International Journal of Computer Engineering and ... - IJCEA [PDF]

Jan 18, 2018 - Finally some chaff points are ... vault V is constructed by the union of unordered set G and chaff point

0 downloads 6 Views 756KB Size

Recommend Stories


International Journal of Computer Engineering and Applications, FUNDAMENTALS OF MACHINE
Life isn't about getting and having, it's about giving and being. Kevin Kruse

International Journal of Computer Engineering and Applications, COMPARATIVE STUDY AND
The happiest people don't have the best of everything, they just make the best of everything. Anony

International Journal of Computer Engineering and Applications, APPLICATION OF GAUSSIAN
Almost everything will work again if you unplug it for a few minutes, including you. Anne Lamott

International Journal of Computer Engineering and Applications, MICROCONTROLLER BASED
No amount of guilt can solve the past, and no amount of anxiety can change the future. Anonymous

International Journal of Engineering and Advanced Technology International Journal of
If you want to go quickly, go alone. If you want to go far, go together. African proverb

International Journal of Engineering and Advanced Technology International Journal of
Don't ruin a good today by thinking about a bad yesterday. Let it go. Anonymous

International Journal of Food Engineering
Don’t grieve. Anything you lose comes round in another form. Rumi

International Journal of Food and Biosystems Engineering
Your big opportunity may be right where you are now. Napoleon Hill

International Journal of Earth Sciences and Engineering - Cafet Innova [PDF]
Editor In Chief: Dr. D. Venkat Reddy. Retired Professor National Institute of Technology Karnataka (NITK) Department of Civil Engineering Karnataka Karnataka,IND. Email: [email protected]. About Journal. The International Journal of Earth Sc

International Journal of Mobile Human Computer Interaction
When you talk, you are only repeating what you already know. But if you listen, you may learn something

Idea Transcript


International Journal of Computer Engineering and Applications, Volume XII, Issue I, Jan. 18, www.ijcea.com ISSN 2321-3469

TEMPLATE PROTECTION SCHEME FOR MULTIBIOMETRIC SYSTEM USING FUZZY VAULT BASED ON FEATURE LEVEL FUSION 1Sandip

Kumar Singh Modak, 2Vijay Kumar Jha

Department of Computer Science & Engineering, Birla Institute of Technology, Mesra (Ranchi)835215 1 Research Scholar,2Associate Professor 1 [email protected], [email protected]

ABSTRACT: Security of multibiometric template is crucial as they contain information regarding multiple traits of the same user. Hence multibiometric template protection is the main focus of this work. Stored biometric template can be replaced by an imposter’s template to gain unauthorized access; a physical spoof can be created from the template to gain unauthorized access to the system and the stolen template can be replayed to the matcher to gain unauthorized access. Multibiometric based systems are more secure compare to the unimodal based system and this is one of the factors for its popularity. Fuzzy vault and fuzzy commitment are two well known schemes for template security. Fuzzy vault is a biometric cryptosystem that secure both the secret key and the biometric template by binding them within a cryptographic framework. In this paper we propose a template protection scheme based on feature level fusion using fuzzy vault .this work extracts feature points from two different biometric modalities namely fingerprint and iris. After the feature level fusion of these two modalities the feature sets are evaluated on a generated polynomial which is turn is derived from a randomly generated key. Finally some chaff points are added that do not lie on the polynomial to form a vault. Since the points lying on polynomial encode the complete information about the template and the secret key, concealing these points secure both the template and secret simultaneously. The main purpose of Embedding algorithm in the process of feature level fusion is to converting different biometric representation into a common representation space.

Keywords: Multibiometric, Fuzzy Vault, Feature level Fusion, Embedding Algorithm, Template Protection, Iris, Fingerprint.

