International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
4. EXPERIMENTS
In this paper, a new method based on DCT has been proposed
in order to achieve hidden communication over the
communication networks. The experiments conducted on the
images given in Figures 1-4. The image given in Fig.1 has been
used as cover image and all of the other images have been
embedded into the cover image. The success of the hiding
process has been evaluated both objectively and subjectively.
The well known MSE value has been used as objective
evaluation measure for hiding performance. And the given
images have been used as objective evaluation measure of the
hiding performance of the proposed method.
The information hiding process has been given below step-by-
step:
l. Compute the frequency domain coefficients of the
cover image by using the DCT with for 8x8 pixels
sized unoverlapping blocks.
2. Use the numerical values of lower sub-triangle of the
DCT blocks in order to hide message information(s).
3. Recode the message information(s) by using the
appropriate tools such as,
a. Rescaling intensity coding,
b. Sampling intensity values,
c. Using block-coding techniques (i.e., run-
length coding).
4. Use one of the computationally secure cryptographic
methods (i.e., position permuting, color shifting, and
vigenerre-image coding) for images in order to
increase the security level of the message.
5. Replace the numerical values, which have been
determined at Step-3, with the values of the
information coefficients, which are still in spatial
domain.
6. Send to stego-object to audiences over the
communication net.
The proposed method has some advantages and disadvantages
that are listed below,
1. The proposed method is a computationally secure;
therefore, the minimum size of the cover image
determines the level of the security of message
information.
2. The proposed method is NOT an information lossless
coding technique until audiences uses frequency
domain.
3. The information hiding capacity of the proposed
method depends on to the methods that are used in
Step-3 of the proposed method and inherit
specifications of the message information(s) images.
4. The proposed method is not based on the bit-change
method; therefore it can produce artificial impulses if
stego-message converted into spatial domain.
In the experiments, The Fig.2, Fig.3, and Fig.4 message images
have been embedded in to first, second, and third band of the
message image, respectively. The stego-message image, Fig.5,
has been converted into spatial domain for visualization.
328
Figure 2: The elevation image of the Earth (2400x1200 pixels
sized).
Figure 3: The lights at the night over the Earth image.
(2400x1200 pixels sized).
4.1 Image Quality Measures
The well-known image quality measures; Mean Squared Error
(MSE), Pearson Correlation Coefficient, and Peak Signal To
Ratio (PSNR) have been used in order to objectively evaluate
the performance of the proposed method. The quality measure
of PSNR is defined with;
A
z
p
PSNR=10 lo — Max — IdB 9)
E10 T MSE
where /max is equal to 255 for 8 bit gray scale images. The
MSE is calculated by using the Eq. (2) given below:
MN
] 2
MSE= MN > > ( i78) (10)
i=l j=l
Interna
erre
Where
horizot
represe
the p
Correl:
Corr
where
Figure
Figur
Table
Stego
Provi
becar
comn
infori