A New Digital Watermarking Algorithm Using Combination of Least Significant Bit (LSB) and Inverse Bit

Article · November 2011with136 Reads
Source: arXiv
Abstract

In this paper, we introduce a new digital watermarking algorithm using least significant bit (LSB). LSB is used because of its little effect on the image. This new algorithm is using LSB by inversing the binary values of the watermark text and shifting the watermark according to the odd or even number of pixel coordinates of image before embedding the watermark. The proposed algorithm is flexible depending on the length of the watermark text. If the length of the watermark text is more than ((MxN)/8)-2 the proposed algorithm will also embed the extra of the watermark text in the second LSB. We compare our proposed algorithm with the 1-LSB algorithm and Lee's algorithm using Peak signal-to-noise ratio (PSNR). This new algorithm improved its quality of the watermarked image. We also attack the watermarked image by using cropping and adding noise and we got good results as well.

Full-text

Available from: Rosziati Ibrahim, May 06, 2014
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 1
A New Digital Watermarking Algorithm Using
Combination of Least Significant Bit (LSB)
and Inverse Bit
Abdullah Bamatraf, Rosziati Ibrahim and Mohd. Najib Mohd. Salleh
AbstractIn this paper, we introduce a new digital watermarking algorithm using least significant bit (LSB). LSB is used because of its little
effect on the image. This new algorithm is using LSB by inversing the binary values of the watermark text and shifting the watermark
according to the odd or even number of pixel coordinates of image before embedding the watermark. The proposed algorithm is flexible
depending on the length of the watermark text. If the length of the watermark text is more than ((MxN)/8)-2 the proposed algorithm will also
embed the extra of the watermark text in the second LSB. We compare our proposed algorithm with the 1-LSB algorithm and Lee’s algorithm
using Peak signal-to-noise ratio (PSNR). This new algorithm improved its quality of the watermarked image. We also attack the watermarked
image by using cropping and adding noise and we got good results as well.
Index Terms— Digital watermarking, Grayscale images, Least significant bit (LSB), PSNR, Watermark text.
——————————
——————————
1 I
NTRODUCTION
rivacy is the ability of an individual or group to
insulate them or information about themselves and
thereby reveal them selectively [1]. Data privacy is the
relationship between collection and dissemination of
data, technology, the public anticipation of privacy, and
the legal issues [2]. Data privacy or data protection has
become increasingly important as more and more systems
are connected to the internet [2]. Watermarking is a
pattern of bits inserted into a digital image, audio or
video file that specifies the file's copyright information
such author, rights and so on [3]. Thus, watermarking
approach is used to make sure of the protection of the
data. However, watermarking is also designed to be
completely invisible. The actual bits representing the
watermark must be scattered throughout the file in such a
way that they cannot be identified and tampered [4].
Thus, the watermarking must be robust enough so that it
can withstand normal changes to the file such as
attacking by adding noise [5].
Contrast to printed watermarks, digital watermarking
is a technique where bits of information are embedded in
such a way that is completely invisible [6]. The problem
with the traditional way of printing logos or names is that
they may be easily tampered or duplicated. In digital
watermarking, the actual bits are scattered in the image in
such a way that they cannot be identified and show
resilience against attempts to remove the hidden data [7].
Media watermarking research is a very active area and
digital image watermarking became an interesting
protection measure and got the attention of many
researchers since the early 1990s [8].
The rest of this paper is organized as follows: Section 2
describes the related work and LSB. Section 3 discusses
the proposed algorithm. Results and discussion is given
in Section 4. The PSNR and its results are shown in
section 5. Discussing attacks on the watermarked images
in section 6 and finally, conclusion will be presented in
Section 7.
2 R
ELATED WORK
In this section we will look into the review of digital
watermarks used for images. It describes the previous
work which had been done on digital watermarking by
using LSB technique and other techniques, including the
analysis of various watermarking schemes and their
results.
