ISSN: 2277-3754 ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 11, May 2014 Watermarking Of Image Signal Based On Discrete Cosine Transform (DCT) 1Huma Vohra, 2Rajiv Dahiya 1 (M-Tech Scholar, PDM College of Engineering, Bahadurgarh, Haryana) 2 (Assistant Professor, PDM College of Engineering, Bahadurgarh, Haryana)

Abstract: - In this paper, we present a digital has been inserted into the host signal should be invisible watermarking techniques based on the discrete cosine by the human eye. Usually each watermarked signal may transform. This is a type of frequency domain watermarking be subjected to intentional or unintentional attacks where that uses the approach of transform domain signal. This method operates in the frequency domain embedding a the signal is chased and an attempt is made to alter or watermark image in a selected set of DCT coefficients. remove the watermark from the watermarked signal. Watermark casting is performed by exploiting the masking Filtering, adding noise, and geometric distortions are characteristics of the Human Visual System, to ensure among these attacks. All video watermarking methods are watermark invisibility. We proposed a 2-dimension image categorized into two domains: uncompressed domain [8]- watermarking algorithm in DCT domain. In order to embed [12] and compressed domain [13], [14]; in which the an image watermark, we split the image watermark into blocks and transform them into DCT domain. Then these DCT uncompressed domain can in turn be divided into spatial coefficients are quantized and adjusted. Finally, we choose the domain [12] and transform domain [8]-[11]. non-zero DCT coefficients of each block to constitute the In this paper, we present a digital watermarking watermark. We split the cover image into blocks and classify techniques based on the discrete cosine transform. This is these blocks based on the human visual system, and transform a type of frequency domain watermarking that uses the them into DCT domain. Experimental results demonstrate that approach of transform domain signal. In order to embed the watermark is robust to most of the signal processing techniques and geometric distortions. Results show the an image watermark, we split the image watermark into performance of DCT based watermarking of gray scale image blocks and transform them into DCT domain. Then these and color image in the presence of image reshaping and JPEG DCT coefficients are quantized and adjusted. compression with 5% and 10% quality. Simulation results The rest of the paper is organized as follows: In reveal that the DCT watermarking outperforms over random Section II, basic steps of digital watermarking is water marks. explained with visible and invisible watermarking. In

section III, attacks in digital images are presented Index Terms: - Image watermarking, Discrete cosine compression and rotation and rescaling is explained. transform, frequency domain watermarking, JPEG Section IV presents the DCT based watermarking. compression. Different steps used with digital watermarking are

presented. Section V, simulation results will be explained I. INTRODUCTION with the help of graphical representation with different A watermarking system can be viewed as a attacks. Section VI, conclusions will be put forward. communication system consisting of three main elements: an embedder, a communication channel and a detector. II. DIGITAL WATERMARKING To be more specific, the watermark information is embedded within the host signal before the watermarked Digital watermarking is the process of embedding signal is transmitted over the communication channel, so information into a digital signal. In visible watermarking, that the watermark can be detected at the receiving end, the information is visible in the picture or video. that is, at the detector. Digital watermarking is a good Typically, the information is text or a logo which technique to prevent illegal distributions of multimedia identifies the owner of the media [7]. In invisible data [1]. Digital watermarking is the imperceptible watermarking, information is added as digital data to insertion of copyright information into multimedia data audio, picture or video, but it cannot be perceived as such. [4], where the information remains detectable as long the An important application of invisible watermarking is to quality of the content itself is not rendered useless. It is copyright protection systems, which are intended to commonly assumed that digital watermarking is the only prevent or deter unauthorized copying of digital media one of several measures that has to be combined to build [5]. a good copy protection mechanism [4]. All digital watermarking schemes could possibly Although there are a variety of digital watermarking partake in the same generic principal of the watermarking methods but the performance of any digital watermarking implementation which are the two watermarking systems must be evaluated on merits such as: transparency, and known as embedding and extracting systems. The robustness, capacity, security, and rapid detection. In a scheme’s input is the watermark itself and it can be such good watermarking algorithm [3]-[7], the watermark that an image or a secret key. The digital watermark can be 274

ISSN: 2277-3754 ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 11, May 2014 formed in many different forms such as a text, a number and the locally generated version do not share the same or even an image. The real use of the key the scheme has spatial pattern anymore. is to compel the security which then can prevent those unauthorized parties from manipulating either from IV. DCT BASED WATERMARKING recovering the watermark. The watermark scheme will In this section, we are presenting the watermarking have an output which is the watermarked data. Fig. 1 algorithm based on DCT with both colour and B/W shows the generic scheme for digital watermarking image. technique. A. DCT TRANSFORM BASED EMBEDDING SYSTEM Let’s X is referred to the original image (cover image),

and its size N1 X N2. Moreover, let us suppose Y as a

notation for the watermark image and its size noted as M1

X M2. The size of the watermark image that we donated as Y is supposed to be smaller than the size of the cover image X. Step 1: Both the cover image X and the watermark Y must be divided into block.

