Implementation of Watermarking in Image Processing using Discrete Wavelet Transform

breezebongΤεχνίτη Νοημοσύνη και Ρομποτική

6 Νοε 2013 (πριν από 3 χρόνια και 5 μήνες)

181 εμφανίσεις

Implementation of Watermarking in Image
Processing using
Discrete Wavelet Transform



Bushra Jamal


Department
of CS/IT

Galgotia Colleg
e of Engineering and Technology


Greater Noida, U. P., INDIA

E
-
mail address:
bushra.jamalrgecmeerut@gmail.com


Alok Kuma
r


Department
of CS/IT

Galgotia Colleg
e of Engineering and Technology


Greater Noida, U. P., INDIA

E
-
mail
address:
godinall.kumar1@gmail.com

Abstract
-

The advancement of computer technologies
and rapid expansion of internet in past years has rapidly
incr
eased the availability of multimedia data throughout
the globe. The digital media brings about convenience to
people but sometimes it can be easily used to be copied,
modified and distributed in an illegal way. The copy right
protection, intellectual prote
ction and material right
protection for authors, owners, buyers and distributors is
necessary and the authenticity of content or matter are
crucial factors to solving problem. In such case the digital
watermarking techniques play an important role as a
val
id solution. The main focu
s of w
atermarking is
developing and introducing new techniques for
watermark embedding and detection. Digital
watermarking techniques have been evaluated according
to robustness, capacity and imperceptibility measures. The
idea of

robust watermarking of images is to embed
information data within the image with an insensible
form for human visual system but in a way that protects
from attacks such as common image processing
operations. In this paper we have given the overview of
wat
ermarking and a suitable method of watermarking for
copyright protection of digital images based on DWT
(Discrete wavelet Transform). Experimental results show
that the embedde
d watermark is quite robust in c
as
e of
various watermark images at high compress
ion ratios and
provides good results in terms of imperceptibility.


Keywords:

Digital watermarking, Wavelet Transform,
Copyright protection


I.

INTRODUCTION


In the late 1990s there was an explosion of interest in
digital systems
for the watermarking of vari
ous
digital
content
s
. The main focus has been on
photographs, audio, and video, bu
t other content
such
as binary images, text, line drawings, three
-
dimensional models, animation parameters,
executable

code, and integrated circuits
have also
been marked. Th
e proposed applications of these
methods are many and varied, and include
identification of the copyright owner, indication to
recording equipment that the marked content should
not be recorded, verification that content has not been
modified since the mar
k was embedded, and the
monitoring of broadcast channels looking for marked
content. A watermark is
basically
a “secret message”
that is embedded into a “cover (original or host)
message”. Digital watermarking is a technique for
inserting information (the
watermark) into an image,
which can be later extracted or detected for variety of
purposes including identification and authentication
purposes.
This

watermark travels with the
cover
message
, even after copying and redistribution. Only
the knowledge of a s
ecret key allows us to extract the
watermark from the cover message. Three very bas
ic
requirements of watermarking
are imperceptibility,
robustness and data hiding.

In this paper, a wavelet
-
based watermarking approach is proposed in which a
visually recogn
izable watermark is added to the
wavelet coefficients of an image. This watermark is a
gray
-
scale image. The extracted watermark is
visually recognizable to claim ownership.
In the
proposed method the

watermark is embedded

into the
lowest sub bands as the
m
agnitude of DWT
coefficients is larger in the lowest bands (LL) at each
level of decomposition and is smaller for other bands
(HH, LH, HL).
The larger the magnitude of the
wavelet coefficient the more significant it is
.


II.

WATERMARKING B
ACKGROUND


A.

General mo
del of digital watermarking


Digital watermarking is considered as an efficient
tool to prove the ownership of digital data. With
digital multimedia distribution over World Wide
Web, Intellectual Property Right (IPR) is more
threatened than ever due to the

possibility of
unlimited copying. One solution would be to restrict
access to the data using some encryption technique.
However encryption does not provide overall
protection. Once the encrypted data are decrypted,
they can be freely distributed or manipu
lated. The
above problem can be solved by hiding some
ownership data into the multimedia data, which can
be extracted later to prove the ownership. This idea is
implemented in bank currency notes. In bank
currency notes, a watermark is embedded which is
us
ed to check the originality of the note. The same
“watermarking” concept may be used in multimedia
digital contents for checking the authenticit
y of the
original content. So, w
atermarking is adding
“ownership” information in multimedia contents to
prove th
e authenticity. This technology embeds a
data, an unperceivable digital code, namely the
watermark, carrying information about the copyright
status of the work to be protected. Thus
watermarking techniques may be relevant in various
application areas inclu
ding Copyright protection,
Copy protection, Temper detection, Fingerprinting,
medical images(MRI, CT scan, Ultrasound ) etc.