Gaurav Bhatnagar et al [9], presented a semi-blind
reference watermarking scheme based on discrete wavelet
transform (DWT) and singular value decomposition
(SVD) for copyright protection and authenticity. Their
watermark was a gray scale logo image. For watermark
embedding, their algorithm transformed the original
image into wavelet domain and a reference sub-image is
formed using directive contrast and wavelet coefficients.
Then, their algorithm embedded the watermark into
————————————————
A. Bamatraf is with the universiti Tun Hussein Onn Malaysia, 86400 Batu
Pahat, Johor, Malaysia. E-mail: abdom45@hotmail.com.
R. Ibrahim is with the universiti Tun Hussein Onn Malaysia, 86400 Batu
Pahat, Johor, Malaysia. E-mail: rosziati@uthm.edu.my.
M. N. M. Salleh is with the universiti Tun Hussein Onn Malaysia, 86400
Batu Pahat, Johor, Malaysia. E-mail: najib@uthm.edu.my.
P
© 20
11
Journal of Co
m
puting
2151
-
9617
http://sites.google.com/site/journalofcomputing/
Page 1
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 2
reference image by modifying the singular values of
reference image using the singular values of the
watermark.
Hao Luo et al [10], proposed a self-embedding
watermarking scheme for digital images. In their
proposed algorithm they used the cover image as a
watermark. It generates the watermark by halftoning the
host image into a halftone image. Then, the watermark is
permuted and embedded in the LSB of the host image.
The watermark is retrieved from the LSB of the suspicious
image and inverse permuted.
Wen-Chao Yang et al [11] used the PKI (Public-Key
Infrastructure), Public-Key Cryptography and watermark
techniques to design a novel testing and verifying method
of digital images. The main idea of their paper is to
embed encryption watermarks in the least significant bit
(LSB) of cover images.
Hong Jie He et al [12], proposed a wavelet-based
fragile watermarking scheme for secure image
authentication. In their proposed scheme, they generated
the embedded watermark using the discrete wavelet
transform (DWT), and then they elaborated security
watermark by scrambling encryption is embedded into
the least significant bit (LSB) of the host image.
Sung-Cheal Byun et al [13], proposed a fragile
watermarking scheme for authentication of images. They
used singular values of singular value decomposition
(SVD) of images to check the integrity of images. In order
to make authentication data, the singular values are
changed to the binary bits using modular arithmetic.
Then, they inserted the binary bits into the least
significant bits (LSBs) of the original image. The pixels to
be changed are randomly selected in the original image.
Gil-Je Lee et al [14] presented a new LSB digital
watermarking scheme by using random mapping
function. The idea of their proposed algorithm is
embedding watermark randomly in the coordinates of the
image by using random function to be more robust than
the traditional LSB technique.
Saeid Fazli et al [15] presented trade-off between
imperceptibility and robustness of LSB watermarking
using SSIM Quality Metrics. In their algorithm, they put
significant bit-planes of the watermark image instead of
lower bit-planes of the asset picture.
Debjyoti Basu et al [16] proposed Bit Plane Index
Modulation (BPIM) based fragile watermarking scheme
for authenticating RGB color image. By embedding R, G,
B component of watermarking image in the R, G, B
component of original image, embedding distortion is
minimized by adopting least significant bit (LSB)
alteration scheme. Their proposed method consists of
encoding and decoding methods that can provide public
detection capabilities in the absences of original host
image and watermark image.
To overcome the drawback of existing techniques, we
would like to introduce a new alternative technique by
inserting watermark text in grayscale images by using
watermarking approach.
3 R
EVIEW OF LSB
In a digital image, information can be inserted directly
into every bit of image information or the more busy
areas of an image can be calculated so as to hide such
messages in less perceptible parts of an image [14],[17].
Tirkel et al [18] were one of the first used techniques
for image watermarking. Two techniques were presented
to hide data in the spatial domain of images by them.
These methods were based on the pixel value’s Least
Significant Bit (LSB) modifications. The algorithm
proposed by Kurah and McHughes [19] to embed in the
LSB and it was known as image downgrading [20].