Step 2: For the purpose of enhancing the perceptual invisibility, we need to take within consideration the understanding of the original image characteristics. Thus, those divided blocks from the watermark which contain such more details or more complex contents should be in fact embedded into those blocks of the original image Fig. 1 generic scheme for digital watermarking technique which also consist of more information.

III. ATTACKS IN IMAGE SIGNALS Step 3: After breaking the original image X into The distortions are limited to those that do not produce blocks, the DCT transform [15] is then applied and excessive degradations, since otherwise the transformed performed for each block of the image X so that each object would be unusable. These distortions also single block can be DCT transformed independently[16]. introduce degradation on the performance of the system. For intentional attacks, the goal of the attacker is to Step 4: Since we have got the image X and the maximize the reduction in these probabilities while watermark Y converted into the frequency domain as in minimizing the impact that his/her transformation the new resultant image A, all we need to do next is produces on the object; this has to be done without knowing the value of the secret key used in the extracting out the middle-frequency coefficients ( FM) watermarking insertion process, which is where all the from the computed image A. In fact, there is a significant security of the algorithm lies [7]. There are many more cause for the reason of choosing the middle-frequency attacks associated with image signal here we are coefficients of A and it is due to two significant folded as considering compression and scaling and rotating. follow:

Compression: This is generally an unintentional attack 1. The first basis is that the human eye indeed is which appears very often in multimedia applications. more sensitive to whatever noise existing within Practically all the audio, video and images that are the frequency components of the low region ( currently being distributed via Internet have been FL), so it’s more sensitive to the noise in those compressed. If the watermark is required to resist lower frequency components than to the different levels of compression, it is usually advisable to components of the higher frequencies. perform the watermark insertion task in the same domain Therefore, we should not replace the watermark where the compression takes place. into those components where the low frequency Rotation and Scaling: This has been the true battle region is placed. horse of digital watermarking, especially because of its 2. The second basis however is the fact that since success with still images. Correlation based detection and the components of the higher frequencies are extraction fail when rotation or scaling is performed on being affected by means of the quantization the watermarked image because the embedded watermark operation for the JPEG lossy compression, 275

ISSN: 2277-3754 ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 11, May 2014 replacing the watermark into this frequency band In this section, simulation results are presented for of the high region ( FH) may discard the watermarking of gray scale image and color image in the watermark while performing the lossy presence of compressed and rescaled images. compression.

Step 5: for any DCT transformed image that has been A. WATERMARKING THE GRAY-SCALE IMAGES divided into the size of 8x8 block out of 64 coefficients Fig. 2 shows a 256x256 gray-scale image of a horse and fig. 3 shown above has a random watermark there will be only two-middle coefficients chosen. embedded into the original image. Fig. 4 shows the Step 6: Afterwards, we will attain the reduced image Watermark detector response to 1000 randomly generated watermarks. Only one watermark matches that is present as well as obtaining the modified digital watermark which in watermarked gray-scale image. both have the size M1 x M2.

Step 7: The final procedure can be performed by mapping the middle-frequency coefficients we achieved into the resulted image in the frequency domain A which leads us to come by the À. The associated result we have got À would then require the inverse function of the DCT for the purpose of accomplishing the watermarked image X`.

B. DCT TRANSFORM BASED EXTRACTION SYSTEM The proper technique for extracting back the invisible watermark from the watermarked image and analyzing this scheme by such following up stages Step 1: The extracting system requires all of the original image X as well as the resultant watermarked Fig. 2 Gray-scale original image image X` and either the digital watermark.

Step 2: The discrete cosine transform function should be applied to both of the watermarked image and the reference image through the decoding process, so that the images can be DCT transformed after being computed and converted into the frequency domain in consequence of the functionality of the DCT [16].

Step 3: Now since we have already applied the DCT for the purpose of decoding the reference image as well as the watermarked image in virtue of the property thereafter is to generate the reduced image which will help us to get rid of the low frequency and high frequency DCT components as long as that reduced image contains only of the middle frequency coefficients. Fig. 3 Gray-scale watermarked image

Step 4: That generated image (reduced image) which is composed of the middle-band frequencies (FM) will then give us the chance for cultivating the polarity pattern. So then we are almost in the last stage to retrieve back the embedded watermark signature.

