EMPHASIZED DIGITAL WATERMARKING ON SINGULAR VALUE DECOMPOSITION IN FREQUENCY DOMAIN

Abstract

Watermarking is an old concept that has been utilized since the early ages of human history in different forms and for a variety of purposes. Digital Watermarking, means of hiding/inserting a message, which can be an image, audio, video or text within the digital media. This hidden/inserted message can later be extracted or detected for variety of purposes. The hiding/inserting process is done in such a manner that it must not cause serious degradation to the original digital media. Although Digital Watermarking is existing for quite a long time, infact for atleast several centuries but it has gained widespread popularity as a research topic in the latter half of the 1990's. Earlier, the meaning and purpose of the watermarks are uncertain. They might have been used for practical functions such as identifying the molds on which sheets of papers were made, or as trademarks to identify the paper maker. On the other hand, they might have represented mystical signs, or might simply have served as decoration. In the past decade, the global rife access of internet technologies makes the com-munication and circulation of digital multimedia contents like images, audio and video very easy. However, this convenience also causes substantial increase in illegal oper-ations such as duplication, modification, forgery, copy-right protection and others in digital media. Therefore, the protection of digital media has become an imperative issue. As the possible solution, Cryptography (or encryption techniques), steganog-raphy and digital watermarking come to our help. However these terminologies have 11 their own philosophies, requirement and limitations. Among these, digital water-marking has drawn much attention of research community to resolve these pressing problems due to its advantages. The major drawback of encryption techniques is that once the data is decrypted by an intruder/attacker, there is no way to protect the data and track its illegal distribution. However, digital watermarking tries to overcome this particular limitation of encryption techniques by providing value-added protection in the form of authentication of data. The major drawback of steganography is that the inserted message is removed easily after any intentional or un-intentional activity of intruder/attacker. On the other hand, watermarking is intended by its develop-ers as the solution to make steganographic process robust to intentional activity of intruder/attacker. The aim of the present thesis is to take the readers through a series of discussions that describe, analyze, explain, and hypothesize about digital watermarking and its usage. In literature, most of the researchers have concentrated their efforts on in-serting a randomly generated Gaussian noise type watermark into the digital media whereas the methodologies proposed in this thesis involve the insertion of meaningful grayscale images/logos as watermark. This thesis, comprising of ten chapters, is con-cerned with the development and presentation of digital watermarking schemes oper-ating in frequency domain and emphasized on singular value decomposition. Chapter 1 gives an overview on the methodology of digital watermarking, as well as on its requirements, classification, applications, and a categorization of attacks on water-marking techniques. A brief literature review of the existing watermarking schemes and organization of the thesis is also presented. The introduction to the necessary concepts, definitions and backgrounds required for this research is presented in chap-ter 2. In chapter 3, a novel reference watermarking scheme is introduced where the watermark embedding stage involves the zeroing of higher frequency coefficients. The frequency coefficients are obtained by the means of wavelet transform and zeroing is iii done by considering directive contrast. Simulation and analysis are presented to evaluate and demonstrate the improved and superior performance of the approach over existing reference watermarking schemes. In chapter 4, a robust reference watermarking scheme is presented by exploring the use of wavelet packet transform in digital watermarking. This chapter also pro-poses a new variant of singular value decomposition namely bi-diagonal singular value decomposition. Experimental evaluation demonstrates that the proposed scheme can withstand a variety of attacks and the superiority of the proposed method is carried out by the comparison with the existing methods. In chapter 5, a new variant of fractional Fourier transform namely multiresolu-tion fractional Fourier transform and the normalized mass matrix for an image are developed and used for watermarking. The technique is shown to have superior perfor-mance and demonstrates robustness to a broad class of intensional or un-intensional degradation. In chapter 6, another framework for reference watermark is proposed using frac-tional wavelet packet transform. The core idea is to create a reference image via fractional wavelet packet transform and a secret rule which is only known to the owner/creator. The feasibility of this method and its robustness against different kind of attacks are verified by computer simulations. In chapter 7, a multipurpose watermarking scheme is proposed. The meaning of the word "multipurpose" is to make the proposed scheme as single watermarking scheme (SWS) or multiple watermarking scheme (MWS) according to our requirement and convenience. To achieve the desired goal, Essentially Non-Oscillatory Point-Value Decomposition and space filling curve are used. The feasibility of this method and its robustness against different kind of attacks are verified by computer simulations and experiments. In chapter 8, encryption and watermarking techniques are integrated for the com-plete protection to improve the authentication of the images. To achieve the desired iv goal, fractional wavelet transform is used. The feasibility of this method and its robustness against different kind of attacks are verified by computer simulations. In chapter 9, a new aspect of watermarking is presented. All above mentioned chapters use a watermark whose size is very small when compare to host image. On the contrary, this is an attempt in which a new watermarking scheme is pre-sented where the size of host image is very small when compare to watermark image. Moreover, three new spectral decomposition (Hankel, Circulant and Topelitz singular value decomposition) are also proposed. Experimental evaluation demonstrates that the proposed scheme can withstand a variety of attacks. In chapter 10, the salient contributions of the work described in this thesis are given with the future perspectives of work in this field.