Image Watermarking in Spatial Domain Using QIM and Genetic Algorithm

IJSRD - International Journal for Scientific Research & Development| Vol. 2, Issue 07, 2014 | ISSN (online): 2321-0613
All rights reserved by www.ijsrd.com 643
Image Watermarking In Spatial Domain Using Qim and Genetic
Algorithm
Renu Sharma1
Mamta Narwaria2
1
M. Tech Student 2
M. Tech (Ph. D)
1,2
Department of Computer Engineering
1,2
Galgotias University, Greater Noida, India
Abstract— Digital watermarking is one of the proposed
solutions for copyright protection of multimedia data. A
watermark is a form of image or text that is impressed onto
paper, which provides evidence of its authenticity. A digital
watermark is digital data embedded in some host document
so as to later prove the ownership of the document. Digital
image watermarking refers to digital data embedding in
images. Robust image watermarking systems are required so
that watermarked images can resist geometric attacks in
addition to common image processing tasks, such as JPEG
compression. Least Significant Bit (LSB) watermarking, is
one of the most traditional method of watermarking which
changes the LSB of individual pixels in correlation with the
watermark. However, pure LSB scheme provides a fragile
watermarking technique which is not acceptable in practical
applications. Also, robustness against geometric attacks,
such as rotation, scaling and translation, still remains one of
the most challenging research topics in pixel based image
watermarking. In this paper, a new pixel-based
watermarking system is proposed, in which a binary logo is
embedded, a bit per pixel, in the pixel domain of an image.
The LSB based watermarking is then quantized using QIM,
augmented with genetic algorithm to produce a
watermarking scheme which is highly robust against
geometrical attacks.
Key words: Image watermarking, LSB, QIM, Genetic
Algorithm.
I. INTRODUCTION
For digital image watermarking systems, geometric attacks,
such as rotation, scaling and translation, do not distort or
remove the embedded watermark, but instead geometrically
and globally modify the watermarked image to make the
watermark decoder (or detector) unable to re-synchronize
the received image. Most existing robust watermarking
systems are block based and/or rely on the correct
synchronization of the image to extract the embedded
watermark. Geometric attacks destroy the synchronization
and render the extracted embedded watermark incorrect or
the extraction process impossible, thus making the
watermark undetectable. Robust image watermarking
systems, which are used to address security concerns, such
as copyright protection or copy control, should guarantee
resistance to geometric attacks. Several systems have been
proposed to address the problem of robustness against
geometric attacks [1]. Exhaustive search techniques try all
possible combinations of the geometric distortion and can be
computationally costly or infeasible. Methods that embed a
reference pattern into an image in addition to the robust
watermark for aligning the received image at the decoder
can impair either the fidelity or general robustness of the
system. Invariant watermarking systems are designed to be
robust only against certain geometric attacks and the
approaches of autocorrelation and implicit synchronization
also suffer from variant problems. Therefore, robustness
against geometric attacks still remains one of the difficult
challenges in image watermarking research [1, 2].
A. Problem Statement
QIM based watermarking techniques are sensitive in the
context that these quickly transits from easily detectable to
virtually undetectable as this parameter varies. Moreover,
with different values of quantization levels, the resultant
image shows a great degree of dissimilarity from the
original unmarked image. For QIM hiding in images, a
scheme is to be designed so that the watermarked image is
robust to, say, a moderate degree of gaussina noise and the
watermark should be easily detectable. The problem
identified in the paper is to present a robust watermarking
technique using QIM, making use of the genetic approach to
make the Mean Square Error as low as possible.
B. Motivation
Information system security is the need of modern times.
Several of these applications relate to copyright notification
and enforcement for audio, video, and images that are
distributed in digital formats. In these cases the embedded
signal either notifies a recipient of any copyright or
licensing restrictions or inhibits or deters unauthorized
copying. For example, this embedded signal could be a
digital "fingerprint" that uniquely identifies the original
purchaser of the copyrighted work. If illicit copies of the
work were made, all copies would carry this fingerprint,
thus identifying the owner of the copy from which all illicit
copies were made. In another example, the embedded signal
could either enable or disable copying by some duplication
device that checks the embedded signal before proceeding
with duplication. Quantization index modulation (QIM)
techniques have been gaining popularity in the data hiding
community because of their robustness and information-
theoretic optimality against a large class of attacks.
II. RESEARCH APPROACH
In this paper, image watermarking is done using QIM
modulation applied after dithering. The watermark
embedding is augmented with genetic approach to keep the
mean square error as low as possible. The detection of the
presence of QIM hidden data, which is an important
consideration when data hiding is used for covert
communication or steganography, is analysed. For a given
host distribution, the techniques presented are able to
quantify detect ability compactly in terms of a parameter
related to the robustness of the hiding scheme to attacks.
In this paper, grayscale images are considered for
watermark embedding. However, the given set of techniques
is directly applicable to colour images with no modification
in any of Red, Green or Blue colour planes. The given
image block is resized to make the height and width, a