Sandip Kumar Singh Modak, Vijay Kumar Jha

1

TEMPLATE PROTECTION SCHEME FOR MULTIBIOMETRIC SYSTEM USING FUZZY VAULT BASED ON FEATURE LEVEL FUSION

[I] INTRODUCTION Multibiometric system can be define as one that combine the outcome obtained from multiple biometric features for the purpose of identification. Unlike a unimodal biometric system that may result in non-universality a multimodal system uses multiple biometric modalities that can result in highly accurate and secure biometric identification system [1].compare to traditional biometric system, multibiometric based system [2] offer several advantages like higher security, user’s convenience, maximum accuracy and high performance. Multibiometric system deals with the multiple templates for the same user corresponding to the different biometric sources. Template protection in a multibiometric system is a challenging task because the template cannot be easily revoked and reissued. Stored biometric template can be replaced by an imposter’s template to gain unauthorized access; a physical spoof can be created from the template to gain unauthorized access to the system and the stolen template can be replayed to the matcher to gain unauthorized access. Multibiometric based systems are more secure compare to the unimodal based system and this is one of the factors for its popularity. Fuzzy vault and fuzzy commitment are two well known schemes for template security [3].biometric cryptosystem can operate in one of three modes namely key release, key binding and key generation. In the key release mode the biometric template and the key are stored in separate entities and whenever the matching is successful the key is release. In the key binding mode both the key and template are bound within the cryptographic framework. Fuzzy vault and fuzzy commitment are comes under this mode. In the key generation mode, the key is derived directly from the biometric data and is not stored in the database [4].in this paper we propose template protection scheme for multibiometric system consisting of fingerprint and iris modalities using fuzzy fault based on feature level fusion. Biometric cryptosystem based on key binding (fuzzy vault, fuzzy commitment) modes are more secure than the key release based system, but it is very difficult to implement due to large intraclasss variation in biometric data. In the case of fingerprint there may be lots of variation may occurs due to translation, rotation, nonlinear distortion, skin condition [5]. The remainder of this paper is organise as follows. Section II discuses the related work proposed earlier in literature for multibiometric using fuzzy vault. Section III describes the feature level fusion technique. Section IV gives details about proposed work. Section V concludes the paper. [2] RELATED WORK Fuzzy vault construct is a biometric cryptosystem that secure both the secret key and the biometric template by binding them within a cryptographic framework [6].fuzzy vault framework thus utilize the goodness of both cryptography and biometric. In fuzzy vault framework, the secret key S is locked by G, where G is an unordered set from the biometric sample.a polynomial P is constructed by encoding the secret S.the polynomial is evaluated by all the elements of the unordered set G.a vault V is constructed by the union of unordered set G and chaff point set C which is not in G[7]. V=G ∪ C Let x denote a biometric template with r elements. The user select a key k, encoded it in the form of polynomial P of degree n and evaluate the polynomial P on all the elements in x.the points lying on P are hidden among a large number of random chaff point that do not lie on P and the union of genuine and chaff point sets constitute the helper data or vault V.in the absence of user biometric Sandip Kumar Singh Modak, Vijay Kumar Jha

2

International Journal of Computer Engineering and Applications, Volume XII, Issue I, Jan. 18, www.ijcea.com ISSN 2321-3469 data it is very hard to identify the genuine point in V and hence the template is secure. During authentication the user provide a biometric query denoted by x’.if x’ overlaps substantially with x, the user can identify many point on V that lie on the polynomial. On the other hand if x and x’ do not have sufficient overlap, it is very difficult to reconstruct the polynomial and the authentication is unsuccessful [8].

Fingerprint Vault Encoding

Fingerprint Vault Decoding

Biometric template protection scheme using fuzzy vault based on feature level was first developed in [8].they uses fingerprint and iris as a biometric modalities and perform the feature level fusion to form the fuzzy vault.in this work fingerprint minutiae encoding was done according to[4],but a two-step novel technique was devised to encode the iriscodes. And finally feature level techniques called concatenation method are used to fuse the feature of fingerprint and iris. The author in [9] suggested a framework based on feature level fusion of three biometric traits namely fingerprint, iris and retina. The combination of three biometric modalities namely fingerprint, iris and retina are analysed by meenakshi in their series of works. They use grouping of retina and fingerprint in [10], fingerprint and iris in [11] and retina and iris in [12]. Brindha and Natarajan [13] proposed a multibiometric template security scheme based on fuzzy vault. They combined the feature of fingerprint and palmprint in their work.