An example of the less predictable or less perceptible
is Least Significant Bit insertion. This section explains
how this works for an 8-bit grayscale image and the
possible effects of altering such an image. The principle
of embedding is fairly simple and effective. If we use a
grayscale bitmap image, which is 8- bit, we would need to
read in the file and then add data to the least significant
bits of each pixel, in every 8-bit pixel.
In a grayscale image each pixel is represented by 1
byte consist of 8 bits. It can represent 256 gray colors
between the black which is 0 to the white which is 255.
The principle of encoding uses the Least Significant Bit of
each of these bytes, the bit on the far right side.
If data is encoded to only the last two significant bits
(which are the first and second LSB) of each color
component it is most likely not going to be detectable; the
human retina becomes the limiting factor in viewing
pictures [17].
For the sake of this example only the least significant
bit of each pixel will be used for embedding information.
If the pixel value is 138 which is the value 10000110 in
binary and the watermark bit is 1, the value of the pixel
will be 10000111 in binary which is 139 in decimal. In this
example we change the underline pixel.
4 P
ROPOSED METHOD
Based on LSB technique, we propose a new watermarking
algorithm. Most of researchers have proposed the first
LSB but our proposed watermarking algorithm is
inversing the watermark text and embedding it in
different order in the traditional LSB. This is because of
the security reason. So, no one will expect that the hidden
watermark text in different order and it is inversed.
Figure 1 shows the framework of the proposed method.
First, we select the image which is a grayscale image and
we will transfer the watermark to binary value after
typing it. Then, we embed the watermark in the image
using the proposed algorithm. Figure 2 shows the
Page 2
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 3
embedding algorithm. Then, we will get the watermarked
image. Then, the receiver will retrieve the watermark
back. The watermark will be extracted from the
watermarked image. Figure 3 shows the extracting
algorithm.
4.1 Embedding Algorithm
In this section, we describe the embedding algorithm.
After we select the image and type the watermark text,
we transfer the watermark text to binary values and
determine the coordinates of the image which the
watermark will be embedded in. First, we will embed the
length of the watermark text in sixteen pixels starting
from the first coordinate which we select until we embed
it in the sixteen pixels in the 1st LSB. Based on the length
of the watermark text, we can know how many copies it
will be embedded and if we are going to embed in the 2
nd
LSB. Before the watermark will be embedded in the
image in the 1
st
LSB, it will be inversed and we will
change the order of embedding. So, if the coordinate X is
even, it will subtract 1 from X and if X is odd, it will add 1
to X. Then, watermarked image will be produced and it
will be saved. Figure 2 shows the embedding algorithm.
4.2 Extracting Algorithm
In this section, we will describe the extracting algorithm.
After receiving the watermarked image, we will get the
length of the watermark text from the 1
st
LSB in the
sixteen pixels starting from the determined coordinates
until we get it from the sixteen pixels. After getting the
length, we can know how many copies the sender has
embedded. So, we can choose any copy to be displayed
and also we can get the embedded watermark in the
second LSB, if there. Then, the algorithm will check if the
coordinate X is even it will subtract 1 from X of if the
coordinate X is odd, it will add 1 to X. After that, it will
get the watermark bit which will be inversed. Finally, the
watermark bits will be transferred to characters which
will be shown as the watermark text. Figure 3 shows the
extracting algorithm.
Input:
1- Cover Image
2- Watermark text.
Output:
Watermarked Image
Begin
1- Check the length of the watermark text to know how
many copies will be embedded in the first LSB and if it
will embed in the second LSB.
2- Embedding the length of the watermark text in the first
LSB.
3- convert the watermark text from characters to bits.
4- Inverse the watermark bit.
5- Check the coordinate of X, if it is odd, the algorithm will
add 1 to X, and if it is even, the algorithm will subtract 1
from X.
6- Embed the watermark bit in the first LSB.
7- Go to 4 until finishing all the watermark.
8- Go to 4 if we need to embed another copy of the
watermark text.
9- Save the Image as bitmap image
End
Fig. 2. Embedding Algorithm
Input:
Watermarked Image.
