Abstract: In this
paper a vigorous watermarking system is proposed by using SVD(Singular  Value Decomposition)and DWT(Discrete Wavelet
Transform). SVD is being used for numerous applications along with supplementary
methods In addition to usual hiding arrangements a random moniker is utilized
to increase its heftinessin contradiction tobogusburglars. The unitary matrices
are utilized to produce a moniker which is going to be embedded into the fourth
level decomposition of shield image. After extraction, image be checked with
the Moniker embedded. If these Monikers are harmonized, unitary matrices are
used to excerpt watermark from watermarked images. Diverse attacks are well-thought-out
and the simulation outcomes show that the extraction of watermark after attacks
had minor effects.

 

Keywords: attacks,
authentication, moniker, SVD, watermarking

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

 

I.                  
Introduction

The
security of data transmitted from one place to other using regular wired or
wireless network is a major concern in the current era. Large amount of secrete
information is being sent using a regular network it may be wired or wireless.
Most of the networks will not employ additional security measures on how the
data is being transferred from one place to other. Hence the users of these
kinds of networks are themselves responsible to take care of their data. An
attempt to transmit images safely was made in this paper. The components of a
digital watermarking system are embedding and extraction. Depending on what
domain of shield image was used to embed the watermark, spatial and transform
or frequencydomain watermarking systems are proposed in the literature.

The chiefadvantage
of transform domain tactics is their extreme robustness to common image falsifications.
Discrete cosine transformis a fundamental technique which was used extensively
in early days of image processing and specifically used in image compression.
Later, the revolution of wavelet has shifted the view of researcher as well as
the people from industry with a wide range of applications. While the DCT
provides an efficient representation for the input image, the DWT in addition
to efficient representation it also enable different mechanisms by which
different image processing tasks can ease their implementations and improves
the performance of their task. The interesting feature of wavelet transform is
that it separates the image in terms of frequency with multiple levels. Some of
the levels are playing a vital role in reconstruction and other a minor role. Now
the secret information can be saved in bands which has not much work in
reconstruction.

This
is somehow or other similar to the LSB based technique of spatial domain 1.
The LSB based technique concentrates in hiding the message bits in the LSB
locations of the cover image. The original LSB bits will be lost and the
importance of these bits is less in the sense that these bits if changed from
zero to one or one to zero results in a change of pixel value by only one,
hence has less effect on the display. In addition to the said watermarking
scheme numerous watermarking techniques robust to symmetrical attacks have been
proposed in the literature 23. The wavelet based watermarking schemes are
found to be robust against multiple attacks like compression, blurring, salt
and pepper noise and many other 4.

A
watermarking scheme will have the following basic components. Watermark is the
prime component of the scheme, which is to be conveyed securely to the
destination which sustains many attacks. Cover image, which is usually large
enough for the watermarking scheme to embed the watermark on to it. Watermarked
image, which is the outcome of the watermarking scheme. Embedding means a process
of hiding the watermark in the cover image. Attack is an activity of modifying
the effective appearance or the effective pixel value plane to a different set.
The extraction is the process of separating the watermark from the watermarked
image 5.

In the
literature a number modifications and improvements has been made to the
watermarking schemes. In 6, Mohammad Ali et.
al presented a blind digital watermarking scheme based on quantization of
Eigen values in Wavelet domain. In the literature fuzzy and artificial neural
networks based techniques 7-9, SVD based techniques 1011, hybrid
techniques 1213, Biometric template based techniques 14, Evolutionary
algorithm based techniques 15, Quadtrees based techniques 16, GEP based
techniques 17 and video watermarking schemes 1819 are proposed. A large
number of surveys are also being conducted 2021.

In
this paper a watermarking scheme is proposed which is a modified version or
improved version of traditional SVD-DWT based watermarking scheme. The rest of
the paper is organized as follows. In section-IIthe basic or standard DWT-SVD
based watermarking scheme was presented. In section-III the authentication
issue of the standard DWT-SVD based technique is described. Section-IV is
concerned the solution of the authentication problem. Section-V presents the simulation
results and the last section concludes the paper.

 

II.               
The Standard DWT-SVD Watermarking
Algorithm

The
standard DWT-SVD watermarking scheme considers the cascade of DWT and SVD as
the main building block for the watermarking scheme. First the DWT will be
applied to the cover image or the carrier image. The DWT as usual decomposes
the cover image into four frequency bands: Low-Low, Low-High, High-Low and
High-High. The Low-Low band characterizeslow frequency, High-Low and Low-Highbands
describe middle frequency and High-High bandcharacterizes high frequency bands,respectively.
The Low-Low band signifies approximate details, High-Low band horizontal
details, Low-High vertical details and High-High band diagonal details of the
image.

These
different bands contain different grades of information in it. The cover image
is basically an image hence the Low-Low band contains all the small variations
which crucial in determining the boundary of different objects present in the
image. Similarly the Low-High and High-Low has some useful information in
reconstruction than that of High-High band of frequency. Hence the High-High
frequency band was selected as the candidate to store the information related
to the secrete data. To provide additional security the SVD is applied to the
High-High band of cover image as well as to the watermark. The concept of this
scheme is to replace the singular values obtained after applying the SVD of
cover image with the singular values of watermark.  It is observed that singular values range
from85to 175 for most of the standard images.