CHAPTER 1

INTRODUCTION
1.1 DIGITAL IMAGE PROCESING

Digital image processing is a rapdily developing area with various raising application in computer
science and engineering. It is very important field for the research work because its techniques are
used in almost all kinds of tasks like human computer interface, medical, visualization, image
enhancement, Law enforcement, artistic effects, image restoration and digital watermarking for
security purpose.

Digital image processing has many beneficial properties over the analogue image processing.
Digital image processing is accomplishing variant computer operation on digital image for various
purposes like enhancing image quality, filtering images from noise. A digital image is a
representation of two dimensional images as a finite set of digital values called picture elements or
pixels. Therefore, processing a digital image by using a digital computer is called digital image
processing.

The digital communication technology, like internet technology confronts various troubles related to
the privacy and security of the data. Security techniques are required because of illegal access of data
without permission. Therefore, it is necessary to protect data in the internet technology. For providing
the security of digital data various techniques are used like encryption, decryption, cryptography,
steganography and digital watermarking. The digital watermarking is an application of the digital
image processing.

1.2 DIGITAL IMAGE WATERMARKING

The digital watermarking is a process of information hiding. There are various techniques for hiding
the information in the form of digital contents like image , text, Audio and video. Basically digital
watermarking is a method for embedding some secret information and additional information in the
cover image which can later be extracted or detected for various purposes like authentication, owner
identification, content protection and copyright protection, etc. Sometimes the scaling factor is used
for the security of the digital content and to protect the data from illegal users and provides the
ownership right for digital data.

An important characteristic of digital watermarking is robustness and imperceptibility against

various types of attacks or common image manipulation like rotation, filtering, scaling, cropping and

1
compression. The efficiency of digital watermarking algorithms is totally based on the robustness of
the embedded watermark against various types of attacks. Digital watermarking method is also used
for the tamper proofing and authentication.

Digital watermarking is a very developing field and used in various application which have been
proved to be successful. The aim of every application is to providing security of the digital content.
The digital watermarking application is Broadcast Monitoring, Digital Fingerprinting, Transaction
Tracking, Copyright Protection, Temper Detection, Data Hiding and Content Authentication etc. In
digital watermarking there is an input image(host image) and an watermark image. There are two
processes occur in digital watermarking i.e. embedding process and Extracting Process. When
Watermark image embed on the cover image for providing security to cover image than this process
is known as Embedding Process and then watermarked image is transmit over the communication
channel. In this stage, when the data is transmitted over the network, some noise is added on the
watermarked image or some attacks are performed on the watermarked image. So, our watermarked
data is either modified or destroyed. The Embedded Watermarked is recovered in Extracted Process.
In Digital Watermarking has two domains, spatial domain and transform domain(frequency
domain). In spatial domain method, the watermark bits are embedded directly into the pixels of cover
image. The spatial domain watermarking methods are simpler and are less robust against various
geometric and non geometric attacks.
In transform domain, the watermark is embedded by changing the coefficient magnitude in a
transform domain using Discrete Wavelet Transform(DWT), Discrete Cosine Transform(DCT),
Singular Vector Decomposition(SVD) .
Transform domain methods can yield more information embedding and more robustness against
many common attacks. But the computational cost is higher than spatial-domain watermarking
methods.
Process of Robustness and optimization that are explained in this:

1. First of all an input image and an watermark image is chosen. We apply DWT on the cover image.
DWT divides image into four sub bands i.e. LL (Low Low), HL (High Low),LH (Low High),HH
(High High) and then LL sub band is again split into sub bands i.e. LL2, HL2, LH2, HH2 known as
2-DWT. After that, apply SVD to smallest LL2 sub band that split the image into u, s, v matrix form
and find out scaling factor of the cover image. Like this, Again apply DWT and SVD on the
watermark image.
2 The Selection of fitness function is based on to the optimize Robustness. PSNR value also has
major role in this paper. Higher the PSNR value provides good quality image.
2
1.2 Advantage Of Digital Image Watermarking:

Robustness
Robustness refers to the schemes resistance against various attacks. Examples of such attacks may be
image processing attacks such as filtering, addition of noise, histogram manipulation, contrast
enhancement etc, and geometric attacks such as cropping, rotation, translation etc.

