SPRINGER BRIEFS IN ELECTRICAL AND COMPUTER ENGINEERING  SIGNAL PROCESSING Amol D. Rahulkar Raghunath S. Holambe Iris Image Recognition Wavelet Filter- Banks Based Iris Feature Extraction Schemes 123 SpringerBriefs in Electrical and Computer Engineering Signal Processing Series editors Woon-Seng Gan, Singapore C.-C. Jay Kuo, Los Angeles, USA Thomas Fang Zheng, Beijing, China Mauro Barni, Siena, Italy For furthervolumes: http://www.springer.com/series/11560 Amol D. Rahulkar Raghunath S. Holambe • Iris Image Recognition Wavelet Filter-Banks Based Iris Feature Extraction Schemes 123 AmolD.Rahulkar Raghunath S.Holambe Department of InstrumentationEngineering Department of Instrumentation AISSMS’ InstituteofInformation SGGS InstituteofEngineering and Technology Technology Pune,Maharashtra Nanded, Maharashtra India India ISSN 2196-4076 ISSN 2196-4084 (electronic) ISBN 978-3-319-06766-7 ISBN 978-3-319-06767-4 (eBook) DOI 10.1007/978-3-319-06767-4 Springer ChamHeidelberg New YorkDordrecht London LibraryofCongressControlNumber:2014938479 (cid:2)TheAuthor(s)2014 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionor informationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purposeofbeingenteredandexecutedonacomputersystem,forexclusiveusebythepurchaserofthe work. Duplication of this publication or parts thereof is permitted only under the provisions of theCopyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the CopyrightClearanceCenter.ViolationsareliabletoprosecutionundertherespectiveCopyrightLaw. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexempt fromtherelevantprotectivelawsandregulationsandthereforefreeforgeneraluse. While the advice and information in this book are believed to be true and accurate at the date of publication,neithertheauthorsnortheeditorsnorthepublishercanacceptanylegalresponsibilityfor anyerrorsoromissionsthatmaybemade.Thepublishermakesnowarranty,expressorimplied,with respecttothematerialcontainedherein. Printedonacid-freepaper SpringerispartofSpringerScience+BusinessMedia(www.springer.com) To our families For their love, encouragement and support Preface The increasing demand onenhanced security has led toan unprecedented interest in automated personal recognition based on biometric system. The biometric system makes use of physiological or behavioral characteristics to recognize individuals. Among all the biometric recognition systems, iris recognition system hasbeen deployedinvariouscriticalapplicationareas (homeland security, border control, web-based services, national ID cards, etc.) because of its unique, stable, and noninvasive characteristics. It is observed that iris image consists of nonuni- form spectral information due to its irregular and random characteristics (tiny crypts, freckles, radial furrows, radial streaks, collarette, pigment spots, etc.). Multi-resolution analysis (MRA)-based technique can be well suited to represent these iris image structures. It is well known that discrete wavelet transform (DWT) is apowerful toolin MRA. Thepowerof DWT is to offer high temporal localization for high frequencies and good frequency resolution for low frequencies. Most of the iris image representation schemes in the literature used off-the-shelf wavelet basis to extract the features. Although there is a defined standard for raw iris data, there is none regarding iris feature representation. Thus, many issues are still open in the field of iris image feature extraction related to the choice offilter bank (FB). The design of FBs and investigations of their properties (near-orthogonality, regularity, time-frequency localization, linear phase, perfect reconstruction, etc.) for image-coding, denoising, com- pression, etc., have been carried out by many researchers. However,effectiveness of the properties in iris pattern recognition has not been addressed in the liter- ature. Several nonideal factors (eyelids occlusion, multiple/separable eyelashes occlusion, reflection (specular, lighting), poor focus, partially opened eyes, motion blur, noncircular shaped of the iris/pupil, etc.) contained in iris images can increase the false rejection rate (FRR). This book focused on the design of critically sampled separable and nonsepa- rable wavelet filter banks (FBs) for effective iris image representation. These systemsareimportantforfeatureextractionalgorithmsduetotheirnonredundancy (critical sampling). In addition, k-out-of-n:A post-classifier is explored to reduce the FRR. Due to the desired properties of these designed FBs like flexible fre- quency response, near-orthogonality, and regularity, the filter banks designed in this book can be more effectively used than the existing FBs in many signal