spie proceedings [spie spie defense, security, and sensing - baltimore, maryland, usa (monday 29...

Enhanced Information Security Employing Orthogonal Code, Steganography and Joint Transform Correlation

M. Nazrul Islam1, Muhammad Faysal Islam2 and Kamal Shahrabi3

1Department of Security Systems, State University of New York, Farmingdale, New York, USA; E-mail: [email protected]

2Department of Engineering Management and Systems Engineering, George Washington University, Washington, DC, USA; E-mail: [email protected]

3Department of Electrical and Computer Engineering Technology, State University of New York, Farmingdale, New York, USA; E-mail: [email protected]

ABSTRACT A novel and robust technique is proposed in this paper for securing confidential information by utilizing orthogonal coding scheme, encoded steganography and nonlinear encryption through joint transform correlation. Different biometric signatures are encoded using individual orthogonal codes and then multiplexed together. The encrypted and multiplexed image is hidden inside a cover image employing a steganography technique, where one from the three least significant bits is chosen using another secret key. A color cover image is utilized which is decomposed into three color components, red, green and blue, so that three different sets of biometric signatures can be embedded into each of the color components. The color stego image is finally encrypted using multiple phase-shifted reference joint transform correlation (MRJTC) technique. The proposed encryption technique is a nonlinear process which increases the security strength significantly against any unauthorized access. The encoded steganography technique reduces the vulnerability that an intruder can retrieve any information from a given image through any steganalysis attack. Finally, the orthogonal coding scheme enhances the robustness by making the biometric information almost inaccessible without authorization. Index Terms: biometric signature, color image processing, orthogonal coding, joint transform correlation, steganography.

INTRODUCTION Tremendous growth in digital technology over the years has also raised the concerns of security the confidentiality and integrity of information. Information security systems require robust techniques to prevent from any unauthorized access, and to verify the authorization without any error in order to give an access to information. Biometric information, including fingerprints, iris, hand geometry, palm print, gestures, have proven to be an efficient tool in establishing the identity of an individual because they are found to contain unique characteristics corresponding to the human being [1 – 4]. However, there are a number of challenges in biometrics-based security systems, including variation and distortion of biometrics with time, place, and environment, non-cooperation of individuals, and preservation of biometric information [5 – 8]. Steganography has been proposed in the literature to secure biometric signatures by hiding them inside a cover image through replacement of the least significant bit (LSB) of the image by the corresponding information bit [9, 10]. However, traditional steganography techniques are vulnerable to steganalysis attacks where an intruder can try to retrieve the secret information by monitoring the LSBs of an unknown image [11, 12]. Also the encryption techniques are typically linear

Optical Pattern Recognition XXIV, edited by David Casasent, Tien-Hsin Chao, Proc. of SPIE Vol. 8748, 87480K · © 2013 SPIE · CCC code: 0277-786X/13/$18 · doi: 10.1117/12.2016394

Proc. of SPIE Vol. 8748 87480K-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms

Informati

L

Orthogo

Enc

Mil il +inI,

.

En

ion 1

Y-nal Coding &

:oded Steganc

If*--

1 Phaca.Chi#a

Transform CI

1crypted Stegc

Information N

1

Multiplexing

:

)graphy

ri Rafaranra

Drrelation

1 Image

7

7Encrypted

Multiplexed Ii

Stego Image

nage

processesinformatio The objecemployingbiometric encryptednamely, rproposed the cover a multipleprocess [1coding, rasecurity ain the samcolor cove

The propoinformatioencryptedselection employing As shownand then superimpoimage fi(xcan be ob

, where thereon without kn

ctive of this pg orthogonal information

d information ed, green andsteganographimage pixels e phase-shifte16]. As the pandom bit repttacks. Additi

me cover imager image.

osed informaon are encryp

d and multipland replacem

g the MRJTC

n in Fig. 2, thmultiplied

osed on a comx, y), and ki(xtained as

e are always nowing the se

aper is to devcoding schemare encryptedis embedded

d blue, and eahy technique a

for informatied reference jproposed inforplacement andionally, the prge while mai

Figure 1: B

tion security pted first usiexed image iment purpose

C technique.