The
major requirements of watermarking system are
perceptual transparency, payload of the watermark,
robustness, security and effic
iency. Robustness of the
watermarking system is its ability to resist various
signal processing attacks. Some of the common
image

processing attacks are JPEG compression,
filtering, cropping, geometric distortions and additive
noise.

Any watermarking syste
m consists of
following two steps; (1) Watermark embedding

and
(2) Watermark

extraction
.



Embedding procedure
: Watermark
embedding is the process in which the secret
image is hidden inside the cover image
without modifying the visibility of the cover
image.

The inputs to the embedding process
are the watermark, the cover object and a
secret key. The key is used to enforce
security and to protect the watermark. The
output of the watermarking scheme is the
watermarked data.



Extraction
procedure
: W
atermark extr
action

is the process of retrieving hidden image
from the watermarked data via certain
mechanism at

receiver
end.


B.

Digital
Watermark Categories and Types of
Algorithms in Use


Digital watermarks are broadly classified into three
categories as
follows [
1]:
Robust, Fragile and Semi
Fragile. In
Robust watermark
the watermark should
be permanently intact to the host signal, and
removing the watermark result in destroying the
perceptual quality of the signal. It is u
sed for
copyright protection,

withstand modera
te to severe
signal
processing

attacks
,

compression, rescaling
,
filtering
)
. Fragile watermark
break very easily under
any modification of the host signal. It is basically
u
sed for tamper detection or as a digital signature.

Semi Fragile watermark
is robust

to some benign
modifications, but brake very easily to other attacks.
It usually
provides

information about the location and
nature of attack and is

designed to break under all
changes that exceed a user
-

specified
threshold

and is
generally
used for data

authentication.

Watermarking techniques can be classified into four
categories according to the type of docume
nt to be
watermarked
as [
3]: Text Watermarking, Image
Watermarking, Audio Watermarking and Video
Watermarking.

In the case of images
, watermarks

can be applied in
spatial

domain,

F
requency domain

and Transform
domain
. In Spatial domain, pixels of one and two
randomly selected subsets of an image are modified
based on perceptual analysis of the original image.
It
is usually based on Spread Spectrum

approach.
In
Frequency domain, values of certain frequencies
are
altered from their original and in

Transform domain
watermark is embedded by modifying the
coefficients of global or block transform using any of
the technique of transform domain e.g., DCT,

DFT,
wavelet

According to human perception the digital
watermarks are classified as visible watermark and
Invisible watermark. Visible watermarking systems
are those in which watermark embedded is visible to
the human visual system (HVS) when image is
vi
ewed [
9
]. Visible watermarking is normally used to
prevent unauthorized access to the data. In invisible
watermarking, watermark embedded is perceptually
invisible to the HVS [6]. The major requirements of
watermarking system are perceptual transparency,
p
ayload of the watermark, robustness, security and
efficiency [1
0
]. Robustness of the watermarking
system is its ability to resist various signal processing
attacks. Some of the common signals processing
attacks are JPEG compression, filtering, cropping,
ge
ometric distortions and additive noise. The digital
watermarking has been used in a wide variety of
applications such as copyright protection, owner
identification, transaction tracking, medical
applicatio
ns and covert communications [11
].

Depending on the

method used for watermark
extraction digital watermarking is divided into blind
,
Semi
-
blind

and non
-
blind wate
rmarking. Blind
watermarking [12
] method does not require original
image to extract the watermark from the watermarked
image,

Semi
-
blind also doe
s not use the original
signal but use some side information and/or the
original watermark
whereas the non
-
blind
watermarking [2] method extracts the watermark by
comparing the watermarked image with the original
image.


C.

Application
and Challenges in Waterm
arking


There is wide variety of applications for digital
watermarking but no “universal” watermarking exists
as requirement of watermarking systems are always
based on application. Although it has to be robust in
general, different level of required robus
tness can be
identified besides the specific characteristics of the
problem which make the watermarking a suitable
solution. Several application are listed below based
on Cox et al [4] and Katzenbeisser and Petitcolas [5]



Owner identification


similar to
copyright
protection, to establish ownership of the content



Copy protection


also known as copy control,
to prevent people from making illegal copies of
copyrighted content.