Output:
Watermark text.
Begin:
1- Get the length of the watermark text from the first LSB.
2- The user can choose which copy he wants if there is more
than one copy.
3- Check the coordinate of X, if it is odd, the algorithm will
add 1 to X, and if it is even, the algorithm will subtract 1
from X.
4- Get the bit from the first LSB.
5- Converse the bit and save it in array.
6- Go to 3 until finishing all the watermark text.
7- Convert the array to characters.
End
Fig. 3. The extracting algorithm
5 E
XPERIMENTAL RESULTS AND DISCUSSION
In our experimental results, we have four 512x512
grayscale BMP images which are shown in figure 4 were
used as cover images. The size of every cover image is 257
kilo-bytes. We have tested different watermark text in
different places of the pixels. As it is known that every
pixel in the grayscale image contains 8 bits. And every bit
have value, for example the first bit in the right has the
Fig. 1. The framework of the proposed method
watermark
Original Image
watermark
Retrieving Watermark
Watermarked Image
Embed Watermark in Image
using watermarking
Algorithm
Page 3
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 4
value 1 and it is called first LSB, and the second bit has
the value 2 and it is called the second LSB, and the third
bit has the value 4 and it is called the third LSB, and the
fourth bit has the value 8 and it is called the fourth LSB
and so on. We have embedded the watermark text in the
first LSB and also in the second LSB and in the third LSB
and in the forth LSB and combined the first with the
second LSB and the first with the third LSB and the first
with the forth LSB and also we combined the second with
the third LSB and also the second with the fourth LSB and
we also combined the third with the fourth LSB. All of
them will be explained in this paper. Table 1 shows the
LSB uses and its maximum capacity and the size of the
watermarked LSB.
(
A
)
(
B
)
(
C
)
(
D
)
Fig. 4. The cover images: (A) Dock (B) Forest (C) Waterfall (D) Toco
Toucan
5.1 The First LSB
Once, we embed maximum capacity of the watermark
text which contains from 32766 bytes and 2 bytes to
embed the length of the watermark text in determined
pixels in the first LSB in the proposed algorithm and the
traditional LSB [19] and Lee's algorithm [14]. Then, we
got the watermarked images without noticeable
distortion. The second time, we embed different
watermark text which contains also from 32766 bytes and
2 bytes to embed the length of the watermark text in the
four images by the proposed algorithm and the
traditional LSB [19] and Lee [14] and we also got
watermarked images without noticeable distortion on
them. By the way, the changes in the first LSB can't be
detectable by the naked eyes because the maximum
change in every pixel is 1.
5.2 The Second LSB
In this algorithm, we embedded the same watermark text
which (we embed in the first LSB) contains from 32766
bytes and 2 bytes to embed the length of the watermark
text in determined pixels in the second LSB. Then, we got
the watermarked images without noticeable distortion
because the maximum change in every pixel is 2.
5.3 The Third LSB
In this algorithm, we embedded the same watermark text
which contains from 32766 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
third LSB and then, we got the watermarked images with
some noticeable distortions in watermarked dock and
Toco Toucan because the maximum change in every pixel
is 4 and it is somehow noticeable.
5.4 The Fourth LSB
When we embedded the same watermark text which
contains from 32766 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
fourth LSB and, we got the watermarked images with
some distortion in all watermarked because the
maximum change in every pixel is 8 and the 8 grade
difference is noticeable.
5.5 Combination First and Second LSB
When we embedded the maximum capacity of the
watermark text which contains from 65532 bytes and 2
bytes to embed the length of the watermark text in
determined pixels in the first and second LSB which is the
proposed algorithm when the watermark text is more
than 32766 bytes in this size of images, we got the
watermarked images without any distortion in the
watermarked images because the maximum change in
every pixel is 3.
5.6 Combination First and Third LSB
When we embedded the same watermark text which
contains from 65532 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
first and third LSB and then, we got the watermarked
images with some distortion in watermarked dock and
Toco Toucan because the maximum change in every pixel
is 5 and it is somehow noticeable.
