Welcome To International Journal of Engineering Research and Development (IJERD)
Welcome To International Journal of Engineering Research and Development (IJERD)
Welcome To International Journal of Engineering Research and Development (IJERD)
s
=
otherwise
t y x f if
y x U
, 5
, ( , 4
) , (
The next scheme describes the embedding algorithm:
Embedding Algorithm
Input: The cover-image and the embedding image message, the seed key and two
modulus numbers m
u
, m
l
.
Output: The stego-image and secret key.
Step1: Find the workable pixel p
c
(i) in cover image C by using the pseudo-random
number generator of seed sk .
Step2: Compressed the embedded-image
-Apply wavelet transform.
-Convert the coding matrix into bit-string B
s
Step 3: Set a threshold value T and the two modulus values m
l
, mu. then compute a
residue,
remainder
g
and the possible capacity in a pixel,
c e
g
, as following
form:
If p
c
(i) >T , compute
u
m
2 ec
Log g =
,
u c remaimder
m mod (i) p g =
Else, compute
ml
2 ec
Log g =
,
l c remaimder
m mod (i) p g =
Where p
c
(i) denotes the intensity of the i-th. Pixel with order of
top-down and left to-right in a cover-image C and
denotes the
truncate value.
Step 3: Compute the absolute difference value,
dv
g
, such that
ev remainder
g g g
dv
=
Where
ev
g
is a value, which is fetched sequentially from Bs with
bits of
ec
g
-length.
Step 4: Embed
dv
g
into the pixel p
c
(i) (here, we define p
s
(i) as the intensity
of the i-th pixel after embedding
ev
g
)by performing the following
Process:
Case I: p
c
(i) < T
Av. pc[I] ps[I] gain , g g if .
g Av gain ,
2
m
g if .
Av. ) ( p ) ( p gain , else
Av. (i) p (i) p gain , g g if .
. g m Av value, adaptable an gain ,
2
m
g if .
2
m
T (i) p
2
m
2.
. g 0 (i) p gain ,
2
m
(i) p if 1.
ev remainder
dv
l
dv
c s
c s ev remainder
dv l
l
dv
l
c
l
ev s
l
c
= >
= s
=
+ = >
= >
< <
+ = <
i i
3. If (T-
2
l
m
)
ev remainder c s c
g g i p i p gain T i p + = < s ) ( ) ( , ) (
Case II : p
c
(i)
T >
A Novel Steganography Method for a Secret True Image
69
1. If p
c
(i)
>(255-
2
u
m
) , gain p
s
(i)=(255-m
u
+1)+g
ev
2. If (T +
2
u
m
)< p
c
(i)
) 1
2
255 + s
u
m
. If g
dv
>
2
u
m
, gain Av=m
u
-g
dv
If g
remainder >
g
ev
,gain p
s
(i)=p
c
(i)+Av
Else, gain p
s
(i)=p
c
(i)-Av
. If g
dv
2
u
m
s ,
gain Av=g
dv
If g
remainder
>g
ev
,gain p
s
(i)=p(i)
c
-Av
Else, gain p
s
(i)=p
c
(i)+Av
3. If T
ev remainder c s
u
c
g g i p i p gain
m
T i p + = + < s ) ( ) ( ),
2
( ) (
Step 5: Hide the embedded-header in the separated file .
Step 6: End.
The header part of the embedded image sending in a separated file.
In the stage of extraction, every bits of embedded image is extracted from the stego-image depending on the
positions that stored using random number generated by seed key and two modulus numbers m
u
, m
l .
additionally
the receiver must know the embedding algorithm and then decompress the extracted bits. The receiver cannot
able to extract the embedded-image without know the embedded positions; therefore this method is more secure.
The next scheme describes the extracting algorithm:
Extracting Algorithm
Input: The stego-image, the seed key and two modulus numbers m
u
, m
l
.
Output: The embedded-image.
Step1: Extract the embedded header of image.
Step 1: Find a workable pixel p
s
(i) in stego-image s by using the
Pseudo-random number generator of seed Sk.
