Himanshu Srivastava / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.449-453449 | P a g ePersonal Identification Using Iris Recognition System, a ReviewHimanshu SrivastavaDepartment of Computer Science & Engineering, Roorkee Institute of Technology, Roorkee (U.K.), Indiahimanshu.firstname.lastname@example.orgAbstractA biometric system provides automatic identificationof an individual based on a unique feature orcharacteristic possessed by the individual. Unlike otherbiometric such as fingerprints and face recognition,the distinct aspect of iris comes from randomlydistributed features. Iris recognition is regarded as themost reliable and accurate biometric identificationsystem available. In this paper, I describe the noveltechniques developed to create an Iris RecognitionSystem available. This paper proposes a personalidentification using iris recognition system with thehelp of six major steps i.e. image acquisition,localization, Isolation, normalization, featureextraction and matching and also these six stepsconsists a numbers of minor steps to complete eachstep. The boundaries of the iris, as papillary andlimbic boundary, are detected by using Canny EdgeDetector & Circular Hough Transformation. We canuse masking technique to isolate the iris image formthe given eye image, this isolated iris image istransformed from Cartesian to polar co-ordinate. Nowfinally extract the unique features (feature vector) ofthe iris after enhancing the iris image and thenperform matching process on iris code using HammingDistance for acceptance and rejectance process. Her Iam giving my review after studying a number ofresearch papers and my proposed technique worksvery well and can be easily implemented.Keywords: Biometrics, Iris Code, Edge Detection,Transformation, Masking, Feature Vector,Hamming Distance.I. IntroductionNow a days, one of the main threats that IT systemand security environment can have, is the possibilityof intruders in the system. This is normally solvedby user authentication schemes based on passwords,secret codes and identification cards or tokens.Schemes based only on passwords or secret codescan be cracked by intercepting the presentation ofsuch a password or by brute force attacks. On theother hand, an intruder can attack systems based onidentification card or token by robbing, copying orsimulating them. As it is a well-known, biometricdeal with identification of individuals based on theirphysical and behavioral features.Biometric solutions, such as identification systemsusing fingerprint, iris, face, and palm print, handgeometry, signature, etc; have many advantagesover the traditional authentication techniques basedon what you know or what you possess. Instead ofcarrying bunk of keys, all those access cards orpasswords you carry around with you, your bodycan be used to uniquely identify you. Among them,iris recognition is tested as the most accuratemanner of personal identification. Thereforenowadays many automatic security systems basedon iris recognition have been deployed worldwidefor border control, restricted access and so on.Figure 1: Texture rich iris image This technique performs better identification thanother biometric identification because I am using aniris which has an extraordinary structure andprovides many interlacing minute characteristicssuch as freckles, coronas, strips, furrow, and cryptsand so on . These visible characteristics,generally called the texture of the iris, are unique toeach subject. The iris patterns of the two eyes of anindividual or those of identical twins babies arecompletely independent and uncorrelated . Allbiometric systems are suffered from two types oferrors as false acceptance and false rejectance. Butiris recognition system gives us lesser falseacceptance and false rejectance rate than any otherpersonal recognition system because iris is protectedbehind the eyelid, cornea and aqueous means that,unlike other biometrics such as fingerprints, thelikelihood of damage is minimal also the iris is notsubject to the effect of aging which means it remainsin a stable from about the age of one until death  so we do not need to update database again andagain over passes time. Thus the entire system isgiving us maximum accuracy and less error rate forpersonal identification in biometric analysis.Some current and future Applications: There isworld wide applications area of iris recognitionsystem. National border controls as living passport,computer login: a password, secure access to bankaccount at ATM machine, ticketless travel,authentication in networking, permission accesscontrol to home, office, laboratory etc, drivinglicenses, and other personal certificates, tracingmissing or wanted persons, anti-terrorism, securityat airports, using as any type of password.Figure 2: Traveler using iris scans security system 