5.7 Combination First and Fourth LSB
When we embedded the same watermark text which
contains from 65532 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
first and fourth LSB, we got the watermarked images
with some distortion in the watermarked images because
the maximum change in every pixel is 9 and it is
noticeable.
5.8 Combination Second and Third LSB
When we embedded the same watermark text which
contains from 65532 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
Second and third LSB, we got the watermarked images
Page 4
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 5
with some distortion in watermarked dock and Toco
Toucan because the maximum change in every pixel is 6
and it is noticeable.
5.9 Combination Second and Fourth LSB
When we embedded the same watermark text which
contains from 65532 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
Second and fourth LSB, we got the watermarked images
with some distortion in watermarked images because the
maximum change in every pixel is 10 and it is noticeable.
Table 1
The different uses of LSB and its maximum capacity and
the size of the 512x512 BMP watermarked images
Image Which LSB Watermark
embedded
Size of
watermarked
image
dock first 32766 bytes 257 KB
forest first 32766 bytes 257 KB
waterfall first 32766 bytes 257 KB
Toco Toucan first 32766 bytes 257 KB
dock second 32766 bytes 257 KB
forest second
32766 bytes 257 KB
waterfall second
32766 bytes 257 KB
Toco Toucan second
32766 bytes 257 KB
dock Third 32766 bytes 257 KB
forest Third 32766 bytes 257 KB
waterfall Third 32766 bytes 257 KB
Toco Toucan Third 32766 bytes 257 KB
dock Fourth 32766 bytes 257 KB
forest Fourth 32766 bytes 257 KB
waterfall Fourth 32766 bytes 257 KB
Toco Toucan Fourth 32766 bytes 257 KB
dock First and second 65532 bytes 257 KB
forest First and second 65532 bytes 257 KB
waterfall First and second 65532 bytes 257 KB
Toco Toucan First and second 65532 bytes 257 KB
dock First and Third 65532 bytes 257 KB
forest First and Third 65532 bytes 257 KB
waterfall First and Third 65532 bytes 257 KB
Toco Toucan First and Third 65532 bytes 257 KB
dock First and Fourth 65532 bytes 257 KB
forest First and Fourth 65532 bytes 257 KB
waterfall First and Fourth 65532 bytes 257 KB
Toco Toucan First and Fourth 65532 bytes 257 KB
dock Second and third 65532 bytes 257 KB
forest Second and third 65532 bytes 257 KB
waterfall Second and third 65532 bytes 257 KB
Toco Toucan Second and third 65532 bytes 257 KB
dock Second and Fourth 65532 bytes 257 KB
forest Second and Fourth 65532 bytes 257 KB
waterfall Second and Fourth 65532 bytes 257 KB
Toco Toucan Second and Fourth 65532 bytes 257 KB
dock Third and Fourth 65532 bytes 257 KB
forest Third and Fourth 65532 bytes 257 KB
waterfall Third and Fourth 65532 bytes 257 KB
Toco Toucan Third and Fourth 65532 bytes 257 KB
5.10 Combination Third and Fourth LSB
When we embedded the same watermark text which
contains from 65532 bytes and 2 bytes to embed the
length of the watermark text in determined pixels in the
third and fourth LSB, we got the watermarked images
with some distortion in all watermarked images because
the maximum change in every pixel is 12 and it is
noticeable.
Table 1 shows different combination of LSB for
embedding the watermark. The embedded watermark
text was increased when we combine 2 LSB.
6 P
EAK SIGNAL TO NOISE RATIO
(
PSNR
)
Notice that, there is no difference between the original
and watermarked images in the first and second LSB by
using our naked eyes. No distortion occurs for these
watermarked images. We found some distortion when we
embed the watermark text in the third and fourth LSB
and also when we combined them. We got the result after
we calculated the Peak signal-to-noise ratio (PSNR).
