Cryptography Reference
In-Depth Information
bit, and to tag watermark bit to the end of the shifted index. That is,
i W =(i<< 1) + W k ,
k∈[0,M W N W −1] .
(12.5)
We next make use of the MDSQ algorithms in [27] for index assignment.
The index assignments in Fig. 12.5, i 1 = l 1 (i W )andi 2 = l 2 (i W ), map the
quantizer output index to the two descriptions i 1 and i 2 .
Referring to the middle part of Fig. 12.5, i 1 and i 2 are transmitted over two
memoryless and mutually independent channels, or the lossy packet networks,
with erasure probabilities p 1 for Channel 1, and for p 2 Channel 2, respectively.
Finally, referring to the right side of Fig. 12.5, both of the transmitted
indices need to be reconstructed, and the embedded watermark needs to be
extracted. At the MD decoder, it first shifts received binary indices i
1 and i
2 to
the right by one bit to smooth away the effects from watermark embedding. It
determines the outcome i from the received indices with the MDSQ decoder.
Next, it does a table look-up process on the determined i using the codebook
C to obtain c
k . After gathering
i and then to obtain the reconstructed block X
all the blocks, the reconstruction image X
is obtain.
MD Encoder with
Watermark Embedding
MD Decoder with
Watermark Extraction
index
assignment
p 1
C
i 1
i
i 1
Channel
1
MD
Decoder
X
l 1 (•)
VQ −1
￿
i W
i Embed
￿
X
VQ
￿
￿
i 2
i 2
C
W
Channel
2
p 2
Gather
last bits
W
l 2 (•)
Fig. 12.5. The structure for embedding one watermark with two descriptions for
transmission in MDC. The two independent channels have mutually independent
breakdown probabilities.
12.4.2 The Single Watermark Extraction Algorithm
In watermark extraction, we do the estimation criterion from received in-
dices for determining the value of the watermark bits. With MDC, if both
Search WWH ::




Custom Search