Imperceptibility
All schemes cannot resist all kinds of attacks, hence schemes are usually made application
dependent. Imperceptibility gives the similarity between the original image and the watermarked.

Security
Security is used to describe the resistance of a scheme against malicious attacks. Many schemes may
be highly robust but may not show security. For instance many
SVD based schemes have security issues due to the false positive problem.

1.3 Disadvantage Of Digital Iamge Watermarking:

Quality decrease
When embed on image decrease its quality.

Reduces social sharing

Social networking sites help spread information and news but your photo are heavily watermarked
they may not receive much attention.

3
1.5 SOFTWARE INTRODUCTION

MATLAB:

1.5.1 Introduction Of Matlab

The name MATLAB stands for MATrix LABoratory. MATLAB was written originally to provide
easy access to matrix software developed by the LINPACK (linear system package) and EISPACK
(Eigen system package) projects.

The MATLAB language provides native support for the vector and matrix operations that are
fundamental to solving engineering and scientific problems, enabling fast development and
execution. MATLAB provides features of traditional programming languages, including flow
control, error handling, and object-oriented programming (OOP). You can use fundamental data
types or advanced data structures, or you can define custom data types.

MATLAB add-on products provide built-in algorithms for signal processing and
communications, image and video processing, control systems, and many other domains. By
combining these algorithms with your own, you can build complex programs and applications

1.5.2 History

Developed primarily by Cleve Moler in the 1970's

Derived from FORTRAN subroutines LINPACK and EISPACK, linear and eigenvalue
systems.
Developed primarily as an interactive system to access LINPACK and EISPACK.
Gained its popularity through word of mouth, because it was not ocially distributed.
Rewritten in C in the 1980's with more functionality, which include plotting routines.
The MathWorks Inc. was created (1984) to market and continue development of MATLAB.
According to Cleve Moler, three other men played important roles in the origins of MATLAB.J. H.
Wilkinson, George Forsythe, and John Todd. It is also interesting to mention the authors of
LINPACK: Jack Dongara, Pete Steward, Jim Bunch, and Cleve Moler. Since another package
emerged: LAPACK. LAPACK stands for Linear Algebra Package. It has been designed to supersede
LINPACK and EISPACK.
4
1.5.3 Strengths:
MATLAB may behave as a calculator or as a programming language
MATLAB combine nicely calculation and graphic plotting.
MATLAB is relatively easy to learn 62
MATLAB is interpreted (not compiled), errors are easy to x
MATLAB is optimized to be relatively fast when performing matrix operations
MATLAB does have some object-oriented elements

1.5.4 Weaknesses:
MATLAB is not a general purpose programming language such as C, C++, or FOR-TRAN
MATLAB is designed for scientificc computing, and is not well suitable for other appli-
cations
MATLAB is an interpreted language, slower than a compiled language such as C++.
MATLAB commands are specic for MATLAB usage. Most of them do not have a direct
equivalent with other programming language commands.

1.5.5 Competition:

One of MATLAB's competitors is Mathematic, the symbolic computation program.

MATLAB is more convenient for numerical analysis and linear algebra. It is frequently
used in engineering community.
Mathematic has superior symbolic manipulation, making it popular among physicists.

5
 CHAPTER 2
LITERATURE SURVEY
2.1 Related Work
Hiren Joshi , Ashish [1] The proposed work is in similar direction where in RMI (Random Matrix
Image) is used in place of Huffman coding. This algorithm increased the robustness in digital
watermark while also enhancing security of production. The experiment part has used random
matrix as a watermark to prevent an attacker for easy attack on the water mark image. The reason for
this is that each image usually has different matrix form with range from 0 to 10. This work is the
use of RMI transition to authenticated user of the image. Without having RMI, none other than
authenticated user can detect and extract watermark from watermarked image.
The limitation of the current work is that the stated technique of digital water marking is used
on grey scale image and not on the coloured image. The experiment performed on greyscale image
having less than 245 greyscale pixel value.