Himanshu Srivastava / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.449-452450 | P a g eFigure 3: Mobile Security Figure 4: Security checking at border This paper represents the review about identificationof a person using iris recognition, after study anumber of papers, articles. My paper is organized asfollows: background work by a number of authors,researcher are given as related work in section II, inmy system, how it works, all the major and minorsteps as image capturing, edge detection, findinginner and outer boundary, isolation, Cartesian toPolar Transformation, enhancing the image,extracting features and matching, the entire mysystem approaches are given as iris recognitionsystem in section III, the entire theme and futurework is given as conclusion in section IV, in the lastall the papers and websites which I have used toprovide this review is given under the headingreferences in section V.II. Related workRelated to this system so many authors present theirtechniques, in this system I studied most of the goodpapers in sequence.Flom and Safir first proposed the concept ofautomated iris recognition . A report waspublished by Johansson in 1992 without anyexperimental results. Since then, iris recognition hasbeen receiving many researchers attention and agreat deal of progress has been achieved in the lastdecade. For instance, Daugman realized an irisrecognition system and the identification accuracy isup to 98.9%. Daugman proposed Integro-Differentiation method to find the pupil and irisboundary  he assumes pupil and iris boundary tobe circular . Daugman’s iris recognitionalgorithm is based on the principle of the failure of atest of statistical independence on iris phasestructure encoded by Quadrature Wavelets. Bolesand Boas hash  calculated Zero-Crossingrepresentation of 1-D Wavelet Transform at variousresolution levels of a virtual circle on an iris imageto characterize the texture of the iris. Iris matchingwas based on two dissimilarity functions. Wildes etal.  represented the iris texture with a LaplacianPyramid constructed with four different resolutionlevels.III. Iris Recognition SystemWhen a subject wishes to be identified by irisrecognition system, his/her eye is first photographedand then a template (iris code) created for his/heriris region. This template is then compared with theother templates stored in a database until either amatching template is found and the subject isidentified or no match is found and the subjectremains unidentified. There are several method foraccomplishing single task as edge detection I haveSobel Operator, Prewitt Operator, Canny EdgeDetection, Roberts method and so on, for boundarydetection as Hough Transformation, Circular HoughTransformation, Integro-Differential Operator,Gradient based approach, Clustering algorithms andso on, for feature extraction I have phase basedmethods, zero crossing and texture analysis basedmethods.As per studied by various papers a well definedsteps have come in my mind which can be disclosefor various readers, researchers and dictated in theform of diagram as shown in figure:5 given below.Figure 5: My SystemImage acquisition: This is our very first step of theentire process. When a person wishes to beidentified by iris recognition system, his/her eye isfirst photographed. The camera can be positionedbetween three and a half inches and one meter to
Himanshu Srivastava / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.449-452451 | P a g ecapture the image. In the manual procedure, the userneeds to adjust the camera to get the iris in focusand needs to be within six to twelve inches of thecamera. This process is much more manuallyintensive and requires proper user training to besuccessful. We must consider that the occlusion,lighting, number of pixels on the iris are factors thataffect the image quality . A number ofresearchers have been used Chinese Academy ofSciences Institute of Automation (CASIA)  andMalaysia Multimedia University (MMU) database to perform their research.Figure 6: Sample iris images Localization: The acquired iris image has