Content authentication


to detect modifications
of the content, as a sign of inv
alid
authentication.



Fingerprinting


sometimes referred as
transaction tracking or traitor tracking, to trace
back illegal duplication and distribution of the
content.



Broadcast monitoring


specifically for
advertisements and in entertainment industries,

to monitor content being broadcasted as
contracted and by the authorized source.



Medical applications


known as invertible
watermarking, to provide both authentication
and confidentiality in a reversible manner
without affecting the medical image in any
way.

Apart of all the implementations and applications of
watermarking there still exists some limitations such
as
-



Lack of protocols, standards and benchmarking.



Lack of comprehensive mathematical theory.



Watermark survival for all attacks.



Relating robus
tness, capacity, perceptual
quality and security.


D.

Watermark
A
ttacks


A watermark attack is an attack on embedding
methods where the presence of a watermark can be
detected by an attacker without knowing the
encryption key. There are various kind of attack

such
as Passive Attack, Collusion Attack, Court of law
attack, Robustness attack, Presentation Attack, and
Counterfeiting attack.

In Passive Attack hacker tries to find out if a
watermark is present in image. In this removal of
watermark is not an aim. It

is Serious for covert
communications. In Collusion Attack hacker uses
several copies of watermarked data (images, video
etc.) to construct a copy with no watermark. He uses
several copies to find the watermark. It is Serious for
fingerprinting application
s.

In Court of law attack
hacker take advantage of legal issues. Robustness
attack is basically intended to remove the watermark
by various techniques such as JPEG compression,
filtering, cropping, histogram equalization additive
noise etc. Presentation At
tack includes
w
atermark
detection failure,
g
eometric transformation, rotation,
scaling, translation, change aspect ratio, line/frame
dropping, affine transformation etc. In Counterfeiting
attack hacker try to render the original image useless,
generate fak
e original image, and create dead lock
problems.



III.

PROPOSED WATERMARKING SCHEME


A.

Wavelet Transform


In order to have more promising techniques,
researches were directed towards watermarking in the
transform domain, where the watermark is not added
to the i
mage intensities, but to the values of its
transform coefficients. Wavelet domain techniques
facilitate to generate and apply selective key
mechanism applicable only to selective portion of the
image.

Discrete Wavelet Transform has been used in digital
ima
ge watermarking more frequently due to its
excellent spatial localization and multi
-
resolution
characteristics, which are similar to the theoretical
model of human visual system. The wavelet
transform decomposes the image into three spatial
directions, i.e
. horizontal, vertical and diagonal.
Hence wavelets reflect the properties of HVS more
precisely.
Wavelet T
ransform decomposes an image
into a set of band limited components which can be
reassembled to reconstruct the original image without
error.















As pointed out

in [7], wavelet
-
based watermarking
methods exploit the frequency information and
spatial information of the transformed data in
multiple resolutions to gain robustness.

Wavelet
Transform is computationally efficient and can be
implem
ented by using simple filter.

Magnitude of
DWT coefficients is larger in the lowest bands (LL)
at each level of decomposition and is smaller for
LL
3

LH
3

LH
2

LH
1

HL
3

HH
3

HL
2

HH
2

HL
1

HH
1


Figure
2
:
Three

Level

2
-
dimensional

DWT model

other bands (HH, LH, HL).

The larger the magnitude
of the wavelet coefficient the more significant it is.

Water
mark detection at lower resolutions is
computationally effective because at every successive
resolution level there are few frequency bands
involved.

High resolution
sub bands

helps to easily
locate edge and textures patterns in an image.

One of
the most s
traightforward techniques is to use a
CDMA sequence in the detail bands according to the
[
8
] given below.


I
Wu,v

= { W
i

+
α

| W
i

| x
i
,

u,v
Є

HL,LH



W
i
u,v
Є

LL,HH

Where W
i

denotes the coefficient of the transformed
image, x
i

the bit of the watermark to be embedded,
and α a scaling factor. To detect the waterm
ark we
generate the same pseudo
-
random sequence used in
CDMA generation and determine its correlation with
the two transformed detail bands. If the correlation
exceeds some threshold δ, the watermark is detected.

There are various advantages of using
discr
ete

w
avelet transform
such as



It

understands the HVS more closely than the
DCT.