The PSNR value was used to evaluate the quality of
the watermarked images. The phrase peak signal-to-noise
ratio (PSNR) is most commonly used as a measure of
quality of reconstruction in image compression [14]. It is
the most easily defined via the Mean Squared Error
(MSE) which for two mXn images I and K where one of
the images is considered as a noisy approximation of the
other (in other words, one is the original and the other is
the watermarked image). MSE is defined as the following
equation (2) and the PSNR is defined in equation (1).
)(10log*10
2
MSE
MAX
PSNR
I
=
(1)
)(10log*20
MSE
MAX
I
=
Where MAX is equal to 255 in grayscale images, MSE
is the mean square error, which is defined as:
=
=
=
1
0
1
0
2
)],(),([
*
1
m
i
n
j
jiKjiI
nm
MSE
(2)
Where I is the original image and K is the
watermarked image.
Typical values for the PSNR are between 30dB and
40dB [14]. If the PSNR of the watermarked image is more
than 30, it is hard to be aware of the differences with the
cover image by the human eyes system. The cover images
are shown in Figure 3. As it is explained, the invisibility
of the watermark in the proposed algorithm is good. And
the original image and the watermarked image cannot be
distinguished by human visibility system (HVS) in some
of the watermarked images. We have calculated the PSNR
of all watermarked images and the result is shown in
table 2 and table 3 and we have done a comparison
between the proposed algorithm and the traditional LSB
[19] and Lee's algorithm [14] when we embedded the
same watermark text and it is shown in Table 4. The
result of PSNR of the four images are more than 54 when
we embed 32766 bytes as a watermark text in the second
time by embedding different watermark text and we
compare between our propose algorithm and the
Page 5
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 6
traditional LSB [19] and Lee's [14] Algorithm and we got
the best results of them.
Table 2
Comparison of the PSNR of the watermarked images in
the first and second and third and forth LSB
Images First
LSB
Second
LSB
Third
LSB
Forth
LSB
Dock 54.5961 48.6361 42.5866 36.5826
Forest 54.6673 48.6080 42.5669 36.5054
Waterfall 54.6216 48.5651 42.5747 36.5716
Toco Toucan 54.6899 48.5925 42.5863 36.5180
Table 3
Comparison of the PSNR of the watermarked images in
the combined LSB
Images
Dock Forest Waterfall Toco
Toucan
First &
Second LSB
48.0080 50.8632 47.9522 47.9606
First & Third
LSB
43.2265 43.2154 43.2277 43.2921
First &
Fourth LSB
37.5494 37.5498 37.6220 37.6079
Second &
Third LSB
41.9395 41.9030 41.9130 41.6953
Second &
Fourth LSB
37.2002 37.1592 37.2387 34.2102
Third &
Fourth LSB
35.8711 35.8387 35.9511 35.8911
Table 4
Comparison of the PSNR of the watermarked images
Between the proposed algorithm and Lee's algorithm [14]
and the traditional LSB [19]
7 A
TTACKS ON THE WATERMARKED IMAGE
We have tested three types of attacks which are cropping
and adding noise and JPEG compression in the
watermarked images. The purpose of these attacks is to
proof the robustness of our algorithm.
7.1 Cropping
We tested the watermarked images by cropping or
resizing the watermarked images from 512x512 pixels to
448x448 pixels as they are shown in Figure 5. In fact, we
lost some of the information after we cropped the
watermarked images but we still have the information
which is in the cropped images. Since the algorithm
embeds many copies of the watermark text if it is not
much, so we still have the information in the cropped
images.
(a) (b)
(c) (d)
Fig. 5. The cropped watermarked images: (A) Dock (B) Forest (C)
Waterfall (D) Toco Toucan
7.2 Adding Noise
We tested the watermarked images by adding noise 'salt
and pepper' in the watermarked images as they are
shown in Figure 6. In fact, we lost little of the watermark
text which does not affect on the watermark text that
much. And also, if the watermark text is not much, that
will give us many copies of the watermark text. So, we
can see the copies to compare the changes.