Image Watermarking In Spatial Domain Using Qim and Genetic Algorithm
(IJSRD/Vol. 2/Issue 07/2014/146)
multiple of 3. The complete image is then partitioned into
blocks of size 3X3 for the purpose of watermark embedding,
and one watermark bit is embedded into one such block,
thereby giving an embedding capacity of N/9 bits where N
is the number of pixels in the original image. Average value
of the pixels in the block is computed and then amount of
noise to be added to the block is computed. This is the
process of dithering in which some noise is added before
quantization to make the watermark embedding uniform
throughout the 3X3 pixel block. Quantization of the average
value of the block is then performed depending on whether
1 or 0 is to be embedded in the 3X3 pixel block.
III. METHODOLOGY
The dithering of the image block can be done in several
ways. The proposed method use genetic approach to give
the maximum value of PSNR (or minimum value of mean
square error) so as to maximize the overall value of the Peak
Signal to Noise Ratio.
The proposed method uses a 3X3 pixel block to
embed 1 bit of watermark information. Hence, an image of
size 1/9 of the host image can be embedded as a watermark
in the host image. Also, the watermark image can be
extracted even if the host image is rotated by any angle
multiple of 900
. Thus, the given technique presents a robust
watermark embedding method against rotation clockwise or
counter clockwise in 900
.
For cutting and cropping attacks, the extraction of
the watermark will produce the same fraction of the
watermark to the corresponding fraction of the host image
used for watermark extraction. Thus, the given technique is
also robust against cutting and cropping attacks.
The given technique of watermark embedding is
robust against addition of noise as far as noise data is added
to the selected portions of the image. In such a case, only
corresponding potion of the watermark image are corrupted
and the rest of the watermark can be detected and extracted.
Even if the watermarked image is subjected to random low
noise data, within a threshold limit, the watermark can be
extracted with substantial probability. However, substantial
random amount of noise added throughout the host image
will corrupt the watermark bits and the watermark cannot be
extracted.
However, changing the brightness or contrast of the
image results in changing pixel intensity values, thereby
producing a change in the quantization levels of the host
image. In such a case, the watermarking information can be
destroyed can therefore, the watermark cannot be extracted.
IV. PROPOSED WORK
A. LSB Based watermarking Illustration
This section illustrates the pure LSB based image
watermarking technique. Consider an 8 X 8 pixel grayscale
image block as shown in figure 3.1.
102 255 170 109 121 214 106 71
186 147 223 231 227 244 41 210
150 222 227 98 211 219 242 207
185 252 67 22 246 71 89 58
233 107 123 46 226 37 195 0
147 179 99 168 90 168 206 253
43 18 96 168 69 73 26 48
141 17 191 129 244 225 64 35
Fig. 3.1 8 X 8 pixel block sample of host image
Also, consider the 8 X 8 pixel binary watermark
shown in the figure 3.2.
1 1 1 0 1 0 0 0
1 1 0 0 1 0 0 1
1 0 0 0 0 1 1 1
0 1 0 0 1 1 1 0
0 1 0 1 1 1 0 1
0 1 1 1 0 0 0 1
0 1 1 1 1 1 0 0
0 0 1 0 1 0 0 0
Fig. 3.2 8 X 8 pixel block sample of watermark (Binary
Image)
The binary values of the pixel corresponding to
decimal values of fig 3.1 are as shown in figure 3.3
1100110
1111111
1
1010101
0
1101101 1111001
1101011
0
1101010 1000111
1011101
0
1001001
1
1101111
1
1110011
1
1110001
1
1111010
0
101001
1101001
0
1001011
0
1101111
0
1110001
1
1100010
1101001
1
1101101
1
1111001
0
1100111
1
1011100
1
1111110
0
1000011 10110
1111011
0
1000111 1011001 111010
1110100
1
1101011 1111011 101110
1110001
0
100101
1100001
1
0
1001001
1
1011001
1
1100011
1010100
0
1011010
1010100
0
1100111
0
1111110
1
101011 10010 1100000
1010100
0
1000101 1001001 11010 110000
1000110
1
10001
1011111
1
1000000
1
1111010
0
1110000
1
1000000 100011
Fig. 3.3 8 X 8 binary pixel values of host image
By replacing the LSB of the individual pixels by
the corresponding binary watermark bits, we get the
watermarked image as shown in the figure 3.4
011001
11
111111
11
101010
11
011011
00
011110
01
110101
10
011010
10
010001
10
101110
11
100100
11
110111
10
111001
10
111000
11
111101
00
001010
00
110100
11
100101
11
110111
10
111000
10
011000
10
110100
10
110110
11
111100
11
110011
11
101110
00
111111
01
010000
10
000101
10
111101
11
010001
11
010110
01
001110
10
111010
00
011010
11
011110
10
001011
11
111000
11
001001
01
110000
10
000000
01
100100
10
101100
11
011000
11
101010
01
010110
10
101010
00
110011
10
111111
01
001010
10
000100
11
011000
01
101010
01
010001
01
010010
01
000110
10
001100
00
100011
00
000100
00
101111
11
100000
00
111101
01
111000
00
010000
00
001000
10
Fig. 3.4 8 X 8 LSB substituted binary values of pixels of
watermarked image
The decimal values corresponding to the
watermarked bits of the host image are shown in figure 3.5
which shows the watermarked image block.
103 255 171 108 121 214 106 70
187 147 222 230 227 244 40 211
151 222 226 98 210 219 243 207
184 253 66 22 247 71 89 58
232 107 122 47 227 37 194 1
146 179 99 169 90 168 206 253
42 19 97 169 69 73 26 48
140 16 191 128 245 224 64 34
Fig. 3.5 8 X 8 LSB substituted pixel values of watermarked
image
The Mean Square Error (MSE) is the average of
squares of the difference between the corresponding bits of
the host image and the watermarked image. For image block
and the watermark shown in figure 3.1 and 3.2 respectively,
the MSE is 0.546875.

