Upload
heavynne-mays
View
48
Download
1
Embed Size (px)
DESCRIPTION
A Neural-Network Approach for Visual Cryptography. 虞台文. 大同大學資工所. Content. Overview The Q’tron NN Model The Q’tron NN Approach for Visual Cryptography Visual Authorization Semipublic Encryption General Access Scheme Conclusion. A Neural-Network Approach for Visual Cryptography. - PowerPoint PPT Presentation
Citation preview
A Neural-Network Approach for Visual Cryptography
虞台文大同大學資工所
Content
Overview The Q’tron NN Model The Q’tron NN Approach for
– Visual Cryptography– Visual Authorization– Semipublic Encryption
General Access Scheme Conclusion
A Neural-Network Approach for Visual Cryptography
Overview
大同大學資工所
What isVisual Cryptography and Authorization?
Visual Cryptography (VC)– Encrypts secrete into a set of images
(shares).– Decrypts secrete using eyes.
Visual Authorization (VA)– An application of visual cryptography.– Assign different access rights to users.– Authorizing using eyes.
What is Semipublic Encryption?
Visual Cryptography (VC)– Encrypts secrete into a set of images
(shares).– Decrypts secrete using eyes.
Semipublic Encryption (SE)– An application of visual cryptography.– Hide only secret parts in documents – Right information is available if and only if a
right key is provided
The Basic Concept of VC
Target Image(The Secret)
Share 2
Share 1AccessScheme
AccessScheme
The (2, 2) access scheme.
The Shares Produced by NN
Target Image(The Secret)
Share 2
Share 1Neural
Network
NeuralNetwork
We get shares after the NN settles down.
Decrypting Using Eyes
Share 2
Share 1
Example: (2, 2)
Target image
Share image2
Share image1
Plane shares are used
Traditional Approach
Naor and Shamir (2,2)
Pixel ProbabilityShares
#1 #2Superposition ofthe two shares
5.0p
5.0p
5.0p
5.0p
WhitePixels
BlackPixels
The Code Book
The VA Scheme
keyshare
user shares(resource 2)
user shares(resource 1)
stacking
stacking
…
…VIP IP P
…VIP IP P
Very Important
Person.
…
The SE Scheme
智慧型系統實驗室資料庫
使用者 Key江素貞 AB陳美靜 CD張循鋰 XY李作中 UV
智慧型系統實驗室資料庫
使用者 Key江素貞 AB陳美靜 CD張循鋰 XY李作中 UV
public share(database in lab)
AB CD XY UV
stacking
usershares
keys
素貞
The SE Scheme
循鋰 美靜 作中
智慧型系統實驗室資料庫
使用者 Key江素貞 AB陳美靜 CD張循鋰 XY李作中 UV
智慧型系統實驗室資料庫
使用者 Key江素貞 AB陳美靜 CD張循鋰 XY李作中 UV
A Neural-Network Approach for Visual Cryptography
The Q’tron
NN Model
大同大學資工所
The Q’tron
i
(ai )
i
(ai )
. . .
0 1 2 qi1
aiQiActive value
Qi{0, 1, …, qi1}IiR
External Stimulus
( )ij j jj
T a QInternal Stimulus
Ni
Noise
Quantum Neuron
The Q’tron
i
(ai )
i
(ai )
. . .
0 1 2 qi1
aiQiActive value
Qi{0, 1, …, qi1}IiR
External Stimulus
( )ij j jj
T a QInternal Stimulus
Ni
Noise
Free-Mode Q’tron
The Q’tron
i
(ai )
i
(ai )
. . .
0 1 2 qi1
aiQiActive value
Qi{0, 1, …, qi1}IiR
External Stimulus
( )ij j jj
T a QInternal Stimulus
Ni
Noise
Clamp-Mode Q’tron
Input Stimulus
InternalStimulus
ExternalStimulus
Noise
NoiseFreeTerm
i
(ai )
i
(ai )
. . .
Noise
Level Transition
Running AsynchronouslyRunning Asynchronously
i
(ai )
i
(ai )
. . .
Energy Function
InteractionAmong Q’trons
Interactionwith
External Stimuli
Constant
Monotonically NonincreasingMonotonically Nonincreasing
The Q’tron NN
Interface/Hidden Q’trons
clamp-modefree-modefree mode Hidden Q’trons
Interface Q’trons
Question-Answering
Feed a question by clamping some interface Q’trons.
clamp-modefree-modefree mode Hidden Q’trons
Interface Q’trons
Question-Answering
Read answer when all interface Q’trons settle down.
clamp-modefree-modefree mode Hidden Q’trons
Interface Q’trons
A Neural-Network Approach for Visual Cryptography
The Q’tron NNs for Visual Cryptography Visual Authorization Semipublic Encryption
大同大學資工所
Energy Function for VC
Visual Cryptography
Image Halftoning
Image Stacking
+
Image Halftoning
Graytone Image
Halftoning
0
255
Halftone Image
0 (Transparent)
1
Graytone image halftone image can be formulated as to minimize the energy function of a Q’tron NN.