he input imagby the respemmon spatialx, y) is the re

chances of decret keys.

velop a novel me, encoded sd and then mud into a colorach color imaalso employs ion hiding purjoint transforrmation securd nonlinear enroposed technntaining negl

Block diagram

PERFORM

system is shoing individuais now embede. The stego

ges containinective orthogl domain. Asspective orth

decoding, com

and efficient steganographyultiplexed togr image, whicage is used toa second set rpose [14, 15rm correlationrity techniquencryption, it ynique can accligible amoun

m of the propo

MANCE AN

own in the blal orthogonal dded into the

image is fi

ng biometric igonal code. ssume that gi(ogonal code.

mpletely or a

approach fory and nonlinegether using och is decompo embed a seof keys to se]. Finally, then (MRJTC) te employs thyields a high ommodate a n

nt of distortio

osed security

NALYSIS

lock diagram codes and te cover imaginally encryp

information aThe individu

(x, y) is the e Thus the en

at least partly

r securing bioear encryptionorthogonal coposed into thrt of biometric

elect one frome stego imagetechnique, whhree-fold encr

level of robunumber of di

on when comp

system

m of Fig. 1. Tthen multiple

ge by using apted using an

are expandeddual encoded expanded formncrypted and m

y, the confid

ometric informn process. Muodes [13]. Theree color chac information

m the three LSe is encrypted hich is a nonryption, orthoustness againsfferent informpared with or

The input biomexed togethera secret key fnother set of

d in one dimeimages are

m of the i’th multiplexed i

dential

mation ultiple en the

annels, n. The SBs of

using nlinear ogonal st any

mation riginal

metric r. The for bit f keys

ension then input

image



Image 1

Image N

EncodedMultiplexI ni riIIuuy

- Expansio

HExpansio

Coy

&ed 1-I Repl

PseudISec

Code 1

n

Code N

n -06

er Image

Bit

lacement r

o- Random I

2 lane

;ret Key

En

Mu

Im

- ea_-._ I.OleyU II

coded &iltiplexedage

nage

In this papapplying t

The set ofthe numbe Next, a canother bired, greenencoded sfor all theusing a seare then re

per, a Walsh the Hadamard

f codes are geer of independ

olor cover imiometric inforn and blue, esteganographye three color ecret key to seeplaced by th

F

Fig

code is empld transform o

enerated in thedent codes.

mage is chosermation, like ach of whichy technique fochannels. Th

elect a bit fromhe correspond

Figure 2: Blo

gure 3: Block

( yxt ,

loyed to prodn 0 (zero) rep

1 =H

⎢⎢⎣

⎡=2n

H

HH

e form of a sq

en to hide thface image. T

h is used to hfor one color che steganogram the three Ling bits of the

ck diagram o

k diagram of

) (∑=

=N

ii yxky

1,

duce the orthopeatedly as de[ ]0=

⎥⎥⎦

⎤____

nn

nn

HH

HH

quare matrix,

he confidentiaThe color imahide one set ochannel. Systaphy process SBs of each pe encrypted im

f the orthogon

the encoded s

) ( )i yxgy ,

ogonal codes,escribed below

where the len

al biometric iage is first spof biometric tems shown inemploys a p

pixel. The selmage.

nal encoding

steganograph

, which can bw [17].

ngth of the m

information, plit into three

signatures. Fn Figs. 2 and

pseudo-randomlected bits fro

technique

hy technique

be implement

atrix is 2m and

which can alcolor compon

Figure 3 showd 3 will be repm plane gene

om the cover i

(1)

ted by

(2)

(3)

d m is

lso be nents,

ws the peated erated image



EncryptioKey - 1

- 2

)0° Phase La

80 °Phase

Shift

70° Phase

Shift

LT

i

T

Input Plaintext

Image

Fourier

ransform

Fourier

ransTOrm

Fourier

'ransform

Fourier

ransform

Joint Powe

Spectrum

Joint Powe

bpectrum

Joint Powe

Spectrum

Joint Powe

Spectrum

I I 90°PI--N SVJ

r - 180 °1

S1

270 °1

SI

CiphertextImage +

hase

tiff'1\('hase Hvlift

'hase

lift

Inverse

FourierTransform

k1

i, Figure 4 sstego imato be encprocessingd(x, y), is

Applying power spe

where u a1800, and

F

shows the bloage. An encrycrypted but hg channels afadded to the

Fourier tranectrum (JPS)

