Efficient Image Encoding and Information Embedding Techniques ???????????????????

1
Efficient Image Encoding and Information
Embedding Techniques ???????????????????

• Advisor Chang, Chin-Chen
• Student Chen, Chang-Chu
• Date June, 17, 2010
• Department of Computer Science and Information
  Engineering, National Chung Cheng University

2
Outline

• Part I VQ encoding technique
• Two-Bounds Triangle Inequality for VQ Codebook
  search
• Part II Information embedding technique
• High Capacity SMVQ-based Hiding Scheme Using
  Adaptive Index

3
Part I VQ encoding technique
4
Vector quantization

• VQ Block-based quantizer
• Applications
• Signal compression (i.e. Image, Speech, )
• Feature recognition
• Information security

5
VQ

• Overview

X
512
6
VQ encoding

• Related works
• Full-search equivalents
• Reduce the range of searching domain
• Decrease the online computational operations
• Partial-search methods
• Organize the codebook by some data structures to
  label a local search domain

7
Proposed scheme

• Triangle inequality
• Reduce the range of searching domain
• Decrease the online computational operations

F
A fixed vector
X
Ci
8
Experiments (1/2)
ms Images Schemes Lena Jet(F16) GoldHill
FS 4360 4400 4390
Mielikainens scheme 2680 2630 3050
Chang and Lins scheme 1 2250 2160 2590
Chang and Lins scheme 2 1900 2030 2200
Chang and Lins scheme 3 1790 1830 2010
Proposed scheme 1030 890 1130
31J. Mielikainen, A Novel Full-Search Vector
Quantization Algorithm Based on The Law of
Cosines, IEEE Signal Processing Letters, vol. 9,
no. 6, Jun. 2002, pp. 175-176. 10C. C. Chang
and I. C. Lin, Novel Full-Search Schemes for
Speeding Up Image Coding Using Vector
Quantization, Real-Time Imaging, vol. 10, no. 2,
Apr. 2004, pp. 95-102.
9
Experiments (2/2)
ms Codebook sizes Schemes 256 512 1024
FS 4390 8670 16270
Mielikainens scheme 3050 5750 10320
Chang and Lins scheme 1 2590 4910 9050
Chang and Lins scheme 2 2200 4040 7260
Chang and Lins scheme 3 2010 3680 6580
Proposed scheme 1130 2010 3840
10
Part II Information embedding technique

• Hiding in VQ and SMVQ index

11
Data hiding
Compressed codes 10111011 11..
11010110 01..
Information
(187)10
(214)10
Internet
Sender
Receiver
Information
12
Data hiding in VQ codes
(Jo and Kim 27)
18
18
46
46
19
27M. Jo and H.D. Kim, A digital image
watermarking scheme based on vector
quantisation, IEICE Transactions on Information
and Systems, vol.E85-D no.6, Jun. 2002,
pp.1054-1056.
13
Side Match VQ (SMVQ)

• Assumption Neighboring pixel intensities in an
  image are pretty similar.

14
State codebook
X (81, 15, 53, 34, 51,?, ?, ?, 91, ?, ?, ?,
49,?, ?, ?)
Codebook(512)
State codebook(8)
15
Data hiding in SMVQ and VQ codes
Secret message




Compressed code
Seedblock
Residualblock
SMVQEmbedding
VQ Embedding
gt THSMVQ
lt THSC



...

...





1
2
3
4
5
6

2n
Codebook
State codebook
16
Extracting data
Secret message




Compressed code
Indicator 1
Indicator 0
SMVQExtracting
VQ Extracting



...

...





1
2
3
4
5
6

2n
Codebook
State codebook
17
Experiments(1/3)
Methods Images Our method Our method Our method Jo and Kims method Jo and Kims method Jo and Kims method Chang and Wus method Chang and Wus method Chang and Wus method
Methods Images Secrets PSNR Bit Rate Secrets PSNR Bit Rate Secrets PSNR Bit Rate
Baboon 22950 23.43 0.177 12372 23.79 0.094 9450 23.76 0.058
Boat 27227 27.35 0.261 13889 27.79 0.106 13978 28.02 0.114
Jet(F16) 29386 28.08 0.305 14273 28.84 0.109 14444 28.93 0.129
Lena 33040 28.80 0.319 14898 29.54 0.114 15016 30.13 0.134
Pepper 30327 28.07 0.295 15036 28.71 0.115 14492 28.88 0.127
SailBoat 24710 26.86 0.236 13870 27.46 0.106 13787 27.58 0.106
27M. Jo and H.D. Kim, A digital image
watermarking scheme based on vector
quantisation, IEICE Transactions on Information
and Systems, vol.E85-D no.6, Jun. 2002,
pp.1054-1056. 12C. C. Chang and W. C. Wu A
steganographic method for hiding secret data
using side match vector quantization, IEICE
Transactions on Information and Systems,
vol.E88-D, no.9, Sep. 2005, pp.2159-2167.
18
Experiments(2/3)
Host images
Embedded images
19
Experiments(3/3)
The results of different parameters weight of
WSED, THSC , THSMVQ, and the size of the codebook
20
Future Research Directions

• Fast VQ codebook search
• Find other geometric models or other estimation
  equations to reduce the range of searching
  domain anddecrease the online computational
  operations
• Data hiding in the compression codes
• Investigate other data compression techniques and
  design the high capacity hiding schemes with
  reversibility property

21
Conclusions

• Part I
• Two-Bounds Triangle Inequality for VQ Codebook
  search
• Part II
• High Capacity SMVQ-based Hiding Scheme Using
  Adaptive Index

22
Thanks for your attention