Diffusion-Based Image Compression in Steganography Paper by Markus - - PowerPoint PPT Presentation

▶
diffusion based image compression in steganography
SMART_READER_LITE
LIVE PREVIEW

Diffusion-Based Image Compression in Steganography Paper by Markus - - PowerPoint PPT Presentation

Aug 30, 2023 133 likes •524 views

Diffusion-Based Image Compression in Steganography Paper by Markus Mainberger, Christian Schmaltz, Matthias Berg, Joachim Weickert, Michael Backes Frank Nedwed 11.12.2012 Contents Explanations Explanations Goals/Motivation Goals

slide-1
SLIDE 1

Diffusion-Based Image Compression in Steganography

Paper by Markus Mainberger, Christian Schmaltz, Matthias Berg, Joachim Weickert, Michael Backes

Frank Nedwed – 11.12.2012

slide-2
SLIDE 2

Contents

  • Explanations
  • Goals/Motivation
  • The algorithm
  • Examples
  • Evaluation
  • Outlook

2

Explanations Goals Algorithm Examples Evaluation Outlook

slide-3
SLIDE 3

What is Steganography?

“the practice of concealing messages or information within other non-secret text

  • r data.”

Here: hiding an image in another one (and getting it out again)

3

Explanations Goals Algorithm Examples Evaluation Outlook

slide-4
SLIDE 4

What is Diffusion?

“the spreading of something more widely” Here: Simulating natural diffusion by partial differential equations

4

Explanations Goals Algorithm Examples Evaluation Outlook

slide-5
SLIDE 5

What is Diffusion?

“the spreading of something more widely” Here: Simulating natural diffusion by partial differential equations

5

Explanations Goals Algorithm Examples Evaluation Outlook

slide-6
SLIDE 6

What do we want to achieve?

Hiding an image (or part of an image) in itself or another picture

  • Without detectable traces
  • In (almost) real-time
  • Without significant loss of quality
  • Retrieving the data with a password

6

Explanations Goals Algorithm Examples Evaluation Outlook

slide-7
SLIDE 7

The algorithm (idea)

  • 1. Choosing pixels as Dirichlet boundary
  • 2. Storing them efficiently
  • 3. Encrypting
  • 4. Embedding them in the cover
  • 5. Recovering the secret
  • 6. Restoring the image with diffusion

7

Explanations Goals Algorithm Examples Evaluation Outlook

slide-8
SLIDE 8

1) Choosing the right pixels

  • Start with a rectangle defined by the

boundaries of the image/censored part

  • Divide it recursively into smaller

rectangles

  • Save characteristic points of the

resulting rectangles

8

Explanations Goals Algorithm Examples Evaluation Outlook

slide-9
SLIDE 9

1) Choosing the right pixels

  • The resulting rectangles are compared

with respect to their Laplace magnitude such that areas with higher contrast will be sampled more accurately

  • the four corner pixels and the middle

are saved

i

f 

9

Explanations Goals Algorithm Examples Evaluation Outlook

slide-10
SLIDE 10

1) Choosing the right pixels

  • The splitting process:

10

Explanations Goals Algorithm Examples Evaluation Outlook

slide-11
SLIDE 11

1) Choosing the right pixels

  • The splitting process:

11

Explanations Goals Algorithm Examples Evaluation Outlook

slide-12
SLIDE 12

1) Choosing the right pixels

  • The splitting process:

12

Explanations Goals Algorithm Examples Evaluation Outlook

slide-13
SLIDE 13

1) Choosing the right pixels

  • The splitting process:

13

Explanations Goals Algorithm Examples Evaluation Outlook

slide-14
SLIDE 14

1) Choosing the right pixels

  • The splitting process:

14

Explanations Goals Algorithm Examples Evaluation Outlook

slide-15
SLIDE 15

1) Choosing the right pixels

  • The splitting process:

15

Explanations Goals Algorithm Examples Evaluation Outlook

slide-16
SLIDE 16

1) Choosing the right pixels

  • The splitting process:

16

Explanations Goals Algorithm Examples Evaluation Outlook

slide-17
SLIDE 17

2) Representation

  • The information of the splitting process

is stored as a binary tree

  • The channels of the corresponding

pixels are quantised to 32 values

  • combined to a bitstream starting with

its length

17

Explanations Goals Algorithm Examples Evaluation Outlook

slide-18
SLIDE 18

3) Encryption

For specialists: The bitstream is encrypted by Advanced Encryption Standard (AES) in Cipher-block Chaining mode (CBC)