and v are the F2700, respect

Figure 4: Blo

ock diagram yption is formhaving randomfter phase shiphase-shifted

fsformation tosignals as giv

( )vuS ,1

+

=

( )vuS ,2

+

=

( )vuS ,3

−

=

( )vuS ,4

−

=

Fourier domatively. Then a

ock diagram o

of the nonlinmed in the sha

m pixel valuifting by 00, 9d keys in each

( ) cyxf ,1 =( ) jyxf ,2 =( )yxf ,3 −=

( ) jyxf ,4 −=o each joint iven by

( )( ) (DvuC

vuF

,

,*

21

+

==

( )( ) (DvujC

vuF

,

,*

22

+

==

( )( ) (uDvuC

vuF

,

,*

23

−

==

( )( ) (DvujC

vuF

,

,*

24

−

==

ain variables. a modified JP

of the MRJTC

near MRJTC-pe of an imag

ues. The encr900, 1800, andh channel, wh( ) (xdyxc , +( ) (xdyxjc , +( ) (xdyxc , +−

( ) (xdyxjc , +image, the m

( )) (uCvu

DvuC

,,

,*

2

+

+=

( )( ) (uCvu

DvuC

,,

,*

2

+

+=

( )) ( vuCvu

DvuC

,,

,*

2

−

+

( )( ) (ujCvu

DvuC

,

,*

2

+

+

The above JPS signal is de

C encryption

-based encrypge having theryption key, d 2700, respe

hich yields fou)yx, )yx, )yx,

)yx, magnitude spe

( )) ( )vuDv

vuD

,

, 2

( )) ( )vuDv

vuD

,,

, 2

( )) ( )vuDv

vuD

,

, 2

( )) ( )vuDvu

vuD

,,

, 2

PS signals areeveloped usin

technique

ption technique same dimenc(x, y), is fe

ectively. The ur joint image

ectra are reco

e again phaseng the followin

ue for securinnsions as the ied to four painput stego imes as given by

orded as four

e-shifted by 00

ng relation.

ng the image arallel mage, y (4) (5) (6)

(7) r joint

(8)

(9)

(10)

(11)

0, 900,



Ein

Steg

:rypted F

io Image Tra

Stegc

ImagE

Decrypti

1FOL

TII CIM

ourier

Insform

Cc

Pseuc

L_Se,

on Key

I

crier=Fringe

b-1 Filter

ever Image

tBit L

lapping I

TJo- Random I

ciane

Icret Key

1

Encoded

-

d FourieInversEi

Transfor

, inuRIr./IcAtImage

&

xl

Stego

m Image

Inverse Fo

where ⊗

Now an informatioencryptedinverse Fo

As can beof the infthe decrypfiltered by

where theunity, and

ourier transfo

denotes a con

Figu

authorized uon from the

d stego imageourier transfo

e obvious fromformation by ption performy a fringe-adju

e filter paramed to overcome

(uS ,

ormation of Eq

nvolution ope

ure 5: Block

Figure 6:

user having encrypted s

e and the resuormation can y

(xe ,m the above eeliminating a

mance in ordeusted filter (F

eters, A(u,v) ae the pole prob

) ( )( vujS

vuSv

,

,,

4

1

−

=

q. (12) yields( ) (cyxs 4, =

eration.

diagram of th

Block diagra

the correct sstego image ult is multiplyield the decr) ({Jy, 1ℑ= −

equation that any unwanteder to remove

FAF) whose tr

( ) [,B

vuH =

and B(u,v) canblem, respect

( )) ( )vuC

vujS

,4

,*

2

=

−+

s the encrypte( ) ( yxdyx ,, ⊗

he MRJTC-ba

am of the bit m

set of keys of Eq. (13).lied by the enrypted image ( ) ( )}vuCvu ,,the proposed

d signal due te any possibleransfer functi

( )( ) (,

,uCvuB

vuA+

n be adjustedtively.

