Digital Signal Processing Reference
In-Depth Information
Fig. 4.29
LSB steganographic method
8
<
w
i
;
Q
if mod .
Q
w
i
;2/
D
c
j
D
e
w
i
w
i
Q
C
1; if mod .
Q
w
i
;2/
¤
c
j
&mod.
Q
w
i
;2/
D
1
(4.13)
:
w
i
Q
1; if mod .
Q
w
i
;2/
¤
c
j
&mod.
Q
w
i
;2/
D
0
where e
w
i
w
i
is the original nonzero quantized
wavelet coefficient, mod (,) is the module operation, and c
j
;j
D
1;:::;Lis the j th
bit in the coarse information bit stream.
The information extracting is simple, which can be obtained by
is the embedded wavelet coefficient,
Q
c
j
D
mod.e
w
i
;2/:
(4.14)
4.4.3
Proposed Two-Description Image Coding Scheme
The proposed two-description image coding scheme based on wavelet domain
partition is sketched in Fig.
4.30
, including encoding and decoding processes. A
similar coding structure can be obtained if spatial partitioning is considered.
4.4.3.1
Information Embedding
The coarse coefficients (coarse information) are embedded into the nonzero quan-
tized wavelet coefficients selected from high frequency to the low frequency for
minimizing the degradation of the fine coefficients with the LSB steganographic
method introduced in Sect.
4.4.2.2
, if L is smaller than the number of the nonzero
quantized wavelet coefficients. There are two points which need to be considered in
the embedding process.
1.
Embedding Capacity
: The embedding capacity depends on the number of
nonzero quantized wavelet coefficients. When the coarse information amount is
larger than the embedded capacity, a combination of embedding and appending
the coarse information can be employed. That means some coarse information is
embedded in the fine information, while the rest will be appended after the fine
information.