The Signal to Noise Ratio (SNR) is a quality metric
related with the watermarked image quality and it is
dependent on MSE in the following way:
∑ ∑
∑ ∑
Peak Signal to noise Ratio can be obtained by
substituting 2552
in the numerator instead if individual pixel
values. The PSNR value for this particular pixel block is
50.75192.
B. Quantization Index Modulation
The following figure illustrates the process of 64 levels
Quantization of a grayscale image with 256 intensity levels.
Fig 3.1 Quantization Illustration
The quantization levels corresponding to
embedding of two different bits in the host signal element
can be done as shown in figure 3.3
Fig 3.2 Two levels of Quantization for binary signal
Embedding
C. Dither Modulation
Quantization can be classified as coarse or fine with general
meaning. Coarse quantization often results in the
appearance of abrupt discontinuities in the signal. In images,
these appear as false contours. One technique used to try to
alleviate this problem is called dithering or pseudorandom
noise quantization. The idea is to add a small amount of
random noise to the signal before quantizing. Sometimes
same noise values are stored and then subtracted for display
or other purposes, although this is not necessary. The idea of
adding noise intentionally might seem odd. But in the case
of dithering, the noise actually serves a useful purpose.
Adding noise before quantization has the effect of breaking
up false contours by randomly assigning pixels to higher or
lower quantization levels. This works because our eyes have
limited spatial resolution. By having some pixels in a small
neighborhood take on one quantized level and some other
pixels take on a different quantized level, the transition
between the two levels happens more gradually, as a result
of averaging due to limited spatial resolution.
Let x ϵ RN
be a host signal vector in which the
watermark message m is to be embedded. The message m is
represented by vector b of binary values of opposite
polarity. Thus bi = ±1, for i = 1,2,...NRm, where Rm denotes
the bit rate. The host signal is decomposed into NRm sub
vectors, of length L = floor (1/Rm), denoted by x1, x2.....
xNRm. In Binary Dither modulation, two L dimensional
uniform quantizers Q-1(.) and Q+1(.) are constructed.
Let the step size Δ be 4, then the two dither vectors
are constructed in the following way:
{
where di (0) and di (1) are the ith
components of two
dither subvectors. The quantization step is a measure of the
distance between two dithered subvectors and thereby
provides a quantitative measure of robustness of the system.
Consider the following example:
x = {251, 255, 252, 247, 230, 241, 245, 253, 247, 230, 241,
230, 241, 245, 230, 241}, m = {1, -1, -1, 1, -1, 1,-1, 1}, N=8
Rm = 2, x1 = {251, 255}, bi = {1}
let x1
denotes the mean of x. Then x1
= 253.
For 128 levels QIM, the quantization levels can be
tabulated as shown below:
Pixel Intensity
Mapping Intensity
(Reconstruction Point)
From To Bit 0 (-1) Bit 1 (+1)
0 To 2 0 2
3 To 5 3 5
. . . . .
. . . . .
. . . . .
253 To 255 253 255
Table 3.1 Two Level Dither Quantization Mapping
For the first block x1
therefore, the quantization
point is 253.
D. Proposed Average Value Based Quantization Indexed
Modulation augmented with Genetic Approach.
For a given image I with width w and height h, a registered
binary logo Z of the dimension 1/3 of Iin both height and
width is selected. Z may be constructed through tiling,
scaling or cropping from a logo that is smaller or larger than
I. Assume that s ≡(i, j), 0 ≤i< w and 0 ≤ j < h, represents a
site in the image I; Is and Zs then represent respectively the
corresponding pixel in I and the bit in the logo Z. The
process of embedding Z into the host image I is as follows.
For pixel Is, a local window centered at s, is
chosen. Let Gs be the set of pixels in the local window, Gs∈
I.
Let μs represent the average value of the pixels in
Gs. For the given embedding bit Zs, QIM embedding
function modifies μs as:
μ's= q(μs+ d(Zs)) − d(Zs)
where q is the scalar quantization function and d is
the corresponding dither modulation obtained by
multiplying Zi with a suitable positive constant. The
corresponding watermarked block is given by
Is
'
= Is±α*( μ's -μs)
Where α ϵ [0, k] is a weighting factor.
The intensities are to be added and subtracted in the
block as per the following scheme, thereby giving next
generation chromosomes from the first generation
0
20
40
0 5 9 13 16 21 25 29 33
to to to to to to to to to
4 8 12 16 20 24 28 32 36
Quantization Level
0
10
20
30
40
50
0 6 121824303642
Bit 0
Bit 1