Step 2: Compute the embedded information as following
Case I: ps (i) <T:
l s remaimder
m mod (i) p g =
l
m
2 ec
Log g =
Case II: p
s
(i)
T >
:
u s remaimder
m mod (i) p g =
u
m
2 ec
Log g =
Step 3: Translate the
remainder
g
into the bits representation to recover the
embedded information, the bit-length for each
remainder
g
is
Determined by the computation of
ec
g
.
Step 4: Decompress the embedded information
- Apply inverse wavelet transform.
Step 4: Recover the embedded-header from the separated file.
Step 5: End.
IV. THE PERFORMANCE MEASUREMENTS
To measure the performance of the proposed method, two different image files with different sizes are
applied. Also, we are used true color images. The method emphases in selecting the image files that its sizes
larger as four times than the image files. In order to explain how the proposed method have high-capacity
A Novel Steganography Method for a Secret True Image
70
feature. We are used subject and object criteria as measures of stego-image and restored message quality. In
object criteria, the following formulas are used:
i) Root mean square error (RMSE
*
).
(2) ,
N) (M
1 M
0 r
1 N
0 c
2
c) I(r, c) (r, I
RMSE
=
(
(
.
=
ii) Peak signal to noise ratio (PSNR
*
): Here we are used two different formulas, one for images as
following:
( )
(3) ,
1 M
0 r
1 N
0 c
2
c) I(r, c) (r, I
N M
1
2
1 L
10
log 10 PSNR
=
.
=
(
(
Where L represents the number of gray levels
.
I
is the stego-image, and I is cover-image with size (MxN).
Whereas, for measuring the quality of restored image, the following PSNR
**
formula is used:
| |
(4) ,
n
2
y(n) x(n)
n
(n)
2
x
10
10log PSNR
=
Where x (n) represent sample of embedded image sequence and y(n) stand for sample of restored image
sequence.
V. SATISFYING THE REQUIREMENTS OF CAPACITY, AND UNDETECTABILITY
In these experiments we attempt to embed an image message its size larger than the cover image. Also
the modulus values m
u
, ml are set of 32 and 16 respectively. In a sense the modulus 32 is set to accommodate
the embedding of 5-bit pattern at the situation that the pixel of embedded image is greater than the threshold
value T picked in our algorithm is empirically set the intensity value of 160.On the other hand, the modulus 16
is set to accommodate the embedding of 4-bit pattern at the condition that the pixel of embedded image is less
than T. We have been tested the proposed method on a number of true color images with different size, to
embed an image message its size as four times. The embedding capacity approximate restored message are
given in table (1).
It can be seen from table 1 that the proposed algorithm with generally higher PSNR values for restored
message. This indicates that the restored file is semi to the origin. Here we dont substitute the similar blocks in
order to make the stego-image without distortion.
Also, the table 2 shows the results of RMSE and PSNR between the cover-images and stego-images.In such
case, the embedding message is totally incorporated into cover-image. This indicates that all images less
distortion and the attacker (steganalysis) cant easily detect the embedded message. This makes the mission of
the attacker is more difficult. The RMSE and PSNR values for stego-images its acceptable and the distortions
are imperceptibility to human vision. It means such distortions will be less noticeable from the viewpoint of
attacker.
Table 1: The results of RMSE and PSNR for the extract image massage.
A Novel Steganography Method for a Secret True Image
71
Table 2: The results of RMSE and PSNR for the stego-images
VI. THE RESULT FOR THE IMPLEMENTATION OF THE PROPOSED SYSTEM
(a) (b)
Figure (3): image1 (a) cover-image, (b) stego-image
(a) (b)
Image2 (a) embedded-image, (b) extract -image
(a) (b)
Figure (4): image3 (a) cover-image, (b) stego-image
(a) (b)
Image4 (a) embeded-image, (b) extract -image
Image
RMSE
(image)
PSNR
(image)
Image1(b) 0.01 95.504
Image3(b) 0.012 87.971
Image5(b) 0.013 85.884
Image7(b) 0.014 80.493
A Novel Steganography Method for a Secret True Image
72
(a) (b)
Figure (5): image5 (a) cover-image, (b) stego-image
(a) (b)
Image6 (a) embedded-image, (b) extract -image
(a) (b)
Figure (6): image7 (a) cover-image, (b) stego-image
(a) (b)
Image8 (a) embedded-image, (b) extract image
VII. CONCLUSIONS
This paper emphases the most application requirements of any success steganography system like
(capacity, invisibility, security, and undetectability and advanced the relation of one to one.