Image Halftoning
Graytone Image
Halftoning
0
255
Halftone Image
0 (Transparent)
1
Graytone image halftone image can be formulated as to minimize the energy function of a Q’tron NN.
In ideal case, each pair of corresponding small areas has the `same’ average graylevel.
The Q’tron NN for Image Halftoning
Plane-G (Graytone image)
Plane-H (Halftone image)
Image Halftoning
Halftoning
Clamp-mode
Free-mode
Plane-G (Graytone image)
Plane-H (Halftone image)
Question
Answer
Image Restoration
Plane-G (Graytone image)
Plane-H (Halftone image)
Restoration
Clamp-mode
Free-mode
Question
Answer
Stacking Rule
+ + + +
The satisfaction of stacking rule can also be formulated as to minimize the energy function of a Q’tron NN.
Stacking Rule
+ + + +
The satisfaction of stacking rule can also be formulated as to minimize the energy function of a Q’tron NN.
The energy function for the stacking rule.
See the paper for the detail.
The Total Energy
+
Share 1 Target
Share 1
Share 2
TargetShare 2
TotalEnergy
Image Halftoning
Stacking Rule
The Q’tron NN for VC/VA
Plane-GS1
Plane-HS1
Share 1
Plane-HS2
Plane-GS2
Share 2
Plane-GT
Plane-HT
Target
Application Visual Cryptography
Plane-GS1
Plane-HS1
Share 1
Plane-HS2
Plane-GS2
Share 2
Plane-GT
Plane-HT
Target
Clamp-Mode
Clamp-Mode
Clamp-Mode
Free-Mode Free-Mode
Free-Mode
Application Visual Authorization
Plane-GS1
Plane-HS1
User Share
Authority
Plane-HS2
Plane-GS2
Plane-GT
Plane-HT
Key Share
Key Share
User Share
VIP IP P
Application Visual Authorization
Plane-GS1
Plane-HS1
User Share
Authority
Clamp-Mode
Free-Mode
Plane-HS2
Plane-GS2
Clamp-Mode
Free-Mode
Plane-GT
Plane-HT
Clamp-Mode
Free-Mode
Key Share
Key Share
User Share
VIP IP P
Producing key Share & the first user share.
Application Visual Authorization
Plane-GS1
Plane-HS1
User Share
Authority
Clamp-Mode
Plane-HS2
Plane-GS2
Clamp-Mode
Free-Mode
Plane-GT
Plane-HT
Clamp-Mode
Some are clamped and some are free.
Key Share
Key Share
User Share
VIP IP P
Producing other user shares.
Application Visual Authorization
Plane-GS1
Plane-HS1
User Share
Authority
Clamp-Mode
Plane-HS2
Plane-GS2
Clamp-Mode
Free-Mode
Plane-GT
Plane-HT
Clamp-Mode
Some are clamped and some are free.
Key Share
Key Share
User Share
VIP IP P
Producing other user shares.
Application Visual Authorization
Plane-GS1
Plane-HS1
User Share
Authority
Clamp-Mode
Plane-HS2
Plane-GS2
Clamp-Mode
Free-Mode
Plane-GT
Plane-HT
Clamp-Mode
Some are clamped and some are free.
Key Share
Key Share
User Share
VIP IP P
Key Share
User Shar
e
User Shar
e
User Shar
e
VIP
IP
P
A Neural-Network Approach for Visual Cryptography
General
Access Scheme
大同大學資工所
Full Access Scheme 3 Shares
朝辭白帝彩雲間朝辭白帝彩雲間
朝 辭 白
帝 彩 雲
間
Shares
Full Access Scheme 3 Shares
朝辭白帝彩雲間朝辭白帝彩雲間
朝 辭 白
帝 彩 雲
間
SharesTheoretically,
unrealizable.
We did it in
practical sense.
Theoretically,
unrealizable.
We did it in
practical sense.
Full Access Scheme 3 Shares
S1 S2 S3
S1+S2 S1+S3 S2+S3 S1+S2+S3
Access Schemewith Forbidden Subset(s)
Anyone knows what is it?Anyone knows what is it?
Access Schemewith Forbidden Subset(s)
人之初性本善人之初性本善
人 之 初
性 本 X
善
Theoretically,
realizable.Theoretically,
realizable.
Shares
Access Schemewith Forbidden Subset(s)
S1 S2 S3
S1+S2 S1+S3 S2+S3 S1+S2+S3
A Neural-Network Approach for Visual Cryptography
Conclusion
大同大學資工所
Conclusion
Different from traditional approaches:– No codebook needed.– Operating on gray images directly.
Complex access scheme capable.
http://www.suchen.idv.tw/
謝謝謝謝