chromosome. Among the next generation chromosomes, the
one with the highest value of PSNR is selected.
For pixel Is, the local window consists of a matrix
of nine pixels, including, along with Is, all the pixels
surrounding it. After setting up the intensity values, the
window is moved to the next block and the same procedure
is repeated again until the whole image is watermarked.
From the first generation chromosome several next
level chromosomes are possible out of which three are
illustrated as shown. The one with highest value of PSNR is
selected.
E. Decoder for the watermark
Let I'
be the image arrived at decoder. This image is
subjected to several distortions in the noisy channel. These
distortions may include Gaussian noise or an attack by the
intruder. For pixel at site s, I'
s ∈ I'
, the average intensity, μ'
s,
of a local window Gs centered at I'
s is used to calculate the
distances between μ'
s and the nearest bit 0 and bit 1
quantizer, denoted as d0
S and d1
S respectively, where Δ
being the quantization step for q. This gives the value of the
watermark bit that is embedded cumulatively in the nine
pixel block of the image. In this scheme, the watermark bit
is embedded in all the nine pixels of the window, thereby
providing an embedding capacity of N/9 bits where N is the
number of pixels in the image to be watermarked. Moreover,
preferably, the dimensions of image should be a multiple of
3.
F. Illustration of watermarking technique
1) Watermark Embedding
Consider the image block shown in the figure given below:
210 214 209 208 213 206
210 215 208 205 212 206
211 214 208 205 214 205
211 214 207 208 215 205
210 213 207 206 215 207
210 215 208 207 214 207
The average intensity values of this pixel block, is
as shown:
211 208.2222
210.5556 209.3333
In this scheme, one can dither modulate the 3X3
window of the pixels to make the average value, a whole
number instead of a natural number. This dither can be done
in such a way to minimize the MSE of the pixel block.
The summation of the values of the nine pixel
block is illustrated as shown:
1899 1874
1895 1884
To make the average value a whole number, the
summation must be a multiple of two. This can be done by
subtracting the remainder (or adding nine minus remainder)
to the average value, when the remainder is divided by
9.This is equivalent to adding prior noise before the actual
embedding of watermark which is referred to as dithering.
One possible dithering is
1899 +0 1874 - 2
1895 + 4 1884 - 3
The corresponding image block after dithering is:
210 214 209 208 213-1 206
210 215 208 205 212 206
211 214 208 205 214-1 205
211+1 214 207 +1 208-1 215-1 205
210 213+1 207+1 206 215-1 207
210 215 208 207 214 207
The above pixel values forms one solution. The
block which gives least MSE is to be chosen.
The average pixel values now obtained are:
211 208
211 209
Consider the portion of the quantization table with
some pre specified quantization levels as shown below:
Pixel Intensity Embedding
From To Bit 0 Bit 1
0 4 0 4
5 10
11 15
16 20 16 20
.
205 210 205 210
211 215 211 215
216 220 216 220
221 225 221 225
.
250 255 250 255
The average watermarked values quantized for
1011, taken row-wise can be obtained with the quantization
scheme:
Watermark
1 0
1 1
Quantized averages
215 205
215 210
Difference between original and modified average values
4 -3
4 1
Let the watermark strength factor be K = 3, then,
one possible image reconstruction would be:
210+2*
4
214+3*
4
209-
2*4
208+2*(-
3)
212+3*(-
3)
206-2*(-
3)
210+2*
4
215+3*
4
208-
2*4
205+2*(-
3)
212+3*(-
3)
206-2*(-
3)
211+2*
4
214+3*
4
208-
2*4
205+2*(-
3)
213+3*(-
3)
205-2*(-
3)
212+2*
4
214+3*
4
208-
2*4
207+2*1 214+3*1 205-2*1
210+2*
4
214+3*
4
208-
2*4
206+2*1 214+3*1 207-2*1
210+2*
4
215+3*
4
208-
2*4
207+2*1 214+3*1 207-2*1
In the above pixel matrix, in each 3X3 pixel
window, the middle column pixels are added with 3 times
the difference of the pixel values between quantized and
unquantized values. This is for the simple reason that to

