下載/瀏覽

Download Report

Transcript 下載/瀏覽

多媒體網路安全實驗室
Lossless Data Hiding for
Color Images
Source:Eighth ISDA ’08 8th International Conference on Intelligent Systems
Design and Applications, 2008.
Authors:Ching-Hui Huang, Chih-Chiang Lee ,
Hsien-ChuWu, and Chung-Ming Wang
Speaker:Cheng Hsu
Date:2010/12/29
多媒體網路安全實驗室
Outline
1. Introduction
2. Related Works
3. The Proposed Method
4. Experimental Results
5. Conclusions
2
多媒體網路安全實驗室
1. Introduction
In this paper, a novel lossless data hiding
scheme is presented to exhibit high datahiding capacity for color palette images.
In the proposed method, each index value
is predicted by using the average of its
four precedent neighbors.
3
多媒體網路安全實驗室
Traditional
Lossless data
datahiding
hidingscheme
scheme
4
多媒體網路安全實驗室
2. Related Works
Tian’s method Embedding phase
( Pi ,Qi )  (200,195)
 Pi  Qi   200  195   395 
A


 197



2
  2 
 2  
d i  Pi  Qi  200  195  5
secret data b  1
d i '  2  d i  b  2  5  1  11
the new piexls ( Pi ' , Qi ' )
 d i '1
11  1
Pi'  A  
 197  
 197  6  203


 2 
 2 
 di ' 
11
Qi '  A     197     197  5  192
2
2
5
多媒體網路安全實驗室
Extracting phase
( Pi ' , Qi ' )  (203,192)
 Pi 'Qi '   203  192   395 
A'  


 197



2
  2 
 2  
d i '  Pi 'Qi '  203  192  11
b  LSB(d i ' )  LSB(11)  1
 d i '  11
di        5
 2  2
 d i  1
 5  1
Pi  A' 
 197  
 197  3  200


 2 
 2 
 di 
5
Qi  A'    113     197  2  195
2
2
6
多媒體網路安全實驗室
3. The Proposed Method
Diagram of data embedding process
多媒體網路安全實驗室
(a) Palette C
(b) Index table I
Color palette image components
8
多媒體網路安全實驗室
3.1 Define the embedding parameter eb
1 4

Step1 : average value xi    xi ,k 
 4 k 1 
difference d i  xi  xi
Step2 : d i '  2  d i  1
predictive template of index
if d i ' may overflow and eb  d i , eb  d i
9
多媒體網路安全實驗室
3.2 Data embedding procedure
10
多媒體網路安全實驗室
3.3 Data extracting procedure
1 4 ' 
Step1 : average value xi '    xi ,k 
 4 k 1 
difference d i '  xi ' xi '
Step 2 :
11
多媒體網路安全實驗室
Step3 : Generate the embedded index tabl e I by all xi and the original
cover image G is obtained by the same palette P and index tabl e I
Step4 : All secret data s is collected by s  s0 s1sB-1
12
多媒體網路安全實驗室
4.Experimental Results
Test image and the embedded
The embeddable positions of test image
多媒體網路安全實驗室
Performance comparisons among lossless data embedding methods on test
palette images
14
多媒體網路安全實驗室
5. Conclusions
In this paper, they have presented a novel
lossless data hiding scheme capable of
providing a high payload for color palette
images.
The difference expansion method has
been applied to embed secret data into
differences and to keep high stego-image
quality after hiding data.
15
多媒體網路安全實驗室