V. SIMULATION RESULTS

276

ISSN: 2277-3754 ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 11, May 2014

Fig. 4 Watermark detector response to 1000 randomly Fig. 7 JPEG Compression image with 5% quality generated watermarks. Only one watermark matches that is Fig. 7 shows the JPEG Compression image with 5% present in watermarked gray-scale image quality. Fig. 8 shows the Watermark detector response to B. WATERMARKING WITH IMAGE SCALING AND JPEG compressed version with 5% quality. Fig. 9 shows RE-SCALING the JPEG Compression image with 10% quality. Fig. 10 Fig. 5 shows the resized image. Fig. 6 shows the shows the Watermark detector response to JPEG Watermark detector response to resized gray-scale image. compressed version with 10% quality.

Fig. 5 resized image

Fig. 8 Watermark detector response to JPEG compressed version with 5% quality

Fig. 6 Watermark detector response to resized gray-scale image C. WATERMARKING WITH JPEG COMPRESSION Fig. 9 JPEG Compression image with 10% quality DISTORTION

277

ISSN: 2277-3754 ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 11, May 2014 [7]. A.H. Tewfik, “Digital Watermarking”, in IEEE Signal Processing Magazine, vol 17, pp 17-88, September 2000. [8]. Yan Liu, Jiying Zhao, "A new video watermarking algorithm based on ID DFT and Radon transform," Signal Processing 90 (2010) 626-639. [9]. Radu O. Preda and Dragos N. Vizireanu, "A robust digital watermarking scheme for video copyright protection in the wavelet domain," Measurement 43 (2010) 1720-1726. [10]. Dooseop Chio, Hoseok Do, Hyuk Choi and Taejeong Kim, "A blind Mpeg-2 video watermarking robust to camcorder recording," Signal Processing 90 (2010) 1327-1332. [11]. Alper Koz and A. Aydin Alatan, "Oblivious Spatio- Temporal Watermarking of Digital Video by Exploiting the Human Visual System," IEEE Transactions on circuits and systems for video technology, vol. 18, no. 3, March 2008. Fig. 10 Watermark detector response to JPEG [12]. B. Mobasseri, M. Sieffert and R. Simard, "Content compressed version with 10% quality authentication and tamper detection in digital video,” VI. CONCLUSIONS Proceeding of IEEE International Conference on Image Processing, vol. 1, 2000, pp. 458-461. In this paper, a digital watermarking techniques based on the discrete cosine transform is presented. This [13]. Dengpan Ye, Changfu Zou, Yuewei Dai and Zhiquan method operates in the frequency domain embedding a Wang, "A new adaptive watermarking for real-time MPEG watermark image in a selected set of DCT coefficients. videos,” Applied Mathematics and Computation 185 (2007) 907-918. We proposed a 2-dimension image watermarking algorithm in DCT domain. In order to embed an image [14]. J. Zhang, A. Ho, G. Qju and P. Marziliano, "Robust video watermark, we split the image watermark into blocks and watermarking of H.64/AVC," IEEE Transactions on Circuits and System-II: Express Briefs 54 (February) transform them into DCT domain. Then these DCT (2007) 205-209. coefficients are quantized and adjusted. Finally, we choose the non-zero DCT coefficients of each block to [15]. Barni Mauro, et al., A DCT-domain system for robust image watermarking. Signal Processing, 1998. 66(3): constitute the watermark. We split the cover image into p.357-372. blocks and classify these blocks based on the human visual system, and transform them into DCT domain. [16]. Yang Cheng-Han, Hui-Yu Huang, and Wen-Hsing Hsu. An Experimental results demonstrate that the watermark is adaptive video watermarking technique based on DCT domain. in 8th IEEE International conference on computer robust to most of the signal processing techniques and and information technology. 2008. Sydney. p. 589-594. geometric distortions.

REFERENCES [1]. Furon T. and P. Duhamel. Copy Protection of Distributed Contents: An Application of Watermarking Technique. in Workshop COST 254: Friendly Exchange through the net. 2000. Bordeaux, France. p. [2]. P. Meerwald, A. Uhl, “Watermark Security Via Wavelet Filter Parameterization”, International Conference on Image Processing, Thessaloniki, Greece, 2001 [3]. P. Meerwald, A. Uhl, “A Survey of Wavelet-Domain Watermarking Algorithms” EI San Jose, CA, USA, 2001 [4]. H. Inoue, A. Miyazaki, T. Katsura “An Image Watermarking Method Based on the Wavelet Transform”, Kyushu Multimedia System Research Laboratory. [5]. F.A.P. Petitcolas, , “Watermarking Schemes Evaluation” ”, in IEEE Signal Processing Magazine, Vol 17, pp 58-64, September 2000 [6]. F.A.P. Petitcolas, “Introduction to information hiding” in Information Techniques for Steganography and Digital Watermarking, S.C. Katzenbeisser et al., Eds. Northwood, MA: Artec House, Dec. 1999, pp 1-11 278