( )) ( )vuD

vuS

,

,3−

ed image as gi)y

ased decrypti

mapping tech

can easily r. Fourier tranncryption keyas given by } ( )yxt ,4= d technique cato correlatione spurious siion is given b

) ]2, v d to avoid hav

iven by

on technique

hnique

retrieve the nsformation y as shown i

an produce ann operation. Tgnals, the sig

by

ving an optica

original biomis applied oin Fig. 5. Th

n uncorruptedTo further engnal in Eq. (

al gain greate

(12)

(13)

metric on the hen an

(14) d form nhance 14) is

(15)

r than



Encode

Multipl

Image

ad &

exed

Code 1

i coi

Code N

Cor

mpression

npression 1-

Correlation

Energy &

Threshold

IVV!!VIQIIVII

Energy &

Threshold

1_, Recovere

jImage 1

_FRecovere

Image N

The encrytechniquerespective The recovencryptedwith the rthe orthogdecryptedexpressed

where K(respectiveplane afteanything zero. Anodemultipl

The propoits performWalsh codone iris sc Figures 8(the cover together tinput imag

F

ypted and mue depicted in Fe stego image

vered hiddend and multiplerespective encgonal code sed, then the outd as

(r

u, v) and T(uely. It can be er the threshobecause of th

other salient fex as this is a

osed informatmance in varde of length 4can, and samp

(a) – 8(d) shoimage using

to form an enges.

Figure 7: Blo

u8ltiplexed inFig. 6. The use pixel bits.

n image fromexed togethercryption key.et used. If ki(tput signal aft

( ) ∫∫ ∑∞ =⎢⎣

⎡=yx

N

k,

u, v) are the observed in Eold operationhe fact that thfeature of the automatically

tion security srious scenario4 was employple text inform

ow the sampleg orthogonal ncoded image

ck diagram o

nformation caser personal k

m the above r. To decrypt Then a thres

(x,y) representer the correla

( )∑=

TvuKN

kk ,1

Fourier transEq. (16) that n, because othe cross-correproposed tecaccomplished

SIMULA

system was sios. Three setsyed for encodimation or han

e binary biomcode. The im

e as shown in

f the orthogon

an then be rekey is employ

step containt specific infoshold operationts the addresation operatio

( ) ( −⎥⎦

⎤ uKuu i,

sforms of theonly the desi

ther biometrielation betwehnique is thatd during the c

ATION RES

imulated usins of differenting purpose. E

nd writing.

metric signatumages were en Fig. 8(e), wh

nal decoding

etrieved from yed to recover

ns multiple bormation, theon is performss code for thon is performe

)−− duyvx,

e spatial domired signatureic signatures een the any tat no additionacorrelation op

SULTS

ng MATLAB t biometric siEach biometr

ures to be encrencoded usinghich contains

technique

m the stego imr the hidden im

biometric infe recovered immed dependinhe ith biometed on the rece

udv

main signals, ke will be prodwill not be

two orthogonal processingperation.

software in oignatures werric set contain

rypted in the g Walsh cods the informat

mage followinmage bits fro

formation thamage is mult

ng on the formtric signature eived signal c

k(x, y) and t(duced at the oable to cont

nal codes is ag step is requi

order to invesre employed ns two fingerp

red (R) chande and multiption of all the

ng the om the

at are tiplied mat of

to be can be

(16)

(x, y), output tribute always red to

stigate and a

prints,

nnel of plexed e four



This is asimple testwhich containssome testbiometric dataand sample text

1234567890

1j íi

II°,,L UB 9 ï I II

r. I m Iid II

!lhill , ...., I

,

1 H

A kRandom

Text

(a) (b) (c) (d)

(e)

Figure 8: Orthogonal coding and multiplexing of biometric signatures for red (R) cover image: (a) – (d)

input biometric information to be encrypted, (e) encrypted and multiplexed image

(a) (b) (c) (d)

(e)

Figure 9: Orthogonal coding and multiplexing of biometric signatures for green (G) cover image: (a) –

(d) input biometric information to be encrypted, (e) encrypted and multiplexed image Figures 9(a) – 9(d) show the second set of binary biometric signatures that were hidden in the green (G) channel of cover image. The encrypted and multiplexed image using orthogonal code is shown in Fig. 9(e), which contains all the four biometric information. Similarly, Fig. 10 shows the third set of biometric signatures that are encrypted and multiplexed together in order to be embedded into the blue component of the cover image.