make the average value to be increased by α for a total of 9
pixels, 9α is to be added to the sum of all 9 pixel values.
The above matrix on solution gives
218 226 201 202 203 212
218 227 200 199 203 212
219 226 200 199 204 211
220 226 200 209 217 203
218 226 200 208 217 205
218 227 200 209 217 205
2) Watermark Detection
Let the watermarked image block; added with Gaussian
noise be as shown in fig.
218 +1 226 201 202 203 212
218 227 200 199 203 212
219 226 200 199 204 +1 211
220 226 +1 200 209 217 203
218 226 200 208 217 205
218 227 200 209 217 +1 205
Moreover, rotated by an angle of 90 degrees
counter clockwise, the following will be the pixel matrix:
212 212 211 203 205 205
203 203 205 217 217 218
202 199 199 209 208 209
201 200 200 200 200 200
226 227 226 227 226 227
219 218 219 220 218 218
The average value of the window corresponding to this
block is
205.1111 210.1111
215.1111 215.1111
Matching with the nearest quantization yields
0 1
1 1
which is the 90 degree counter clockwise rotation of original
watermark.
1 0
1 1
Thus, the watermark remains intact on counter
clockwise or clockwise rotations of multiples of 90 degrees
and the extracted watermark is obtained rotated in the
similar way. This is due to the symmetric property of the
3X3 pixel window which is used to embed 1 bit binary
value. The mean square error for the image block is 46.67,
thereby giving the PSNR value 30.31.
V. ANALYSIS OF PROPOSED WORK
A. Quantization Levels Illustration
Consider the grayscale image of Lena shown in Figure 4.1.
Image of dimensions 512 X 512 is considered to make the
visual artifacts clear which are produced as a result of
quantization.
Fig. 4.1: Lena Image with 256 Quantization Levels (8 bit
Grayscale image)
Fig. 4.2: Lena Image with 128 Quantization Levels
Code Mapping
Pixel Intensity Range Mapping Intensity Value
0 - to - 1 0
2 - to - 3 3
4 - to - 5 5
. .
. .
. .
254 - to - 255 255
Table 4.1: Code Book For 128 Level Quantization
Code Mapping
0 - to - 3 0
4 - to - 7 7
8 - to - 11 11
. .
. .
. .
252 - to - 255 255
Table 4.2 Code Book For 64 Level Quantization
Fig 4.4 Lena Image with 32 Quantization Levels
Code Mapping
Pixel Range Mapping Intensity Values
0 - to - 7 0
8 - to - 15 15
16 - to - 23 11
. .
. .
. .
248 - to - 255 255