Wavelet coded image is a multi
-
resolution
description of image. Hence an image can be
shown at different levels of resolution and can be
sequentially processed from low resolu
tion to
high resolution.




DFT and DCT are full frame transform, and
hence any change in the transform coefficients
affects the entire image except if DCT is
implemented using a block based approach.
However DWT has spatial frequency locality,
which means i
f signal is embedded it will affect
the image locally. Hence a wavelet transform
provides both frequency and spatial description
for an image.

Wavelet transforms use wavelet filters to transform
the image. There are many available filters, although
the mos
t commonly used filters for watermarking
are
:
-
Haar Wavelet Filter, Daubechies Orthogonal
Filters
,
D
aubechies Bi
-
Orthogonal Filters,
Symlet
wavelets
.
Each of the filters decomposes the image
into several frequencies.

Single level decomposition
gives four f
requency representations of the images.


B.

Proposed
Watermark
Embedding
S
cheme


Watermark embedding steps of proposed techniques
are as follows:

Step 1

Set the gain factor for embedding.

Step 2


Read in the cover image

Step 3

Determine size of watermarked (c
over)
image.

Step 4


Read in the message image and reshape it
into a vector.

Step
5

Decompose the cover image using simple
‘Haar’ Wavelet into four non overlapping multi

resolution coefficient sets: LL1, HL1, LH1 and HH1.

Step
6

Perform second level DWT on

LL1 to give 4
coefficients: LL2, HL2, LH2 and HH2.

Step
7

Repeat decomposition for LL2 to give next
level compo
nents: LL3, HL3, LH3 and HH3
.

Step
8

set the value of
block size

Step 9

Re
-
formulate the grey
-
scale watermark
image into a vector of zeros and o
nes.

Step
10

Process

the image in blocks

Step
11

Add PN
-
seque
nces to h3 and v3

components
when message = 0, according to the formula shown
below:


I
Wu,v

= { W
i

+
α

| W
i

| x
i
, u,v
Є

HL,LH





W
i
u,v
Є

LL,HH

Step
12


Perform IDWT.

Step
13

Convert back to uint8.

Step
14

Write watermarked image to file.

Step
15

Display watermarked image.

The general framework
for this proposed technique is
based on the concept of embedding watermark image
into the host image by applying DWT technique and
obtain watermarked image by IDWT. It is
diagrammatically represented as shown below:














C.

Proposed Watermark Extract
ion Scheme


Watermark extraction steps of proposed techniques
are as follows:

Step 1

Read in the watermarked object.

Step 2

Determine size of watermarked image.

Step 3

Decompose watermarked image

using
Haar
wavelet up to 3 levels to get HL3 Coefficients.

S
tep 4

Read in original watermark.

Step 6

Determine size of original watermark

Step 7

Initialize message to all ones.

Step 8

Process the image in blocks

Step
9

Add PN
-
sequences to h3 and v3

components
when message = 0.

Step 10

Reshape the message vector and

display
recovered watermark.

Step 11

Display recovered message.


Figure

3:

General Framework of
DWT based
Watermark embedding

scheme



Step 12

Calculate the quality of recovered image by
using
PSNR (Peak Signal
-
to
-
Noise Ratio)

function

PSNR = 10log
10

(255
2
/MSE) dB

Step 13


Calculate the Accuracy rate of recovered
image by u
sing AR function

AR=CP/NP

Where NP is the number of pixels of the watermark
image and CP is the number of correct pixels in the
watermark image that is retrieved from the attacked
image.


IV.

EXPERIMENTAL RESULTS


In t
he test set

for this watermark evaluatio
n
experiment,

image
is
randomly selected from the
internet. Matlab 7.7.0 software platform is use to
perform the experiment. The PC for experiment is
equipped with an Intel Pentium(R) dual
-

core CPU
Personal laptop and 2GB RAM.

The experimental results are

represented in the
following figures for watermarked image in DWT
domain while taking the different values of gain
factor K. The following results are for first level
decomposition and analysis for the third level
decomposition is in progress and results
are yet to be
achieved. Various observations for experiment are
depicted in table I. Due to its computationally
efficient modeling of the HVS, the wavelet domain
offers perhaps the most promising environment for
robust watermarking. The algorithm does not
offer
any problem for retrieving the small watermark from
the watermarked image along with only minimal
degradation of the cover image during embedding.
Watermarking in the wavelet domain appears to show
the most promising results. The algorithm described
here is one of the most simplistic available in the
wavelet domain, and yet the results are still excellent.
These results tend to reinforce the common belief in
wavelet domain as the most promising domain for
digital watermarking. A final consideration is

the size
of the watermark being embedded. Use of a smaller
watermark allows larger blocks to be used, increasing
the strength of correlation and thus system
robustness. It’s worth noting that watermark recovery
could be improved by using a smaller waterma
rk and
increasing the block size used for embedding. This
should reduce the number of errors from normal
detection, as well as improving watermark
robustness.