ryt-----\ Sign Here

(a) (b) (c) (d)

(e)

Figure 10: Orthogonal coding and multiplexing of biometric signatures for blue (B) cover image: (a) –

(d) input biometric information to be encrypted, (e) encrypted and multiplexed image

(a) (b)

Figure 11: Effect of steganography on a color cover image: (a) image before steganography, and (b)

image after embedding twelve different binary biometric signature images

Figure 11 shows the effect of steganography process on a color cover image. The color image of Fig. 11(a) was used as the cover image to hide all the twelve biometric information depicted in Figs. 8 – 10. The cover image was split into its color components, each of which then contained the respective biometric information through the bit replacement process. The resultant stego image is shown in Fig. 11(b). It can be obvious from the two images that there is no visible change in the given image even after embedding twelve biometric signatures. Histogram analyses were carried out on the individual color channels which also did not show any significant changes. Additional color images were employed as the cover images with varying background and contents. It was observed that the original images and their corresponding stego images offer identical view to regular eyes. Even the histograms of the original and resulted stego images offer no difference to raise a suspicion about image alteration. The stealth capability of the proposed technique was also investigated by using blind weighted stego image analysis to estimate the LSB change rates [18]. The simulation results are summarized in Table 1. Close change rate between a specific channel of the original cover image and



resulted stego image indicates that it would be very difficult to determine if any information is hidden using a blind steganalysis method.

Table 1: Weighted stego image change rates for original cover image and stego image

Weighted Stego

Original Cover Image

After Steganography

Cover Image 1 Red 0.00069297 0.00071582Cover Image 1 Green 0.00138481 0.00140686Cover Image 1 Blue 0.00017203 0.00019150Cover Image 2 Red 0.00131138 0.00122920Cover Image 2 Green 0.00167064 0.00152484Cover Image 2 Blue 0.00094766 0.00114984Cover Image 3 Red 0.00919333 0.00838849Cover Image 3 Green 0.00329215 0.00334655Cover Image 3 Blue 0.00528524 0.00569346Cover Image 4 Red 0.04288447 0.04337175Cover Image 4 Green 0.08928525 0.09171086Cover Image 4 Blue 0.14010596 0.15743530

The stego image of Fig. 11(b) is then nonlinearly encrypted using the MRJTC technique, where the FAF parameters were chosen to as A(u,v) = 1.0 and B(u,v) = 10-4. The image is placed in a joint input image along with the encryption key generated in the form of an image as shown in Fig. 12(a). The resultant encrypted stego image is shown in Fig. 12(e), which makes it obvious that the confidentiality of the input information is well-protected because of the nonlinear encryption process. Computer simulation experiment verified that the original stego can easily be recovered from the encrypted image of Fig. 12(b) without losing any information. Further investigations proved that the decryption process is so robust that it was still successful in retrieving the information under noisy environment and even when part of the image was occluded.



(a) (b)

Figure 12: MRJTC-based encryption technique: (a) stego image, and (b) encrypted stego image

CONCLUSION An efficient information security system is developed in this paper to protect biometric signatures from any unauthorized access. The technique employs orthogonal coding scheme to encode multiple biometric information and multiplex together which makes it almost impossible to decode even a part of the secret information. Though the simulation results presented in the paper included four biometric signatures, the orthogonal coding scheme can be extended to theoretically any number of information. The encoded biometric information is then embedded into a color cover image using a second set of keys to choose one from the three least significant bits. The steganography technique is so efficient in making the information completely hidden such that any steganalysis method can easily fail to retrieve any information. The stego image is finally encrypted using a modified joint transform correlation technique which offers a high level of security through incorporation of a nonlinear encryption process, however, involving a rather simple architecture. Simulation experiments verify that the proposed technique can yield an effective system for securing information, including biometric signatures, personal identification information, as well as confidential documents, and to verify the identity in real time.

ACKNOWLEDGMENT This work is supported by a research grant from the Department of Energy, USA.