Code Mapping
0 - to - 15 0
16 - to - 31 31
32 - to - 47 47
. .
. .
. .
240 - to - 255 255
Code Mapping
0 - to - 31 0
32 - to - 63 63
64 - to - 96 96
. .
. .
. .
224 - to - 255 255
Fig 4.7 Lena Image with 4 Quantization Levels
Code Mapping
0 - to - 63 0
64 - to - 127 127
128 - to - 192 192
192 - to - 255 255
Fig 4.8 Lena Image with 2 Quantization Levels (Black and
White Image)
Code Mapping
0 - to - 127 0
128 - to - 256 256
Consider the binary logo shown in Figure 4.10 with
dimensions 150 X 99.
Fig. 4.9: Binary Logo as watermark
The watermarked image with the binary logo is shown in
figure 4.11
Fig 4.10 Watermarked image (watermark being embedded
in the top right corner)
VI. WATERMARK EXTRACTION
The extracted watermark under Gaussian noise addition can
be done with the techniques described in section 3.4 and is
shown in the figure 4.12 below:
Fig. 4.11: Watermark Extraction under Gaussian Noisy
Perturbation Channel
As the watermark is embedded symmetrically in 3 X 3 block
of pixels, the watermark embedding is robust against any
rotation of 90, 180, and 270 degree and will produce the
same output, rotated by the same amount clockwise or
anticlockwise. However, under other degrees the extracted
watermark is fragile and can be face false negative. The
extracted watermark is illustrated in the following table:
Table 4.8 Watermark Extracted Under Various Distortions
Levels

A. Quantization Levels and PSNR
For the image Lena of dimensions 513 X 513, and the
binary watermark (copyright symbol) of dimension 171 X
171, the PSNR metric can be plotted as a function of
quantization levels as shown in the table given below:
Quantization Levels PSNR
128 45.71
112 41.42
96 39.91
80 31.54
64 29.38
48 27.54
32 24.28
16 20.19
VII. CONCLUSION AND FUTURE SCOPE
Our detection-theoretic results for grayscale images show
that the ease with which QIM can be detected depends
strongly on the host statistics and quantization levels along
with the strength parameters. Specifically, with large
quantization values or large window size of QIM based
hiding, the watermarking becomes robust and becomes easy
to detect. This characteristic does hold for typical transform
domain image data, which has strong robustness. While the
knowledge of host distribution assumed in this detection-
theoretic analysis does not hold for image data (where the
statistics can vary significantly from image to image),
standard genetic algorithm techniques are shown to perform
well for image fidelity. QIM is inherently easily detectable
if the quantization levels are small and leads to a lossy
compression. The detectability could be reduced by
reducing the design level of robustness against attacks, or by
reducing the embedding rate. More fundamentally, this work
considers currently proposed QIM schemes, to make them
robust against quarter rotations in clockwise or
anticlockwise direction, while at the same time, keeping the
Mean Square Error as low as possible. This work also
presents an issue of whether it is possible to design QIM
schemes that are both robust and covert, and point to some
recent theoretical results that indicate the potential for such
schemes.
REFERENCES
[1] F. Hartung and M. Kutter, "Multimedia
watermarking techniques," Proceedings of the
IEEE, vol. 87, pp. 1079-1107, 1999.
[2] M. D. Swanson, M. Kobayashi, and A. H. Tewfik,
"Multimedia dataembedding and watermarking
technologies," Proceedings of the IEEE, vol. 86,
pp. 1064-1087, 1998.
[3] B. Chen and G. W. Wornell, "Quantization index
modulation: A class of provably good methods for
digital watermarking and information embedding,"
IEEE Transactions on Information Theory, vol. 47,
pp. 14231443, 2001.
[4] B. Chen and G. W. Wornell, "Quantization index
modulation methods for digital watermarking and
information embedding of multimedia," Journal of
Vlsi Signal Processing Systems for Signal Image
and Video Technology, vol. 27, pp. 7-33, 2001.
[5] M. Wu and B. Liu, "Data hiding in image and
video: Part I - Fundamental issues and solutions,"
IEEE Transactions on Image Processing, vol. 12,
pp. 685-695, 2003.
[6] M. Wu, H. Yu, and B. Liu, "Data hiding in image
and video: Part II - Designs and applications,"
IEEE Transactions on Image Processing, vol. 12,
pp. 696-705, 2003.
[7] E.A. Lee and D.G. Messerschmitt, Digital
Communication, 2 nd edn., Boston, MA: Kluwer
Academic Publishers, 1994.
[8] R. Zamir and M. Feder, “On Lattice Quantization
Noise,” IEEE Transactions on Information Theory,
vol. 42, 1996, pp. 1152– 1159.
[9] F.Perez-Gonzalez,C Mosquera, Mauro Barni and
A. Abrardo, “Rational Dither Modulation: a high-
rate data-hiding method robust to gain attacks,
“IEEE Trans. On Signal Processing, vol. 53, no.
10, pp. 3960-3975, October 2005, Third
supplement on secure media.
[10]Qian Zhang and Nigel Boston, “Quantization Index
Modulation using the E8 lattice”
[11]Brian Chen and Gregory W. Wornell “Digital
Watermarking And Information Embedding Using
Dither Modulation”
[12]F. Guerrini, R. Leonardi, and M. Barni, “Image
watermarking robust to non-linear value-metric
scaling based on higher order statistics,” in
Proc.Int. Conf. Acoustics, Speech Signal Process.
(ICASSP), Toulouse,France, May 2006.
[13]Shiju K P ,Tamil Selvi P “Performance Analysis Of
High Dynamic Range
[14]Image Watermarking Based On Quantization Index
Modulation”.
[15]Prabhishek Singh, R S Chadha, “A Survey of
Digital Watermarking Techniques, Applications
and Attacks”, International Journal of Engineering
and Innovative Technology (IJEIT) Volume 2,
Issue 9, March 2013
[16]Vidyasagar M. Potdar, Song Han, Elizabeth Chang,
“A Survey of Digital Image Watermarking
Techniques”, 3rd International Conference on
Industrial Informatics (INDIN 2005) IEEE.
[17]Namita Chandrakar, Jaspal Bagga, “Performance
Comparison of Digital Image Watermarking
Techniques: A Survey”, International Journal of
Computer Applications Technology and Research
Volume 2– Issue 2, 126 - 130, 2013
[18]Deepshikha Chopra, Preeti Gupta, Gaur Sanjay
B.C., Anil Gupta, “Lsb Based Digital Image
Watermarking For Gray Scale Image”, IOSR
Journal of Computer Engineering (IOSRJCE),
2012
0
20
40
60
128 96 64 32
PSNR