[3] FEATURE LEVEL FUSION Feature level fusion plays an important role in the process of data fusion. Advantage of feature level fusion including derive the most discriminatory information from multiple feature sets

Sandip Kumar Singh Modak, Vijay Kumar Jha

3

TEMPLATE PROTECTION SCHEME FOR MULTIBIOMETRIC SYSTEM USING FUZZY VAULT BASED ON FEATURE LEVEL FUSION

involve in fusion. Feature fusion is capable of deriving and gaining the most effective and least dimensional feature set that benefit the final decision [14].feature level fusion is consider more powerful and effective than the other level of fusion like score and decision level fusion. The main reason of its effectiveness is the feature set contains richer information about the input biometric data. Feature level fusion is expected to provide better recognition result. However score level fusion and decision level fusion are more popular in the literature and there is not much research on feature level fusion. The main difficulty in feature level fusion is cases where the features are not compatible.e.g eigen-coefficient of face and minutiae sets of fingerprints. The main goal of feature level fusion is to combine the feature set from different sources and combine them into a single one. Two popular feature leve fusion are: serial and parallel feature fusion. Serial feature fusion works by simply concatenating two or more feature vector into a single feature vector. Suppose first feature vector of p-dimension and second feature vector of q-dimension then the fused feature vector will be of (p+q) dimension. Parallel feature fusion on the other hand combine the two feature vector into a complex vector z=x+iy (i is imaginary) [15].the most successful feature level fusion method now a day is CCA (canonical correlation analysis) which uses the correlation between two set of feature to find two sets of transformation such that the transformed feature have maximum correlation across the two feature sets [16]. But CCA based feature fusion often suffer from the small sample size problem. The kernel trick (KCCA) is an effective way to solve the small size problem and non-linear problem [17].

[4] PROPOSED WORK In this paper we propose a template protection scheme using fuzzy vault based on feature level fusion of fingerprint and iris. Initially in the enrolment phase the feature sets of fingerprint and iris are stored in respective database. Fuzzy vault Encoding: the first step of vault construction is feature extraction from fingerprint and Iris. The following are the steps for feature extraction. A. Fingerprint Feature Extraction: fingerprint pre-processing is the first step required to enhance the quality of image before feature extraction step. The main objective of pre-processing step is to reduce the noise which is produce during the capture of fingerprint image. Histogram equalization (HE) technique is used to enhance the image quality. HE flattens and stretches the dynamic range of the image histogram [18].the improve image after HE is shown in fig 1 which is used for feature extraction purpose.

Fig 1 a) input fingerprint image

b) Histogram Equalized Image

Sandip Kumar Singh Modak, Vijay Kumar Jha

4

International Journal of Computer Engineering and Applications, Volume XII, Issue I, Jan. 18, www.ijcea.com ISSN 2321-3469

Haar Wavelet Transform: wavelet transform is a mathematical tool based on many layer function decomposition. Haar wavelets transform technique, one most popular amongst wavelets is applied for feature extraction from fingerprints. If f(x, y) represent an image signal its Haar wavelet Transform is equal to two 1D filter as shown below.

From this image a single 1x60 feature vector is extracted by row wise serialization. This itself treated as extracted feature set of fingerprint.

Sandip Kumar Singh Modak, Vijay Kumar Jha

5

TEMPLATE PROTECTION SCHEME FOR MULTIBIOMETRIC SYSTEM USING FUZZY VAULT BASED ON FEATURE LEVEL FUSION

Vault Decoding Iris Feature Extraction: Pre-processing of the Iris image is required before feature extraction.pre-processing comprises of Localization, Segmentation and Normalization. The localization and segmentation are performed by canny edge detection [19] and circular Hough transform [19] to detect the iris boundaries and reduce their radius and center.

Iris Localization, segmentation and normalization

We apply Hough transform first for Iris /sclera boundary and then to Iris /pupil boundary for more accurate segmentation. The output of this step results in storing the radius and x, y parameters of inner and outer circles. The extracted iris region is fixed into rectangular block by remapping each point within the iris region to polar co-ordinate using rubber sheet model developed by daughman [19].

