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
-
Upload
william-crawford -
Category
Documents
-
view
217 -
download
0
Transcript of Damageless Information Hiding Technique using Neural Network Keio University Graduate School of...
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
Outline
Background Motivation Current Problem Proposed Method Experiment results Future Work Coclusion
Background
Emergence of the InternetContents are widely distributed
Information hiding provides reliabilityDigital watermarking for Digital Rights Manag
ementSteganography for covert channel
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
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
Proposed Method
ApproachEmbed model to pattern recognition model
Neural network as classifier (extraction key)
Only proper extraction key will lead to proper hidden signal
Why use neural network?
Has abilities ofTolerance to noiseError correction and complementationAdditional learning characteristic
Multi-layered Perceptron ModelBackpropagation Learning (Supervised Learni
ng)
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)
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)
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
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
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)
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
Backpropagation learning
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
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
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
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
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
Further experiments
Can proposed method extract from high pass filtered image or jpeg image
？
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
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
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
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
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
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
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
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
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
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
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
Conclusion
Information hiding technique without embedding any data into target content by using neural network
Ability to collaborate with other information hiding algorithm
Thank you
Appendix
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
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
Sigmoid function Input function for multi-layered perceptron model sigmoid = look like letter of S
xy
exp1
1
x y
Selection of feature values
８
８
Feature subblock
Has DC value and various values of AC (low, middle, high)
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
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
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