[19]C. Parthasarathy, Dr. S. K. Srivatsa, “Increased
Robustness Of Lsb Audio Steganography By
Reduced Distortion Lsb Coding”, Journal of
Theoretical and Applied Information Technology
[20]Amit Singh, Susheel Jain, Anurag Jain, “Digital
watermarking method using of second Least
Significant Bit (LSB) with the inverse of LSB ”,
International Journal of Emerging Technology and
Advanced Engineering Web Site: www.ijetae.com,
February 2013
[21]Abdullah Bamatraf, Rosziati Ibrahim And Mohd.
NajibMohd. Salleh, “A New Digital Watermarking
Algorithm Using Combination Of Least Significant
Bit (Lsb) And Inverse Bit”, Journal Of
Computing, April 2011.
[22]Gurpreet Kaur, Kamaljeet Kaur, “Image
Watermarking Using LSB (Least Significant Bit)”,
International Journal of Advanced Research in
Computer Science and Software Engineering, April
– 2013.
[23]Ms. E. Suneetha, Ms. P. Sridevi, Smt. D. Swetha,
Ms. E. Sumalatha, “Steganography using LSB
algorithm and RGB decomposition”, Journal Of
Information, Knowledge And Research in
electronics and communication engineering, 2012.
[24]Harshitha K M, Dr. P. A. Vijaya, “Secure Data
Hiding Algorithm Using Encrypted Secret
message”, International Journal of Scientific and
Research Publications, June 2012.

Image Watermarking in Spatial Domain Using QIM and Genetic Algorithm

More Related Content

What's hot (20)

Viewers also liked (12)

Similar to Image Watermarking in Spatial Domain Using QIM and Genetic Algorithm (20)

More from ijsrd.com (20)

Recently uploaded (20)

Image Watermarking in Spatial Domain Using QIM and Genetic Algorithm