[5] EMBEDDING ALGORITHM: The embedding algorithm transform a biometric feature representation Xm into a new feature representation Zm, where Zm=€m(Xm) for all m=1,2,.....M The input representation X can be a real value feature vector, a binary string, or a point set. Whereas the output representation Z could be a binary string or a point set that could be secured Sandip Kumar Singh Modak, Vijay Kumar Jha

6

International Journal of Computer Engineering and Applications, Volume XII, Issue I, Jan. 18, www.ijcea.com ISSN 2321-3469 using fuzzy commitment or fuzzy vault. Fuzzy commitment is used if Z is a binary string, whereas a fuzzy vault is used if Z is a point sets [9].

Suppose Alice select the polynomial 𝑝(𝑥) = 𝑥 2 − 3𝑥 + 1,where the co-efficient (1,-3,1) encode her secret key k.if her unorder set after feature level fusion is A={-1,-2,3,2},then she obtain the polynomial projection as {A,P(A)}={(-1,5),(-2,11),(3,1),(2,-1)}.now she add two chaff points C={(0,2),(1,0)} that do not lie on p,to find the final set R= {(-1,5),(-2,11),(3,1),(2,1),(0,2),(1,0)}and this R constitute a vault

[6] FUZZY VAULT DECODING Now if Bob can separate at least 3 points from R that lie on P, he can reconstruct P ,hence decode the secret representation as the polynomial co-efficient (1,-3,1).otherwise he will end up with incorrect p and he will not be able to access the key k. We try to unlock the vault with the query feature.we have N query feature pints (Q) ,U1*,U2*,U3*,........UN* and the points to be used in polynomial reconstruction are found by comparing Ui*,i=1,2,3.....N with the abscissa values of the vault V.if any Ui*,i=1,2,3....N is equal to V ,the corresponding vault point (vl,wl) is added to the list of points to be used. We use langrage interpolation to construct the polynomial [20].

L={(v1,w1),(v2,w2),(v3,w3)........} The corresponding polynomial is (𝑈−𝑉2)(𝑈−𝑉3)…….(𝑈−𝑉𝐷)

(𝑈−𝑉1)(𝑈−𝑉3)……(𝑈−𝑉𝐷)

P*(u) =(𝑉1−𝑉2)(𝑉1−𝑉3)……(𝑉1−𝑉𝐷)w1 + (𝑉2−𝑉1)(𝑉2−𝑉3)……(𝑉2−𝑉𝐷)w2+.................... (𝑢−𝑉1)(𝑢−𝑉2)…….(𝑢−𝑉𝐷)

...................+(𝑉𝐷+1−𝑉1)(𝑉𝐷+1−𝑉2)…..(𝑉𝐷+1−𝑉𝐷)wD+1

[7] CONCLUSION In this paper we propose a template protection scheme for multibiometric system using fuzzy vault based on feature level fusion.multibimetric system accumulate evidence from multiple user traits or more than one biometric trait of a user to recognise a person. Template security is essential to protect both the integrity of the biometric system and the privacy of the users. We have also proposed different embedding algorithm for transforming biometric representation. As we use

Sandip Kumar Singh Modak, Vijay Kumar Jha

7

TEMPLATE PROTECTION SCHEME FOR MULTIBIOMETRIC SYSTEM USING FUZZY VAULT BASED ON FEATURE LEVEL FUSION

fuzzy vault as a template protection scheme, so before feature level fusion of fingerprint and iris it is mandatory to convert these feature sets into a point-set feature. These point set are used for the construction of vault. We have shown that the multibiometric vault can secure template from multiple modalities like fingerprint and iris.we have also demonstration that the multibiometric vault provide better recognition performance as well as higher security compare to the unibiometric vault.fuzzy vault has been widely used for proving security and for that some chaff point which is not originally lie on genuine point are added to construct the vault.