to bepreprocessed to detect the iris, which is an annularportion between the pupil (inner boundary) and thesclera (outer boundary). The first step in irislocalization is to detect pupil which is the blackcircular part surrounded by iris tissues. The center ofpupil can be used to detect the outer radius of irispatterns. The important steps involved are:1. Pupil detection2. Outer iris localizationWell-known methods such as the Integro-Differential Operator, Hough Transform and ActiveContour models have been successful techniques indetecting the boundaries. The iris localizationproposed by Tisse et al. is a combination of theIntegro-Differential Operator and the HoughTransform. The Hough Transform is used for aquick guess of the pupil center and then the Integro-Differential Operator is used to accurately locatepupil and limbus using a smaller search space .But in this paper I am using Canny Edge Detectionfor detecting edges in the entire eye image after thatapplying Circular Hough Transformation fordetecting outer boundary of iris by using pupilcenter and inner boundary of iris.Figure 7: localized iris image Isolation: Now the task is to isolate the iris. In theimages used, there is some presence of the white ofthe eye. This was done by using a maskingtechnique as here I am choosing best techniqueamong other so I will use Gaussian Mask and thencropping the image to minimize the area that doesnot contain any edge data. The mask is a circularone which has the same radius as the iris. It thuspasses all pixels that are contained in the circlewhich are all the pixels forming the iris. By makinguse of the center and the radius which are calculatedin advanced step, we set the polar coordinatesystem. In this coordinate system, the feature of theiris is extracted.Figure 8: Isolated iris image Normalization: For the purpose of accurate textureanalysis, it is necessary to compensate thisdeformation. Since both the inner and outerboundaries of the iris have been detected so it iseasy to map the iris ring to a rectangular block oftexture of a fixed size. The Cartesian to polarreference transform suggested by Daugman authorizes equivalent rectangular representation ofthe zone of interest as shown figure. In this way Icompensate the stretching of the iris texture as thepupil changes in size, and unfold the frequencyinformation contained in the circular texture in orderto facilitate next feature extraction. Also this processis very necessary because feature extraction andmatching process becomes easy.Figure 9: Depicted how iris image is converted to PolarCoordinateThus the following set of equations  is used totransform the annular region of iris into polarequivalent.I(x(ρ, θ), y(ρ, θ)=>I(ρ,θ)Withxp(ρ,θ)=xp0(θ)+rp*cos(θ)yp(ρ,θ)=yp0(θ)+rp*sin(θ)xi(ρ,θ)=xi0(θ)+ri*cos(θ)yi(ρ,θ)=yi0(θ)+ri*sin(θ)Where, rp and ri are respectively the radius of pupiland the iris, while (xp(θ), yp(θ)) and (xi(θ), yi(θ)) arethe coordinates of the pupillary and limbicboundaries in the direction θ. The value of θ belongsto [0; 2], ρ belongs to [0; 1].Figure 10: Unwrapped image Enhancement and denoising: The normalized irisimage still has low contrast and may have non-uniform illumination caused by the position of lightsources. In order to obtain more well-distributed
Himanshu Srivastava / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.449-452452 | P a g etexture image, we enhance iris image by means oflocal histogram equalization and remove highfrequency noises by filtering the image with anappropriate filter.Figure 11: Iris image after enhancement and denoising Feature Extraction: Lim et al.  also use theWavelet Transform to extract features from the irisregion. Both the Gabor Transform and the HaarWavelet are considered as the Mother Wavelet.Laplacian of Gaussian (LoG) is also used inprevious papers. In my paper, using Haar Wavelet,decomposing upto 4thlevel a feature vector with 87dimensions is computed. Since each dimension hasa real value ranging from -1.0 to +1.0, the featurevector is sign quantized so that any positive value isrepresented by 1 and negative value as 0. Thisresults in a compact biometric template consisting ofonly 87 bits.Storing and Matching: Now this is our final phasefor completing our system. Here we will store the 87bit iris code or template in our database for futurematching and this matching is done with the help ofan efficient matching algorithm here we are usingHamming Distance algorithm for the recognition