Table
I
:
RESULTS UNDER SEVERAL EXECUTIONS

Gain
F
actor

Execution
T
ime

PSNR

AR

K=1

96.4866

51.
1115

81.7391

K=2

95.4414

53.2435

86.2650

K=3

94.3124

55.1205

90.1845

K=4

93.2122

56.1047

91.0649

K=5

92.0625

59.7828

92.5308

K=6

92.0012

60.2461

93.5701















































Watermarked Image
Watermarked Image
Watermarked Image
Watermarked Image
Watermarked Image
Watermarked Image

Figure
4
:
watermarked Lena image

f
or k= 1
, 2, 3….

6

V.

CONCLUSION


In our proposed watermarking method, a

grayscale
visual watermark image is inserted into the host
grayscale image using the Discrete Wavelet
Transform, where the copyright of Watermark is
printed. The experimental results have confirmed that
this technique is robust is nature. We have describe
d
the comprehensive ov
erview of watermarking and
give detailed
view of proposed technique. The
protection of intellectual property rights is perhaps
one of the last major barriers to the “digital world.”
While the techniques presented in this paper are not

foolproof, they can help validate claims of ownership
that are required for intellectual property law
enforcement. On
-
going research is currently
conducted where the project contributions and scope
are as follows:

1. More improved results for one level de
composition
as well as to achieve highly validated results for two
and three level decompositions
.

2. Security aspects against different types of attacks
will be examined both qualitatively and
quantitatively for a comprehensive set of different
image type
s.


REFERENCES


[1]. Rey, C.; Dugelay, J. L. (2002): A Survey of Watermarking
Algorithms for Image Authentication, EURASIP Journal on
Applied Signal Processing, pp. 613

621.

[2]. Burt, P. J.; Adelson, E.H. (1983): The Laplacian pyramid as a
compact image c
ode, IEEE Transactions on Comm.,31(4), pp. 532
-
540.

[3]. Chandra, M.; Pandey, S. (2010): A DWT Domain Visible
Watermarking Techniques for Digital Images, International
Conference on Electron
ics and Information Engineering.

[4] Cox, I. J.; Miller, M. L.; Bl
oom, J. A. (2002): Digital
Watermarking, Morgan Kaufmann Publishers.

[5] Katzenbeisser, S.; Petitcolas, F. A. P. (2000): Information
Hiding Techniques for Steganography and Digital Watermarking,
Artech House.

[6]. Craver, S., et al. (1998): Resolving Right
ful Ownerships with
Invisible Watermarking Techniques: Limitation, Attacks and
Implications, IEEE Journal on Selected Areas in Communication,
16(4), pp. 573
-
586.

[7] Woo, C. S.; Du, J.; Pham, B. (2005): Performance Factors
Analysis of a Wavelet
-

based Wate
rmarking Method, Proc. 3rd
Australasian Information Security Workshop (AISW2005),
CRPIT, vol. 44, pp.89
-
97.

[8] Langelaar
, G.; Setyawan, I.; Lagendijk, R. L.
(2000):Watermarking Digital Image and Video Data, IEEE Signal
Processing Magazine, Vol 17,pp 20
-
43

[
9
].Huang, C. H.; Wu, J. L. (2004): Attacking Visible
Watermarking Schemes, IEEE Transactions on Multimedia, 6(1),
pp.1
6
-
30.

[1
0
].Miller, M. L.; Cox, I. J.; Linnartz, J. P. (1999): A review of
watermarking principles and practices, Published in Digital Signal
Processing in Multimedia Systems.

[11
].Morimoto, N.; Digital Watermarking Technology with
Practical Applications, I
nforming science special issue on
multimedia informing technologies, 2(4), pp. 107
-
111.

[12
].Wong, P. H. W.; Au, O. C.; Yeung, Y. M. (2003): A Novel
Blind Multiple Watermarking Technique for Images, IEEE
Transactions on Circuits and Systems for Video Techn
ology,
13(8), pp. 813
-
830.