REFERENCES: [1]. C.Prathipa and L. Latha,”A survey of multimodal biometric fusion and template security techniques.”IJARCET volume 3, pp 3511-3516, 2014. [2]. Ross, Arun A., Karthik Nandakumar, and Anil Jain. Handbook of multibiometrics. Vol. 6. Springer Science & Business Media, 2006. [3] Vetro, A., and N. Memon. "Biometric system security." Tutorial presented at Second International Conference on Biometrics, Seoul, South Korea. 2007. [4]. Nandakumar, Karthik, Anil K. Jain, and Sharath Pankanti. "Fingerprint-based fuzzy vault: Implementation and performance." IEEE transactions on information forensics and security 2.4 (2007): 744757. [5]. Maltoni, D., Maio, D., Jain, A., & Prabhakar, S. (2009). Handbook of fingerprint recognition. Springer Science & Business Media [6]. Juels, Ari, and Madhu Sudan. "A fuzzy vault scheme." Information Theory, 2002. Proceedings. 2002 IEEE International Symposium on. IEEE, 2002. [7]. Meenakshi, V. S., and G. Padmavathi. "Security analysis of password hardened multimodal biometric fuzzy vault with combined feature points extracted from fingerprint, iris and retina for high security applications." Procedia Computer Science 2 (2010): 195-206. [8]. Nandakumar, Karthik, and Anil K. Jain. "Multibiometric template security using fuzzy vault." Biometrics: Theory, Applications and Systems, 2008. BTAS 2008. 2nd IEEE International Conference on. IEEE, 2008. [9]. Nagar, Abhishek, Karthik Nandakumar, and Anil K. Jain. "Multibiometric cryptosystems based on feature-level fusion." IEEE transactions on information forensics and security 7.1 (2012): 255-268. [10]. Meenakshi, V. S., and G. Padmavathi. "Security analysis of password hardened multimodal biometric fuzzy vault with combined feature points extracted from fingerprint, iris and retina for high security applications." Procedia Computer Science 2 (2010): 195-206. [11]. Meenakshi, V. S., and G. Padmavathi. "Secure and revocable multibiometric templates using fuzzy vault for fingerprint and iris." International Conference on Advances in Information and Communication Technologies. Springer Berlin Heidelberg, 2010. [12]. Meenakshi, V. S., and G. Padmavathi. "Retina and iris based multimodal biometric fuzzy vault." IJCSIS International Journal of CSIS 7.2 (2010).

Sandip Kumar Singh Modak, Vijay Kumar Jha

8

International Journal of Computer Engineering and Applications, Volume XII, Issue I, Jan. 18, www.ijcea.com ISSN 2321-3469 [13]. Brindha, V. Evelyn, and A. M. Natarajan. "Multi-modal biometric template security: Fingerprint and palmprint based fuzzy vault." Journal of Biometrics and Biostatistics 3.6 (2012): 1-6. [14]. Yang, Jian, et al. "Feature fusion: parallel strategy vs. serial strategy." Pattern Recognition 36.6 (2003): 1369-1381. [15].Haghighat, Mohammad, Mohamed Abdel-Mottaleb, and Wadee Alhalabi. "Discriminant correlation analysis: Real-time feature level fusion for multimodal biometric recognition." IEEE Transactions on Information Forensics and Security 11.9 (2016): 1984-1996. [16]. Xu, Xiaona, Yue Zhao, and Haijun Li. "The study of feature-level fusion algorithm for multimodal recognition." Computing Technology and Information Management (ICCM), 2012 8th International Conference on. Vol. 1. IEEE, 2012. [17] Shawe-Taylor, John, and Nello Cristianini. Kernel methods for pattern analysis. Cambridge university press, 2004. [18].Besbes F., Trichili H. and Solaiman B. (2008) 3rd International IEEE Conference on Information and Communication Technologies, From Theory to Applications, Syria, 1 5. [19]. Daugman J. (2004) Video Technology, 14(1), 21 -39.

IEEE

Journals

on

Circuits

and

Systems

for

[20]. Uludag, Umut, Sharath Pankanti, and Anil K. Jain. "Fuzzy vault for fingerprints." International Conference on Audio-and Video-Based Biometric Person Authentication. Springer Berlin Heidelberg, 2005.

Sandip Kumar Singh Modak, Vijay Kumar Jha

9

Smile Life

When life gives you a hundred reasons to cry, show life that you have a thousand reasons to smile

Get in touch

© Copyright 2015 - 2024 PDFFOX.COM - All rights reserved.