pixels that vary only in intensity, not color. Gray-scale imagestherefore have only a single band. Without loss of generality,the remaining discussion will focus on gray-scale images. Thediscussion is easily extended to cover color images by notingthat a color image is the composition of three individual gray-scale images representing the red, green and blue bands. Thetypical gray-scale image has an 8-bit depth which is sufficientto represent 256 unique intensity values ranging from black to white [9]. A brief review of binary representation will beinstructive when interpreting bit-level pixel data in the contextof a digital image. An 8-bit binary numeral has the generalform
A
7
2
7
+
A
6
2
6
+
...
+
A
1
2
1
+
A
0
2
0
where
A
n
represents a single binary digit. In a digital imageit is clear that
A
7
is the most significant bit and indicateswhether the pixel value is greater than 127. A common meansof converting a grayscale image to a binary (i.e. black-and-white) image is to extract the
A
7
bit from each pixel. Bycontrast,
A
0
embodies relatively little information and, in thecontext of a digital image, can generally be understood as anoise channel.
B. Overview of Steganograhy
Steganography hides secret messages under the cover of a carrier signal so it cannot be seen or detected [6], [8],[11]. Steganography technique should generally possess twoimportant properties: good visual/statistical imperceptibilityand a sufficient payload. The first is essential for the securityof hidden communication and the second ensures that a largequantity of data can be conveyed [10]. Two levels of protectioncan be done if the message is encrypted before hiding it, soit must be decrypted before reading it. Invisible watermarkingis treated as a subset of steganography [10].
Fig. 1. Embedding of a secret message into the three least significant channelsof a cover image
The difference is that steganography conceals a messageso that this hidden message is the object of the communi-cation where in watermarking; the hidden message providesimportant information about the cover media, such as au-thentication or copyright. Steganography, in the simplest case,capitalizes on this overabundance of information by replacingthe noise channels (i.e. the least significant bit channels)with an arbitrary secret message. Figure 1 gives an overviewof a steganographic process flow. A source image, hereafterreferred to as a cover, is viewed as 8 information carryingchannels. A secret message is spread over the least significantchannels (in this case the three least significant channels)with the modified channels re-combined to obtain an output,hereafter referred to as the stego image, that visually resemblesthe cover image and contains the injected message.III. PVD M
ETHOD FOR
G
RAY
-L
EVEL
I
MAGE
The pixel-value differencing (PVD) method [7] segmentsthe cover image into nonoverlapping blocks containing twoconnecting pixels and modifies the pixel difference in eachblock (pair) for data embedding. A larger difference in theoriginal pixel values allows a greater modification. The hidingalgorithm is described as follows:1) Calculate the difference value
d
i
for each block of twoconsecutive pixels
P
i
and
P
i
+1
,
d
i
=
P
i
+1
−
P
i
2) Find the optimal
R
i
of the
d
i
such that
R
i
=
min
(
u
i
−
k
)
, where
u
i
≥
k
,
k
=
|
d
i
|
and
R
i
∈
[
l
i
,u
i
]
3) Decide
t
bits of secret data which are hidden with each
d
i
, i.e. each block of two consecutive pixels is definedas
t
=
log
2
(
w
i
)
where
w
i
is the width of the
R
i
4) Read
t
bits binary secret data one by one according toStep 3, and then transform
t
into decimal value
b
. Forinstance, assume a binary secret data is
101
, then
b
= 5
.5) Calculate the new difference value
d
i
using:
d
i
=
l
i
+
b
,for
d
i
≥
0
or
d
i
=
−
(
l
i
+
b
)
, for
d
i
<
0
6)
P
i
and
P
i
+1
are modified to hide t secret data by the fol-lowing formula:
(
P
i
,P
i
+1
) = (
P
i
−
m
2
,P
i
+1
+
m
2
)
:
d
i
∈
odd
or
(
P
i
,P
i
+1
) = (
P
i
−
m
2
,P
i
+1
+
m
2
)
:
d
i
∈
even
where
m
=
d
i
−
d
i
. Finally, we compute thevalues of
(
P
i
,P
i
+1
)
which represent the secret data.7) Repeat Steps 1-6, until all secret data are hidden intothe cover image and the stego-image is obtained.In the extraction phase, the original range table is necessary.It is used to partition the stego-image by the same method asused to the cover image. The extraction phase is implementedas follows:1) Calculate the difference value
d
i
between each twosuccessive pixels for each block
(
P
i
,P
i
+1
)
from thefollowing formula :
d
i
=
|
P
i
+1
−
P
i
|
2) Find the optimum
R
i
of the
d
i
just as in Step 2 in thehiding phase.3) Obtain
b
by subtracting
l
i
from
d
i
. The
b
value repre-sents the value of the secret data in decimal.4) Convert
b
into binary then find number of bits
t
fromthe secret data, where
t
=
log
2
(
w
i
)
[7]
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 8, November 2010157http://sites.google.com/site/ijcsis/ISSN 1947-5500