oftwo samples that is reference template andenrollment template. It is basically uses an exclusiveOR (XOR) function between two bit patterns.Hamming Distance is a measure, which delineatethe differences, of iris codes. Every bit of presentediris code is compared to the every bit of referencediris code, if the two bits are the same, e.g. two 1‟s ortwo 0‟s, the system assigns a value „0‟ to thatcomparison and if the two bits are different, thesystem assigns a value „1‟ to that comparison. Theformula for iris matching isshown as follows:HD =1/N∑ (Pi ⊕ Ri) where N is the dimension of feature vector, Pi is theith component of the presented feature vector whileRi is the ith component of the referenced featurevector.Here we can set the threshold value if the resultantvalue is greater than threshold value, then reject theiris else accepted the user iris.As per review I am giving my thought in matchingprocess which can be represented by a diagramnamed storing and matching process in figure: 12.Figure 12: Storing and matching processIV. ConclusionIn this review paper I show how a person can beidentified by a number of ways but instead ofcarrying bunk of keys or remembering things aspasswords we can use us as living password, whichis called biometric recognition technology it usesphysical characteristics or habits of any person foridentification. In biometrics we have a number ofcharacteristics which we are using in our recognitiontechnology as fingerprint, palm print, signature,face, iris recognition, thumb impression and so onbut among these irises recognition is besttechnology for identification of a person. I can saythat this technology is not completely developed andwe need a number of scientists, researchers anddeveloper who can work on this technology and cancomplete the dream of Mr. Daugman by applyingthe uses of iris recognition in each and every fieldwhere security is needed by the human being.References http://www.amazingincredible.com/show/82-the-incredible-human-eye (last referredon 26 April 2013) P.W. Hallinan, “Recognizing HumanEyes,” Geomtric Methods Comput.Vision,vol. 1570, pp.214-226, 1991. J. Daugman and C. Downing, “EpigeneticRandomness, Complexity, and Singularityof Human Iris Patterns,” Proc. Royal Soc.(London) B: Biological Sciences, vol. 268,pp. 1737-1740, 2001. R. Wildes, J. Asmuth, G. Green, S. Hsu, R.Kolczynski, J. Matey, S. McBride, “AMachine-vision System for IrisRecognition,” vol. 9, pp.1-8, 1996. R.P. Wildes, “Iris Recognition: AnEmerging Biometric Technology,”Proceedings of the IEEE, Sept. vol.85,pp.1348-1363, 1997. http://www.cl.cam.ac.uk/~jgd1000/applics.html (last referred on 26 April 2013)
Himanshu Srivastava / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.449-452453 | P a g e http://www.forbes.com/global/2010/0412/entrepreneurs-hector-hoyos-iris-scanners-identity-we-see-you.html (last referred on26 April 2013) http://pinktentacle.com/2006/11/iris-recognition-technology-for-mobile-phones/(last referred on 26 April 2013) http://www.freedomsphoenix.com/News/081163-2010-12-31-were-next-napolitano-in-afghanistan-to-see-iris-scanning-at.htm L. Flom and A. Safir, “Iris RecognitionSystem,” U.S. Patent, No. 4641394, 1987. A. Bertillon. La couleur de Piris. “Revuescientifique,” vol. 36, no. 3, pp. 65-73,1885. J.G. Daugman, “High Confidence VisualRecognition of Persons by a Test ofStatistical Independence,” IEEE Trans.Pattern Anal. Mach. Intelligence, vol. 15,no.11, pp.1148-1161, 1993. W. Boles and B. Boas hash, “A HumanIdentification Technique Using Images ofthe Iris and Wavelet Transform,” IEEETrans. on Signal Processing, vol. 46, no. 4,pp.1185-1188, 1998. Kevin W. Bowyer, Karen Hollingsworth,Patrick J. Flynn, “Image Understanding forIris Biometrics: A Survey, ComputerVision and Image Understanding,” Vol.110, Issue 2, pp. 281-307, 2008. http://www.idealtest.org/dbDetailForUser.do?id=4 (last referred on 26 April 2013) http://pesona.mmu.edu.my/~ccteo/ (lastreferred on 26 April 2013) Li Ma, Yunhong Wang, Tieniu Tan, “IrisRecognition Using Circular SymmetricFilters,” National Laboratory of PatternRecognition, Institute of Automation,Chinese Academy of Sciences, P.R.China. C.Tisse L. Martin L. Torres and M. Robert,“Person Identification Technique UsingHuman Iris Recognition," Proc. VisionInterface, pp. 294-299, 2002. S. Lim, K. Lee, O. Byeon, T. Kim,“Efficient iris recognition throughimprovement of feature vector andclassifier,” ETRI Journal, Korea, vol.23, no. 2, 2001.