Anumol Joseph1 , K. Anusudha[2] This paper deals with a new DWT and SVD watermarking
technique in which two watermark images are embedded in the HL and LH bands of the host image
after two level DWT decomposition of the host image using HAAR wavelet by modifying the
singular values of the host image with that of watermark images. The performance was evaluated
using RMSE, PSNR and correlation coefficient. Simulation results show that this method is robust
against various types of attacks.

Podili Brahma Srinu , B.J agadeesh [3] The proposed algorithm is a robust, blind digital image
watermarking method based on Fuzzy- BPNN system in hybrid transform domain i.e. combination of
both DCT and DWT transform techniques. Experimental results shows that the proposed method is
robust to many processing attacks by giving better NCC values and for transparency it yields good
PSNR values. Median filtering and Resizing attacks give less NCC values when compared to other
processing attacks.

Divjot Kaur Thind, Sonika Jinadal[4],In this paper a new watermarking is used which combines
DCT(Discrete Cosine Transform), DWT(Discrete Wavelet Transform), SVD(Singular Value
Decomposition) in which watermarking is performed on high frequency band and then various
attacks have been applied.

6
Madhuri Rajawat , D S Tomar[5],This paper presents techniques, applications, classifications,
attacks and tampering detection in watermarking. This paper improves the security of image by
combining new algorithm for digital watermarking and tampering detection. Authors worked on
RGB components for enhancing security and robustness.

Aparna J R, Sonal Ayyappan[3], It proposes a block based image watermarking algorithm in which
cryptographic algorithm are used to tell about the positions of the cover(original) image in which the
watermark is to be embedded.
Two different keys are generated using Diffie Hellman Key Exchange algorithm and using these keys
the positions of cover(original) image to which the watermark bits are to be embedded.

Preeti Parashar, Rajeev Kumar Singh[4], The authors have given an survey of existing digital
image watermarking techniques. The results of various techniques have been compared on the basis
of outputs.

Lalit Kumar Saini, Vishal Shrivastava[5],This paper includes the following sections like overview
of watermarking, Types of watermarking, Classification and applications of image watermarking and
threats and performance evaluation in image watermarking.

Ms.Mahejabi Khan, Mr. Ajay Kushwaha, Mr. Toran Verma[6], Introduced a secure and robust
watermarking algorithm based on the combination of image interlacing, DWT & DCT techniques

7
 CHAPTER 3
OBJECTIVES OF THESIS

From the conclusions of literature survey, the main objectives of our work are:
1. Survey of existing recent image techniques water marking.
2. Propose a optimal scaling matrix generation algorithm using differential evaluation.
3. Implementation proposed algorithm in matlab.
4. Comparatively analysis for proposed algorithm exiting work.

8
 CHAPTER 4

METHODOLOGY

4.1 Research Methodology

A. Discrete Wavelet Transform:
Discrete Wavelet transform (DWT) is an mathematical tool for decomposing an image. The
transform is based on waves, and these waves are small called wavelets, of frequency that are varying
and limited duration. The DWT divides the input image into four sub-bands LL (Low-Low), LH
(Low-High), HL (High-Low) and HH(High-High). The LL (Low-Low) sub band represents the
coarse-scale coefficients of DWT while the LH, HL and HH sub bands represent the fine scale
coefficients of DWT. For next coarser scale of wavelet coefficients, the LL sub band are further
processed until the size of cover image and watermark image will same. LL sub band are divided into
LL2, LH2, HL2 and HH2 bands. In which LH2, HL2, and HH2 contains the highest frequency band
while LL2 contains the low frequency band.

