is left after the watermarking it analyze the relation between the security and the interval.

5 Conclusions In this paper, a secure visible and removable colorwatermarking scheme adopted in DWT domain is proposed. A secure template is generated which modulates the coefficients of original image and

ISBN: 978-1-61804-169-2

[2] Pan Yinghui, “Digital Watermarking Particle Swarm Optimization Based on Multi-wavelet”, JCIT: Journal of Convergence Information Technology, Vol. 5, No. 3, pp. 38-45, 2010. [3] Yingkun Hou, Chunxia Zhao, Yong Cheng, Zhengli Zhu, “Image Watermarking Resynchronization to Geometric Attacks in DWT Domain”, JDCTA: International Journal of Digital Content Technology and its Applications, Vol. 4, No. 4, pp. 88-98, 2010 [4] MS Kankanhalli, KR Ramakrishnan, “Adaptive Visible Watermarking of Images,” Proc. IEEE Int l Conf. Multimedia Computing and Systems, IEEE CS Press, 1999, pp. 68-73. [5] Yasuyuki Nakajima, Akio Yoneyama, and Yoshinoti Hatori, “A Fast Logo Insertion Algorithm for MPEG Compressed Video”, Consumer Electronics, 2003. ICCE. 2003 IEEE International Conference on Publication Date: 17-19 June 2003 pp.38-39, 2003 [6] Yongjian Hu, Sam Kwong, Jiwu Huang, “An Algorithm for Removable Visible Watermarking”, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 16, No. 1, pp. 129-133, 2006. J. Meng, S.F

93

Recent Advances in Telecommunications, Signals and Systems

[7] Ying, Y., X. Sun, H. Yang, C.T. Li and R. Xiao, "A Contrast-Sensitive reversible visible image watermarking technique", IEEE Transactions on Circuits and Systems, No. 9, pp.656-667, 2009.

Assessment: From Error Visibility to Structural Similarity”, IEEE Trans. Image processing, vol. 13, NO. 4, April 2004,

[8] Mnajunatha, P.R. and K. Shivaprakash,"A Robust wavelet-based watermarking scheme for copyright protection of digital images", IEEE Second International Conference on Computing, Communication and Networking Technologies, pp:19,2010. [9] Ali A. Elrowayati, Jamal R. Elbergali, “Medical Images Compression using Lifting Wavelet and Adaptive Wavelet Back Propagation Neural Networks”, Mosharaka International Conference on Communications, Propagation and Electronics (MICCPE2010), Amman Jordan, 2010. [10] D. Donoho, I. Johnstone, G. Kerkyacharian, and D. Picard, “Density estimation by wavelet thresholding,” Technical Report Stanford University, 1993. [11] Shaou-Gangmlaou, Shih-Tse chen, Shu-Nien chao" Wavelet-based lossy-to-Lossless Medical Image Compression using dynamic VQ And SPIHT Coding," , Biomedical Engineering Application ,Basis & Communication., Vol. 15 No. 6,pp. 235- 242, December 2003. [12] Zakaria Suliman Zubi, Ali A. Elrowayati, IBTCDWT Hybrid Coding of Digital Images, the Proceedings of the 3rd International Conference on APPLIED INFORMATICS and COMPUTING THEORY (AICT '12) Barcelona, Spain, October 1719, 2012. [13] I. El-Fegh, Zakaria Suliman Zubi, Ali A. Elrowayati, and Faraj A. El-Mouadib. 2009. Handwritten Arabic words recognition using multi layer perception and Zernik moments. In Proceedings of the 10th WSEAS international conference on evolutionary computing (EC'09), Nikos E. Mastorakis, Anca Croitoru, Valentina Emilia Balas, Eduard Son, and Valeri Mladenov (Eds.). World Scientific and Engineering Academy and Society (WSEAS), Stevens Point, Wisconsin, USA, 46-51. [14] Zhou Wang, Alan Conrad Bovik, Hamid Rahim

Sheikh and Eero P. Simoncelli,

ISBN: 978-1-61804-169-2

“Image Quality

94