18

Explanations Goals Algorithm Examples Evaluation Outlook

slide-19
SLIDE 19

4) Hiding the information

Encrypted bitstream 11bit binary blocks 7bit ternary blocks Using our password, we get a pseudo- random permutation On those positions we hide the ternary bits using mod-3 matching

19

Explanations Goals Algorithm Examples Evaluation Outlook

slide-20
SLIDE 20

4) Hiding the information

Mod-3 matching: Value v, ternary bit t: = 1 v := v - 1 v – t mod 3 = 2 v := v + 1 = 0 v := v Special cases for v = {0, 1, 254, 255}

20

Explanations Goals Algorithm Examples Evaluation Outlook

slide-21
SLIDE 21

Special case: Censoring

  • Needs additional information about

censored area

  • Boundary information improves

reconstruction

  • Characteristic pixels on boundaries are

not saved

21

Explanations Goals Algorithm Examples Evaluation Outlook

slide-22
SLIDE 22

Recovering the image

The recovered secret serves as Dirichlet boundary data for the PDE: has the two eigenvalues is a contrast parameter

   

u u D div u

t

   

D

1

1 

2 2 2

1 1  

u   

 

22

Explanations Goals Algorithm Examples Evaluation Outlook

slide-23
SLIDE 23

Examples/Evaluation

400x400 grayscale Cover 20kb 768x584 RGB Secret 140kb

23

Explanations Goals Algorithm Examples Evaluation Outlook

slide-24
SLIDE 24

Diffusion process

24

Explanations Goals Algorithm Examples Evaluation Outlook

slide-25
SLIDE 25

Examples

160x160 RGB Cover 10kb 192x146 RGB Secret 10kb Low quality

25

Explanations Goals Algorithm Examples Evaluation Outlook

slide-26
SLIDE 26

Examples

26

160x160 RGB Cover 10kb 192x146 RGB Secret 10kb High quality

Explanations Goals Algorithm Examples Evaluation Outlook

slide-27
SLIDE 27

Censoring

Censoring: (Original)

27

Explanations Goals Algorithm Examples Evaluation Outlook

slide-28
SLIDE 28

Censoring

Censoring: (censored cover)

28

Explanations Goals Algorithm Examples Evaluation Outlook

slide-29
SLIDE 29

Censoring

Censoring: (Reconstruction)

29

Explanations Goals Algorithm Examples Evaluation Outlook

slide-30
SLIDE 30

Censoring

768x584 RGB Secret

30

Explanations Goals Algorithm Examples Evaluation Outlook

slide-31
SLIDE 31

Comparison

31

Explanations Goals Algorithm Examples Evaluation Outlook

slide-32
SLIDE 32

Comparison

32

Explanations Goals Algorithm Examples Evaluation Outlook

slide-33
SLIDE 33

Evaluation

Fig.: Compression ratio vs. MSE MSE :=

33

 

2 1 1 , ,

1  

 

 

M m N n n m n m

u f N M

Explanations Goals Algorithm Examples Evaluation Outlook

slide-34
SLIDE 34

Evaluation

  • Combines many state of the art

techniques

  • Protected against detection/recovering
  • (almost) real-time
  • High quality of the results
  • Vulnerable to changes of the cover

34

Explanations Goals Algorithm Examples Evaluation Outlook

slide-35
SLIDE 35

Outlook

  • Faster algorithms/implementations

could further speed up this method

  • Better choice of important pixels could

improve the quality

  • Extension of this method to image

sequences, etc.

35

Explanations Goals Algorithm Examples Evaluation Outlook

slide-36
SLIDE 36

Quiz

Secret (10kb) Mask Cover (30kb) Recovered secret 100 x 100 grayscale Cover

36

Explanations Goals Algorithm Examples Evaluation Outlook

slide-37
SLIDE 37

References

37

  • Try it out for yourself!

http://stego.mia.uni-saarland.de/

  • Mainberger, M., Schmaltz, C., Berg, M., Weickert, J.,

Backes, M.: Diffusion-Based Image Compression in Steganography

  • Schmaltz, C., Weickert, J., Bruhn, A.: Beating the

Quality of JPEG 2000 with Anisotropic Diffusion

  • Oxford dictionary
  • Image Processing and Computer Vision – Lecture

Notes 2011, J. Weickert

  • An Implementation of Key-Based Digital Signal

Steganography: Lecture Notes in Computer Science Volume 2137, 2001, pp 13-26

  • http://compressionratings.com
  • http://www.aishack.in

Explanations Goals Algorithm Examples Evaluation Outlook