Fig 1: 1- Level DWT Decomposition

Fig 2: 2-Level DWT Decomposition

9
 B. Singular Vector Decomposition(SVD):
SVD is an linear algebra technique and used to solve mathematical problems. SVD approach used in
watermarking because of the fact that singular values obtained after the decomposition of image
matrix are stable. The SVD belongs to orthogonal transform which decompose the matrix into three
matrices of same size. we apply SVD to last LL band of input image that decompose the input image
into I_u, I_s and I_v matrix form. Like this apply svd to the lowest LL band of watermark image that
decompose the image into W_u, W_s and W_v. Through all this our main focus are on scaling matrix
that are used to control the strength of watermarked image.
4.1 .1
Proposed algorithms:

A. Embedding Process:
Read both input image and watermark image.
Use 1-level DWT to decompose the input image into four sub-bands (i.e. LL, LH, HL, HH).
Use 2-level DWT haar to again split LL into four sub-bands (i.e. LL2, LH2, HL2, HH2).
Apply SVD to LL2 sub band that split the image into matrix form that extract the singular
values.
Apply DWT to the watermark image to decompose the watermark image into four sub bands
Apply SVD to LL sub band to extract singular values.
Apply optimal scale matrix, that gives best values and stores in alpha variable.
Apply inverse DWT to get the watermarked image.
Calculate the PSNR value for input host image and watermarked image using formula.
B. Extraction Process:
Apply 2 level haar DWT to decompose the watermarked image into four sub bands.
Apply SVD to LL2 sub band to extract the singular values.
Apply 2 level haar DWT on input image to decompose image into four sub bands(i.e, LL2,
LH2, HL2, HH2)
Apply SVD to LL2 sub band .
Compute the singular matrix of extracted image.
watermark image to decompose the watermark image into sub bands(i.e,
WLL,WLH,WHL,WHH)
Apply SVD to WLL sub band to extract the singular values.
10
 Apply inverse DWT to get the extracted watermark image.
Check out the correlation coefficient of watermarked image and extracted watermark image
that will be noiseless.
C. Optimal Scaling :

Optimal Scaling helps for providing the best values. In Optimal Scaling, we used two main
factors:

Differential Evolution(DE):
DE algorithm worked on a population of candidate solutions(called agents). These candidate
solutions are moved around in the search space and creating new candidate solution by
combining the existing ones according to its simple formula and if the new position of an agent is
an improvement than it is accepted otherwise the new position is discarded. Pseudo code of DE
are Formally, let be the cost function which must be minimized or fitness function which must be
maximized. Let designate a candidate solution (agent) in population. The basic DE algorithm can
then be described as follows:
Initialize all agents with random positions in search-space.
Until a termination condition is met (e.g. number of iterations performed, or an adequate
fitness reached), repeat the following:
For each agent in the population do:
Pick three agents , from the population at random, they must be distinct from each
other as well as from agent .
Pick an random index ( being the dimensionality of the problem to be optimized).
Compute the agent's new position as follows:
For each , pick an number that are uniformly distributed
If or then set otherwise set
(In essence, the new position is outcome of binary crossover of agent with
intermediate agent .
If then replace that agent in the population with the improved candidate
solution, that is, replace with in the population.
Pick the agent from the population that has the highest fitness value or lowest the
cost value and return it as the best candidate solution.

11
4.1.3
Quality Measures:
Some of quality measures used in image watermarking are:
1) Normalized Correlation(NC):
It tells about the comparison between original(input) image and extracted image.NC targeted to 1.0.
=1
NC=
=1 2 =1 2

Wi= Original Watermark,

Wj=Extracted Watermark

2) Mean Square Error(MSE):

MSE in watermarking is to measures the average of square of errors between original image and
watermark image.

MSE= 1MN ()
Where M, N is pixel values in host image
Wij = Pixel value of Watermarked Image
Hij = Pixel value of Original Image

3) Peak Signal to Noise Ratio(PSNR):PSNR determines the Efficiency of Watermarking with respect
to the noise.
It is given by:
PSNR=10*log(/MSE)
Where p= maximum value in original image

12
 SOFTWARE TOOL

MATLAB
The name MATLAB stands for MATrix LABoratory. MATLAB was written originally to provide
easy access to matrix software developed by the LINPACK (linear system package) and EISPACK
(Eigen system package) projects.

