Enhancing Iris Biometric Recognition System with Cryptography
and Error Correction Codes: A Preliminary Study

Abstract
Major challenge on iris and most biometricidentifier is the intra user variability in the acquired identifiers.Iris of the same person captured in different time may differ dueto the signal noise of the environment or the iris camera. In ourproposed approach, Error Correction Code, ECC is thereforeintroduced to reduce the variability and noise of the iris data. Inorder to find best system performance, the approach is testedusing 2 different distance metric measurement functions for theiris pattern matching identification process which are HammingDistance and Weighted Euclidean Distance. An experimentwith the CASIA version 1.0 iris database indicates that ourresults can assure a higher security with a low false rejection orfalse acceptance rate.
I. INTRODUCTION
In today society, security has been a major concern and isbecoming increasingly important. Cryptography has becomeone of the most effective way and has been recognized as themost popular technology for the security purposes. Historyshows that human can remember only short password [1],most of the user even tend to choose password that can beeasily guessed using dictionary or brute force attack.Thislimitation has triggered the utilization of biometric to producestrong cryptographic key. Biometric is unique to eachindividual and it is reliable. For years, securities to the worldare mostly based on what one knows such as password, PIN,or a security question such as mother’s maiden name. Thissecurity feature is easily forgotten, stolen, shared andcracked. Biometric is a measurement of the human bodybiological or physical characteristics to determine the humanidentity of (who you are?). There are different types ofbiometric technologies available today which includefingerprints, face, iris/retina, hand geometry, signature, DNA,keystroke, tongue and etc. Biometric offered an inextricablylink from the authenticator to its owner, which cannot done bypasswords or token, since it cannot be lent or stolen [2].Biometric is used to enhance the privacy and securityweaknesses exist in the current security technology such assimple password or PIN authentication. Another advantage ofbiometric is that it can detect and prevent multiple ID’s.Although biometric is unique among all individuals, it is notpossible to use biometric as a direct cryptography key for thesystem due to the difference bits occur in the template duringevery authentication. Biometric images or templates arevariable by nature which means each new biometric sample isalways different. Noises and error may occur in the capturedimage due to burst or background error and hence thegenerated template is different during every authentication.There is also awareness concerning the privacy and thesecurity of personal information due to the storage of thebiometric templates. The loss or compromise of biometrictemplates may end up to unusable biometric. Humanphysiological biometric is hardly change or do modification ifthe image template has been stolen or compromised.This paper presented an ideal biometric authenticationsystem which fulfils the important properties such asnon-repudiation, privacy, and security. The important idea toachieve these properties is by combining both the irisbiometrics and password. The intra personal variation (noisesand error occurred in the genuine iris code) is solved byintroducing the Error Correction Codes (ECC). The ErrorCorrection Codes helps to correct errors on the genuineidentification making the threshold value to be smaller whichsignificantly helps in improving the user separation of theimpostor verification. In this paper, two different distancemetric function used in the template matching process will becompared. The next section provides the review of theliterature work, followed by the methodology andexperimental results. Finally, the last section gives theconclusion remarks and the future work of this paper.II. LITERATURE REVIEWA. Biometric TechnologiesBiometric has become a very popular topic and gained highinterest from the researcher for around 25 years. Biometricrefers to the identification and verification of a humanidentity by measuring and analyzing the physiological or thebiological information of a human. The term “biometric”come from Greeks gives the meaning of “the measurement oflife”. In recent century, there is a wide varieties ofidentification and authentication options for the securitywhich is something you have (key, smartcard, ID badges) andsomething you know (password, PIN, security questions). Sowhy do most security professionals will still choosebiometrics to be the “best” method for the identification andauthentication purpose? The answer is biometric come fromthe nature and it is the determination of “who you are”. Theuniqueness of biometric is an undeniable reality known by all.Biometric can be basically divided into two categories whichare structural (face, hand geometry or veins patterns,fingerprints, iris patterns, retina patterns, DNA) andbehavioral (voice, signature, keystroke). The diagram isshown in Figure 1.B. Error Correction CodesError Correction Codes are widcommunication channels. Role of this to correct errors that could occurover a noisy channel. Changing onegood cryptography system may caunreadable. Therefore, a way of deterror occur is needed. To overcombiometric data, Error Correctiontherefore been suggested. Dependioccur on the biometric identifier, erruse to eliminates the differences betverification readings [4]. Reed Soloby Irving Reed and Gus Solomontheir paper title “Polynomial CodFields” [5]. Reed Solomon Codescorrecting codes which is able to strahave an excellent burst correction cadescribed by a tuple (n,k,t) whereSolomon codes, k is the length of tnumber of error to be corrected.C. Template MatchingThere are several matching methrecognition. Some example of the feare such as Euclidean Distance [7]Standard Related coefficient [7][8][and etc. In this research, two diffemetric which is Hamming DistanceDistance are being applied to the sperformance to corresponding threshevaluated.D. Past WorksThe earliest history work on iris biThe concept of Daugman has becommodel for the researcher in this parhave been several researches on integbiometric in these recent years. Theidentify on biometric cryptography iin the acquired biometriccryptography method such as RSA,or suitable to use for storing the bioencrypted form and perform the mchanges in the bits of the featurebiometric may lead to a huge differencrypted feature. To solve thetechniques have been introducedThese techniques can be basically cawhich is Cancellable biometric, Keyand Key Generating Cryptosystem.Cancellable biometric possessed ta biometric template is being comproissues using different transform [Davida store the error correction bmight have lead to some leakage ofThe assumption of iris code bit cFacegerprintPalm/RetinaDNAVoicenatureystrokedely used in the field ofe Error Correction Codesr during the transmissione bit in the ciphertext of aause the plaintext to betecting and correcting theme the noisiness of then Codes technique hasing on the type of errorror correction codes weretween the enrollment andomon Code is introducedon January 21, 1959 ondes over Certain Finiteis a block based erroraight forward decode andapability. This block coden is the length of Reedthe message, and t is thehod for image biometricfeature matching methods], Absolute Distance [7],[9], Neural Network [10]erent matching distancesand Weighted Euclideansystem and each of theirhold value is recorded andiometrics is by Daugman.mes a standard referencerticular field [11]. Theregrating cryptography intoe major challenge that wes the intra user variabilityidentifiers. TraditionalAES or DES is not idealometric templates in andmatching directly. Minore set extracted from therence in the results of theabove problems, someby several researchers.ategorized into three typesy Binding Cryptosystem,the characteristic of that ifomised; a new one can be12]. In this paper [12],bits in the database thatf the user biometric data.changes among different

DOWNLOAD FULL REPORT
http://utopia.csis.pace.edu/cs615/btas20...%20153.pdf