Damageless Information Hiding Technique using Neural Network Keio University Graduate School of...

42
Damageless Information Hiding Te chnique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe

Transcript of Damageless Information Hiding Technique using Neural Network Keio University Graduate School of...

Page 1: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Damageless Information Hiding Technique using Neural Network

Keio University

Graduate School of Media and Governance

Kensuke Naoe

Page 2: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Abstract

An information hiding technique without embedding any data to target contentPattern recognition model

Neural network as classifier (extraction key)

Advantage and disadvantage

Page 3: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Outline

Background Motivation Current Problem Proposed Method Experiment results Future Work Coclusion

Page 4: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Background

Emergence of the InternetContents are widely distributed

Information hiding provides reliabilityDigital watermarking for Digital Rights Manag

ementSteganography for covert channel

Page 5: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Motivation and current problem

Use one information hiding algorithm with another to strengthen the security of the content Digital watermarking Steganography FIngerprinting

There are many great information hiding algorithm but have difficulties to collaborate possibility of obstructing previously embedded data Applying another information hiding algorithm might re

sult in recalculation of fingerprint for the content

Page 6: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Research Objective

To hide or to relate certain information without embedding any information to the target content

Ability to collaborate with another information hiding algorithm to strengthen the security

Page 7: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Proposed Method

ApproachEmbed model to pattern recognition model

Neural network as classifier (extraction key)

Only proper extraction key will lead to proper hidden signal

Page 8: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Why use neural network?

Has abilities ofTolerance to noiseError correction and complementationAdditional learning characteristic

Multi-layered Perceptron ModelBackpropagation Learning (Supervised Learni

ng)

Page 9: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Proposed Method (Embedding)1.Frequency Transformation of content

Hidden signal as teacher signal

2.Selection of feature subblock

3.Use feature values as input value for neural network

4. Generation of classifier (extraction key)

Coordinate of feature subblocks (extraction key)

Page 10: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Proposed Method (Extraction)1.Frequency Transformation of content

Hidden signal as output signal

2.Selection of feature subblock

3.Use feature values as input value for neural network

4. Applying the classifier (encryption key)

Coordinate of feature subblocks (encryption key)

Page 11: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

What is neural network?

neuron ( nervous cell ) It only has a function of receiving a signal and dispatc

hing signal to connected neuron When organically connected, it has ability to process

a complicated task

A network built with these neurons are called neural network Multi layered perceptron model

Often used for non-linear pattern classifier

Page 12: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Calculation of network

Input value of neuron Sum product of network weight and output values from

previous layer

jxj

yj

y1 yi yN

w1j wijwNj

N

iiijj ywx

1

Page 13: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Generating classifier (extraction key)

1.Frequency Transformation of content

Hidden signal as teacher signal

2.Selection of feature subblock

3.Use feature values as input value for neural network

4. Generation of classifier (encryption key)

Coordinate of feature subblocks (encryption key)

Page 14: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Patterns and signals to concealpatter

nhidden signal

pattern

hidden signal

pattern

hidden signal

1 00000 11 01010 21 10100

2 00001 12 01011 22 10101

3 00010 13 01100 23 10110

4 00011 14 01101 24 10111

5 00100 15 01110 25 11000

6 00101 16 01111 26 11001

7 00110 17 10000 27 11010

8 00111 18 10001 28 11011

9 01000 19 10010 29 11100

10 01001 20 10011 30 11101

        31 11110

        32 11111

Page 15: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Backpropagation learning

Page 16: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

input pattern

sign

al v

alue

Page 17: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 2

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

input pattern

sign

al v

alue

Page 18: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 3

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

input pattern

sign

al v

alue

Page 19: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 4

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

input pattern

sign

al v

alue

Page 20: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

input pattern

sign

al v

alue

Page 21: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Further experiments

Can proposed method extract from high pass filtered image or jpeg image

Page 22: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 1

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

sign

al v

alue

original highpass

Page 23: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 2

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132

pattern

sign

al v

alue

original highpass

Page 24: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 3

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

sign

al v

alue

original highpass

Page 25: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 4

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

sign

al v

alue

original highpass

Page 26: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 5

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

sign

al v

alue

original highpass

Page 27: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 1

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

outp

ut s

igna

l

original jpeg

Page 28: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 2

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

outp

ut s

igna

l

original jpeg

Page 29: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 3

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

outp

ut s

igna

l

original jpeg

Page 30: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 4

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

outp

ut s

igna

l

original jpeg

Page 31: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Network 5

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132pattern

sign

al v

alue

original jpeg

Page 32: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Future work

Because it relies on the position of feature sub block, it is weak to geometric attacksRotation, expansion, shrinking

Key sharing has to rely on another security technology

Page 33: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Conclusion

Information hiding technique without embedding any data into target content by using neural network

Ability to collaborate with other information hiding algorithm

Page 34: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Thank you

Page 35: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Appendix

Page 36: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Tradeoffs for information hiding

Watermarking

(Digital Right Management)

Steganography

(Covert Channel)

Fingerprinting

(Integrity check)

Capacity

(Amount of data to be embedded)

Not important

Small amount is enough

Important

More the better

Not Important

More the better

Robustness

(tolerance against attack to the container)

Important

Must not be destroyed

Not important

Content and hidden data are

not related

Important

Should be weak against alteration

Invisibility

(transparency of hidden data)

Important

Should not disturb the content

Important

Existence should be kept secret

Not Important

Existence can be informed

Page 37: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Three layered perceptron model

Three layer model Feed forward model

Input function Sigmoid function

Backpropagation learning x1 xi xM

i

j

k

jkw

ijw

Input layer

Hidden layer

Output layer

Page 38: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Sigmoid function Input function for multi-layered perceptron model sigmoid = look like letter of S

xy

exp1

1

x y

Page 39: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

Selection of feature values

Feature subblock

Has DC value and various values of AC (low, middle, high)

Page 40: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

number of hidden neuron=10 threshold=0.05

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

level of alteration (increase with step of 0.1)

perc

enta

ge

selected feature sub blocksother sub blocks

Page 41: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

number of hidden neuron=10 threshold=0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

level of alteration (increase with step of 0.1)

perc

enta

ge

selected feature sub blocksother sub blocks

Page 42: Damageless Information Hiding Technique using Neural Network Keio University Graduate School of Media and Governance Kensuke Naoe.

number of hidden neuron=20 threshold=0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

level of alteration (increase with step of 0.1)

perc

enta

ge

selected feature sub blocksother sub blocks