The process of compression gives as the ability to hide data its size larger than the cover data. Thus, we can hide
a message its size larger than the size of cover-media and make the size of message less limited in some size. At
this point, here is the powerful of the proposed method by increasing the capacity of cover-media and reducing
the limits of size that facing most data hiding techniques. We find from experiments that the increased size of
message have no much effects on the quality of stego-image and still imperceptible. Additionally, the
compression process add another layer of protection for secrete message.
Also, after the embedding stage is completed, the stego-image less distortion or distortion-free depending on the
values of RMSE and PSNR as we see in tables (1, 2), respectively. This is because the change in the values of
data of cover-image is not found or not perceptual and the quality of stego-image still good. These results
A Novel Steganography Method for a Secret True Image
73
reflected the ability of human visual system. This point will reflect on the steganalysis, and makes the process of
analyzing the produced steganography very difficult.
For the security requirement, the algorithm produces a sequence of secret key that send independently to
increase the difficulty of steganalysis on these stego-images. The receiver can extract the embedded message by
using that secret key only.
REFERENCES
[1]. X. Luo, F. Liu and P. Lu. A LSB Steganography Approach Against Pixels Sample Pairs Stegonalysis.
International Journal of Innovative Computing, Information and Control, Vol. 3, No. 3, pp. 575-588,
June 2007.
[2]. William Stallings, Cryptography and Network Security, Principles and Practice, Third edition, Pearson
Education, Singapore, 2003.
[3]. M. Vetterli and J. Kovacevic. Wavelets and Subband Coding. Prentice Hall, 1995.
[4]. M. A. B. Younis and A. Janta. A New Steganography Approach for Image Encryption Exchange by
Using the Least Significant Bit Insertion. International Journal of computer science and network
security, vol.8, no.6, June 2008.
[5]. C. C. Lin, and W. H. Tsai, "Secret Image Sharing with Steganography and Authentication," Journal of
Systems and Software, 73(3):405-414, December 2004.
[6]. Kafa Rabah. Steganography - The Art of Hiding Data. Information technology Journal 3 (3) -2004.
[7]. [2] S. Lyu and H. Farid. Steganalysis Using Higher-Order Image Statistics. IEEE Transactions on
Information Forensics and Security, Vol. 1, No. 1, pp. 111-119, March, 2006.
[8]. B. K. Alsberg, A. M. Woodward, and D. B. Kell, An introduction to wavelet transforms for
chemometricians: a time-frequency approach, Chemometrics and Intelligent Laboratory Systems,Vol
37, No. 2, pp. 215-239, 1997.
[9]. R. C. Gonzalez, and R. E. Woods, Digital Image Processing, 2nd Edition, Prentice-Hall, Inc., 2002.
[10]. A. N. Akansu and M. J. Y Smith. Subband and Wavelet Transforms: Design and Applications Kluwer
Academic Publishers, 1996.
[11]. M. K. Bashar, T. Matsumoto, and N. Ohnishi, Wavelet transform-based locally orderless images for
texture segmentation, Pattern Recognition Letter, Vol. 24, No. 15, pp. 2633-2650, 2003.
[12]. H. G. Hwang, H. J. Choi, B. D. Kang, H. K. Yoon, H. C. Kim, S. K. Kim, and H. K. Choi,
Classification of breast tissue images based on wavelet transforms using
[13]. discriminant analysis, neural network and SVM, Proceedings of the 7th International Workshop on
Enterprise Networking and Computing in Healthcare Industry, Busan
[14]. , Korea, pp. 345349, 2005.
[15]. S.Wang and K.Yang, a Scheme of High Capacity Embedding On Image Data
[16]. using Modulo Mechanism, WISA, Souel, Korea, (2001).
[17]. Y.K.Lee and L.H.Chen, high Capacity Image Sreganographic Model, IEE
[18]. proceeding Vision, Image and Signal Proceeding, 147, 3, 288,(2000) .