REFERENCES

[1] A. K. Jain, K. Nandakumar and A. Nagar, “Biometric template security,” EURASIP Journal on Advances in Signal Processing, vol. 2008, ID. 579416, 2008.

[2] R. Sanchez-Reillo, C. Sanchez-Avila and A. Gonzalez-Marcos, “Biometric identification through hand geometry measurements,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, no. 10, pp. 1168 – 1171, 2000.

[3] Y. C. Feng and P. C. Yuen, “Binary discriminant analysis for generating binary face template,” IEEE Transactions on Information Forensics and Security, vol. 7, no. 2, pp. 613 – 624, 2012.

[4] M. Fouad, A. El Saddik, Z. Jiying and E. Petriu, “A fuzzy vault implementation for securing revocable iris templates,” Proceedings on IEEE International Systems Conference (SysCon), pp. 491 – 494, 2011.



[5] L. Delac and M. Grgic, “A survey of biometric recognition methods,” Proceedings of International Symposium on Electronics in Marine, pp. 184 – 193, 2004.

[6] D. Bala, “Biometrics and information security”. Proceedings of the 5th annual conference on Information security curriculum development (InfoSecCD '08) ACM, pp. 64 – 66, 2008.

[7] J. E. Mills and S. Byun, “Cybercrimes against consumers: could biometric technology be the solution?,” IEEE Internet Computing, vol. 10, no. 4, pp. 64 – 71, 2006.

[8] A. Chen and V. Chandran, “Biometric template security using higher order spectra,” Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 1730 – 1733, 2010.

[9] A. Cheddad, J. Condell, K. Curran and P. McKevitt, “Biometric inspired digital image steganography,” Proceedings of 15th Annual IEEE International Conference and Workshop on the Engineering of Computer Based Systems (ECBS 2008), pp. 159 – 168, 2008.

[10] X. Luo, Z. Hu, C. Yang and S. Ga, “A secure LSB steganography system defeating sample pair analysis based on chaos system and dynamic compensation,” Proceedings of The 8th International Conference on Advanced Communication Technology, vol. 2, pp. 1014 – 1019, 2006.

[11] A. D. Ker, “Steganalysis of embedding in two least-significant bits,” IEEE Transactions on Information Forensics and Security, vol. 2, no. 1, pp. 46 – 54, 2007.

[12] M. Goljan, J. Fridrich and T. Holotyak, “New blind steganalysis and its implications,” Proceedings of SPIE in Security, Steganography and Watermarking of Multimedia Contents VIII, vol. 6072, pp. 1 – 13, 2006.

[13] M. N. Islam, “Encryption and multiplexing of fingerprints for enhanced security,” Proceedings of IEEE Long Island Systems, Applications and Technology Conference (LISAT), 2011.

[14] M. F. Islam and M. N. Islam, “A secure approach for encrypting and compressing biometric information employing orthogonal code and steganography,” SPIE Proceedings in Optical Pattern Recognition XXIII, 2012.

[15] M. F. Islam and M. N. Islam, “Efficient and robust information security system employing color encoded steganography and orthogonal encryption,” Proceedings of IASTED International Conference on Visualization, Imaging and Image Processing, July 2012.

[16] M. N. Islam, “Color image encryption using multiple reference joint transform correlation,” Proceedings of IEEE Long Island Systems, Applications and Technology Conference (LISAT), 2012.

[17] F. R. K. Chung, J. A. Salehi and V. K. Wei, “Optical orthogonal codes: design, analysis and applications,” IEEE Transactions on Information Theory, vol. IT-35, pp. 595 – 604, 1989.

[18] R. McEvoy, J. Curran, P. Cotter and C. Murphy, “Fortuna: cryptographically secure pseudo-random number generation in software and hardware,” Proceedings of Irish Signals and Systems Conference (IET), pp. 457 – 462, 2006.

[19] A. D. Ker and R. Böhme, “Revisiting weighted stego-image steganalysis,” Proceedings of SPIE in Security, Forensics, Steganography, and Watermarking of Multimedia Contents X, vol. 6819, San Jose, CA, pp. 0501 – 0517, 2008.



spie proceedings [spie spie defense, security, and sensing - baltimore, maryland, usa (monday 29...

Documents