Cryptography Reference
In-Depth Information
locations for information embedding
(PSD)
locations for adjusting complexity
(PAC)
Fig. 8.7.
Information assignment.
Approach to Adjust Complexity
In order to adjust the complexity of blocks in binary images, we use only
half of the pixels in the image. The remaining half of the pixels are used for
adjusting complexity. Pixels of the images are divided into two groups,
group
A
and
group B
. Pixels in
group A
are used for secret information and pixels
in
group B
are used for adjusting complexity. The pixels of group A are the
pixels for secret data, and are termed
PSD
. The pixels of group B are the
pixels for adjusting complexity, and are termed
PAC
. The locations of the
PSD and PAC correspond to a checkerboard pattern. Fig. 8.7 shows the bit
assignment of PSD and PAC in the images.
In the proposed method, we redefine noisy patterns as
0.5−δ≤α(P )≤0.5+δ.
(8.8)
where P is a binary image and δ is a constant coe
cient satisfying 0 <δ≤0.5.
Next, to generate the binary patterns for embedding, we divide a bit-
plane of the cover image into mm size blocks. Let P
i
(i =1, 2,,N)be
the noise-like blocks and C
i
(= α(P
i
)) be the complexity of P
i
. The com-
plexity histogram of blocks that are replaced with secret data is denoted by
h
ORG
(c), (0.5−δ≤c≤0.5+δ). h
ORG
(c) is the number of blocks of com-
plexity c. In order to represent the complexity histogram of blocks that have
been replaced with secret data, we define the another histogram denoted by
h
EMB
(c), (0.5−δ≤k≤0.5+δ), and initialize h
EMB
(c) to 0.
Firstly, we map a binary sequence having the size of (mm)/2 extracted
from secret data on PSD in P
i
. Then, we set the target complexity of P
i
by
the following steps.
a-1) C
org
and e are initialized as C
i
and 0 respectively.
a-2) Let k
c
and k
s
be C
org
+ e and C
org
−e.Thek
c
is the target complexity
if the following condition is satisfied,
h
ORG
(k
c
) >h
EMB
(k
c
)andk
c
≤0.5+δ. (8.9)
The k
s
is also the target complexity if the following condition is satisfied,
h
ORG
(k
s
) >h
EMB
(k
s
)
and
0.5−δ≤k
s
.
(8.10)