Imag
e
The first watermark text
which contain from 32766
bytes
The second watermark text
which contain from 32766
bytes
Propos
ed
Algori
thm
Lee's
Algori
thm
LSB
Algori
thm
Propos
ed
Algori
thm
Lee's
Algori
thm
LSB
Algori
thm
Dock
54.5961 53.7041 53.6950 54.4636 53.8333 53.8282
Fores
t
54.6673 53.7650 53.7511 54.5066 53.8906 53.8720
Wate
rfall
54.6216 53.7310 53.7216 54.4895 53.8452 53.8330
Toco
Touc
an
54.6899 53.7707 53.7727 54.5402 53.9229 53.9034
Page 6
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 7
(a) (b)
(c) (d)
Fig. 6. The watermarked images with salt and pepper noise: (A) Dock
(B) Forest (C) Waterfall (D) Toco Toucan
7.3 JPEG Compression
As it is known about LSB, Least Significant Bit is weak in
JPEG compression because the image lost the most of LSB
of the watermarked image. So, we have a simple solution
of this problem which is to convert the watermarked BMP
image to JPEG image. Then, we will calculate the
difference between the two watermarked images by using
equation (1).
Difference = watermarked BMP image
– watermarked JPEG image. (1)
Then, the Difference array will be sent with the
watermarked image. If the watermarked image is JPEG,
we will implement equation (2) to get the watermarked in
BMP format:
Watermarked BMP image = Watermarked JPEG image
+ Difference. (2)
After that, we can retrieve the watermark text back
by using the proposed extracting algorithm from the
watermarked BMP image.
8
C
ONCLUSION
This paper proposed a new LSB based digital
watermarking scheme with the combination of LSB and
inverse bit. The experimental result shows that the
proposed algorithm maintains the quality of the
watermarked image. This paper also shows the
experimental results when combining different positions
of LSB such as the second LSB and the third LSB and
fourth LSB and the combination between them. The
proposed algorithm is also tested using Peak signal-to-
noise ratio (PSNR) and the result of PSNR is compared
with the traditional LSB [19] and Lee's algorithm [14]. We
also attack the watermarked image by using cropping and
adding noise and we got good results as well. Therefore,
this new digital watermarking algorithm can be used to
embed watermark inside the image.
A
CKNOWLEDGMENT
This research is supported by Fundamental Research
Grant Scheme (FRGS) Vote 0738.
R
EFERENCES
[1] Muftic, S. (2010), Privacy Issues and Solutions in Real and in
Digital Worlds. Presentation at the AAAS 2010 Conference – San
Diego, February 20, 2010. In
‘http://ec.europa.eu/dgs/jrc/downloads/jrc_aaas2010_privac
y_muftic.pdf’
[2] Ibrahim, R. and Kuan, T. S. (2010) , Steganography Imaging (SIS):
Hiding Secret Message inside an Image. Proceedings of the World
Congress on Engineering and Computer Science 2010, San
Francisco, USA.
[3] Cramer C. (2005), About Digital Watermarking. from
http://www.willamette.edu/wits/idc/mmcamp/watermarking.htm.2005
[4] Serrão, C. and Guimarães, J. (1999), Protecting Intellectual
Proprietary Rights through Secure Interactive Contract Negotiation,
Springer-Verlag Berlin Heidelberg 1999.
[5] Gulati, K. (2003), Information Hiding Using Fractal Encoding.
Thesis for master degree, Mumbai, India.
[6] Mandhani, N. K. (2004), Watermarking Using Decimal
Sequences”, Master thesis submitted to the Graduate Faculty of
the Louisiana State University, India.
[7] Katzenbeisser, S. and Petitcolas, F.A.P., (2000). Information
hiding techniques for steganography and digital watermarking.
Artech House Publishers.
[8] Nagra, J., Thomborson, C. and Collberg, C. (2002), a functional
taxonomy for software watermarking, in M. Oudshoorn, ed., `Proc.
25th Australasian Computer Science Conference 2002', ACS, pp.
177-186.
[9] Bhatnagar, G. and Raman, B. (2008), A new robust reference
watermarking scheme based on DWT-SVD, Elsevier B.V. All rights
reserved.