The MATLAB language provides native support for the vector and matrix operations that are
fundamental to solving engineering and scientific problems, enabling fast development and
execution. MATLAB provides features of traditional programming languages, including flow
control, error handling, and object-oriented programming (OOP). You can use fundamental data
types or advanced data structures, or you can define custom data types.

MATLAB add-on products provide built-in algorithms for signal processing and
communications, image and video processing, control systems, and many other domains. By
combining these algorithms with your own, you can build complex programs and applications

KEY FEATURES
High-level language for numerical computation, visualization, and application development.

Interactive evironment for iterative exploration, design, and problem solving.

Mathematical functions for linear algebra, statistics, Fourier analysis, filtering, optimization,
numerical integration, and solving ordinary differential equations.

Built-in graphics for visualizing data and tools for creating custom plots.

Development tools for improving code quality and maintainability and maximizing Performance.
Tools for building applications with custom graphical interfaces.

Functions for integrating MATLAB based algorithms with external application

and languages such as C, Java, .NET, and Microsoft Excel.

13
 REFERENCES

[1] Divjot Kaur Thind, Sonika Jindal, A Semi Blind DWT-SVD Video Watermarking,
International Conference on Information and Communication Technologies(ICICT), pp:1661-
1667, 2015.
[2] Madhuri Rajawat, D S Tomar,A Secure Watermarking and Tampering detection technique
on RGB Image using 2 Level DWT, proc. IEEE 5th International Conference on
Communication Systems and Network Technologies, pp:638-642, 2015.
[3] Aparna J R, Sonal Ayyappan,Image Watermarking using Diffie Hellman Key Exchange
Algorithm, proceeding of the International Conference on Information and Communication
Technologies(ICICT), pp:1684-1691, 2015.
[4] Parashar Preeti., Rajeev Kumar Singh , A Survey: Digital Image Watermarking Techniques
, International Journal of Signal Processing, (ijsip), pp:111-127,2014.
[5] Lalit Kumar, Vikas Shrivastava, A Survey of Digital Watermarking Techniques and its
Applications , International Journal of Computer Science Trends and Technology (IJCST),
pp:70-73,2014.
[6] Ms. Mahejabi Khan, Mr. Ajay Kushwaha, Mr. Toran Verma,A new digital image
watermarking algorithm based on image interlacing,DWT,DCT, IEEE International
Conference on Industrial Instrumentation and Control(ICIC) college of Engineering
Pune,India. May 28-30,2015, pp:885-890.
[7] Urvi H.Panchal, Rohit Srivastava,A Comprehensive Survey on Digital Image Watermarking
Techniques, IEEE 5th International Conference on Communication Systems and Network
Technologies, pp:591-595, 2015.
[8] D. Vaishnavi, T.S.Subashini,Robust and Invisible Image Watermarking in RGB Color
Space using SVD, procedia computer science 46, International Conference Information and
Communication Technology(ICICT), pp:1770-1777, 2015.
[9] GhoutiL, BouridaneA and Ibrahim MK, Digital image watermarking using balanced
multiwavelets, IEEE Transactions on Signal Processing, 54(4), pp. 1519-1536, 2006.
[10] Podili Brahma Srinu, B.Jagadeesh, Copyright Protection of Digital Image using FIS-
BPNN Technique in Hybrid Transform Domain , International Journal of Innovative
Research in Science, Engineering and Technology (An ISO 3297: 2007 Certified
Organization).

14
[11] Hiren Joshi,Ashish Jani , A Novel Digital Watermarking Algorithm using Random
Matrix Image, International Journal of Computer Applications (0975 8887) Volume 61
No.2, January 2013
[12] Anumol Joseph ,, K. Anusudha , Robust watermarking based on DWT SVD,
International Journal of Signal & Image Processing, Issue. 1, Vol. 1, October 2013
[13] Tulika Bhuyan , vinay kumar srivastava and Falgun Thakkar , Shuffled SVD based
Robust and Secure Digital Image Watermarking, International Conference on Electrical,
Electronics, and Optimization Techniques (ICEEOT) - 2016

15