Ph.D Thesis

What the mind of man can conceive and believe, it can achieve.

- Napolean Hill 

Institute: Department of Mathematics,

                 Indian Institute of Technology Roorkee, India.

Title: Emphasized Digital Watermarking on Singular Value Decomposition in Frequency Domain

Description: In the past decade, the global rife access of internet technologies makes the communication and circulation of digital multimedia contents like images, audio and video very easy. However, this convenience also causes substantial increase in illegal operations 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), steganography and digital watermarking come to our help. However these terminologies have their own philosophies, requirement and limitations. Among these, digital watermarking 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 developers 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 inserting a randomly generated Gaussian noise type watermark into the digital media whereas the methodologies proposed in this thesis involve the insertion of meaningful gray-scale images/logos as watermark. This thesis, comprising of ten chapters, is concerned with the development and presentation of digital watermarking schemes operating 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 watermarking 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 chapter 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 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 proposes 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 multiresolution fractional Fourier transform and the normalized mass matrix for an image are developed and used for watermarking. The technique is shown to have superior performance and demonstrates robustness to a broad class of intentional or un-intentional degradation.

In chapter 6, another framework for reference watermark is proposed using fractional 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 complete protection to improve the authentication of the images. To achieve the desired 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 presented where the size of host image is very small when comparing 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.

Supervisor: Dr. R. Balasubramanian, 

                      Associate Professor, 

                     Department of Mathematics, 

                     Indian Institute of Technology Roorkee, India. 

Examiner: Dr. Prabir Kumar Biswas, 

                   Professor, 

                   Department of Electronics & Electrical Communication Engineering,

                   Head of the Centre,  Computer & Informatics Centre, 

                   Indian Institute of Technology Kharagpur, India. 

Duration: Registration: 27th July 2006

                 Submission:   30th September 2009

                 Defense:         16th August 2010

Financial Support during PhD: Council of Scientific and Industrial Research (CSIR), New Delhi