[10] Luo, H, Chu, S. H. and Lu, Z. M. (2008), Self Embedding
Watermarking Using Halftoning Technique, Circuits Syst Signal
Process (2008) 27: 155–170
[11] Yang, W. C., Wen, C. Y. and Chen, C. H., (2008), Applying
Public-Key Watermarking Techniques in Forensic Imaging to
Preserve the Authenticity of the Evidence. Springer-Verlag Berlin
Heidelberg 2008.
[12] He, H. J., Zhang, J. S. and Tai, H. M., (2006), A Wavelet-Based
Fragile Watermarking Scheme for Secure Image Authentication.
Springer-Verlag Berlin Heidelberg 2006
[13] Byun, S. C., Lee, S. K., Tewfik, A. H. and Ahn, B. H., (2003), A
SVD-Based Fragile Watermarking Scheme for Image Authentication.
Springer-Verlag Berlin Heidelberg 2003
[14] Lee, G. J., Yoon, E. J. and Yoo, K. Y. (2008), A new LSB based
Digital Watermarking Scheme with Random Mapping Function”, in
2008 IEEE DOI 10.1109/UMC.2008.33
[15] Fazli, S. and Khodaverdi, G (2009), Trade-off between
Imperceptibility and Robustness of LSB Watermarking using SSIM
Page 7
JOURNAL OF COMPUTING, VOLUME 3, ISSUE 4, APRIL 2011, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 8
Quality Metrics, in 2010 IEEE DOI 10.1109/ICMV.2009.68
[16] Basu, D., Sinharay, A. and Barat, S. , (2010), Bit Plane Index Based
Fragile Watermarking Scheme for Authenticating Color Image. IEEE,
DOI 10.1109/ICIIC.2010.53
[17] Titty, T., Steganography: Reversible Data Hiding Methods for
Digital Media. Bachelor project.
[18] Tirkel, A., Rankin, G., Schyndel, R. V., Ho, W., Mee, N., AND
Osborne, C. 1993. Electronic watermark. In Proceedings of DICTA.
666–672.
[19] Kurah, C. AND Mchughes, J. 1992. A cautionary note on image
downgrading. In Proceedings of the IEEE Computer Security
Applications Conference. Vol. 2. IEEE Computer Society Press,
Los Alamitos, CA,153–159.
[20] Zheng, D. , Liu, Y. , Zhao, J. and El Saddik, A. ,(2007), A Survey
of RST Invariant Image Watermarking Algorithms, ACM
Abdullah Bamatraf recived his B.Sc degree in Computer science
from Hadramout University of Science and Technology, Mukalla. He
is currently pursuing his study for Master degree in a New Digital
Watermarking Algorithm Using Combination of Least Significant Bit
(LSB) and Inverse Bit at Universiti Tun Hussein Onn Malaysia. His
research area includes Image Processing and Least Significant Bit.
Rosziati Ibrahim is with the Software Engineering Department,
Faculty of Computer Science and Information Technology, Universiti
Tun Hussein Onn Malaysia (UTHM). She obtained her PhD in
Software Specification from the Queensland University of
Technology (QUT), Brisbane and her MSc and BSc (Hons) in
Computer Science and Mathematics from the University of Adelaide,
Australia. Her research area is in Software Engineering that covers
Software Specification, Software Testing, Operational Semantics,
Formal Methods, Data Mining, Image Processing and Object-
Oriented Technology.
Mohd.Najib Mohd.Salleh is a senior lecturer at Faculty of Computer
Science and Information Technology, Universiti Tun Hussein Onn
Malaysia since 2001. He had Bachelors degree in Computer Science
from Universiti Pertanian Malaysia, Selangor in 1988. He received a
Master degree in Computer Science in Information System from
Universiti Teknologi Malaysia in 2000. He completed his PhD in
Data Mining from Universite De La Rochelle, France in 2008. His
doctoral thesis was on decision tree modeling with incomplete
information in classification task problem. His research interests
include uncertainty in decision science, decision theory, artificial
intelligence in data mining and knowledge discovery.
Page 8