processing applications like pattern classification, data-compression, vii viii Preface watermarking,denoising,etc.Inthisbook,wehave evaluatedtheperformanceof the designed FBs forextraction offeaturesofthe iris. However,theseFBs can be used effectively toextract features fromface,fingerprint, palm-print, ear, etc., for automatic person verification (identification). Abriefintroductiontobiometricsingeneralandirisinparticularispresentedin Chap. 1. The motivation along with a brief review of the previously published related work (Iris recognition algorithms, one-dimensional filter-banks, and two-dimensional filter-banks) is also presented in this chapter. Chapter 2 explains the design of a new class of triplet half-band filter bank (THFB).ThepropertiesoftheproposedclassofTHFBareinvestigatedtoextract the discriminating iris features. The details of THFB-based feature extraction processincludingpost-classifierareexplainedinthischapter.Wealsoprovidethe experimental results to show the effectiveness of the proposed technique. In Chap. 3, a nonseparable, nonredundant, multiscale combined directional wavelet filter bank (CDWFB) is constructed by the combination of directional waveletfilterbank(DWFB)androtateddirectionalwaveletfilterbank(RDWFB). This chapter also discusses the iris feature extraction algorithm based on a combination of CDWFB and post-classifier. Experimentation is carried out to evaluate the performance of the schemes. Chapter 4 explains the iris feature extraction scheme based on 2-D nonsepa- rable, nonredundant, multiscale hybrid finer directional wavelet filter bank and classification using fused post-classifier under nonideal environmental conditions. ThischapteraddressestheissueinthedesignofDWFBandextendstheproposed class of THFB for the 2-D nonseparable filter bank. Simulation results for the proposed algorithm are also presented in this chapter. Chapter 5 addresses the issue in the design of proposed nonseparable FBs and presentsthedesignofthenewclassoftriplethalf-bandcheckerboardshapedfilter bank (THCSFB). This chapter also describes the directional ordinal measures (DOMs) for iris image representation using the designed class of THCSFB. The experimental results are provided to demonstrate the performance of this method. This book provides the new results in wavelet filter banks-based feature extraction, and the classifier in the field of iris image recognition. It provides the broad treatment on the design of separable, nonseparable wavelet filter banks. It brings together the three strands of research (wavelets, iris image analysis, and classifier). This book contains the compilation of basic material on the design of waveletsthatavoidsreadingmanydifferentbooks.Thematerialonseparableand nonseparable wavelet design has been reorganized significantly so to provide an easier path for newcomers and researchers to master the contents. Acknowledgments We are grateful to many teachers, colleagues, and researchers, who directly or indirectlyhelpedusinpreparingthisbook.WearethankfultoDr.L.M.Waghmare, Director, Shri Guru Gobind Singhji Institute of Engineering and Technology, NandedandDr.P.B.Mane,Principal,Prof.H.P.Chaudhari,Head,Departmentof Instrumentation Engineering, AISSMS Institute of Information Technology, Pune for their motivation and constant support while preparing the manuscript. WearealsothankfultoDr.B.M.Patre,Dr.R.H.Chile,Dr.S.T.Hamde,Dr.V.G. Asutakar,Prof.R.S.Jamkar,Dr.V.S.Thool,Dr.R.V.Sarwadnya,andMr.R.P. Borse.Wewouldliketoacknowledgeourcolleagues,whohaveinvolvedindirectly withthiswork,Dr.D.V.Jadhav,Dr.N.S.Nehe,Dr.M.S.Deshpande,Dr.P.K. Ajmera,Mr.J.P.Gawande,Mrs.S.P.Madhe,andMr.S.S.Gajbhar.Weoweour lovingthankstoourstaffmembersMr.B.D.UpareandMr.A.J.Shindefortheir valuablesupport.Finally,wewishtoacknowledgeDr.ChristophBaumann,Editor, andSathyaSubramaniamfortheirunusuallygreathelpandeffortsduringtheperiod ofpreparingthemanuscriptandproducingthebook. Amol D. Rahulkar Raghunath S. Holambe ix Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Biometrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Requirement of Biometrics Systems. . . . . . . . . . . . . . . . . . . . . 2 1.3 Iris as a Biometric. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.4 Strengths and Weaknesses of the Iris as a Biometric. . . . . . . . . 4 1.4.1 Strengths of Iris Biometric. . . . . . . . . . . . . . . . . . . . . . 4 1.4.2 Weaknesses of Iris Biometric. . . . . . . . . . . . . . . . . . . . 5 1.5 Performance Measures of Iris Recognition System . . . . . . . . . . 5 1.6 Nonideal Iris Recognition: A New Challenge . . . . . . . . . . . . . . 5 1.7 A Brief Review On . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.7.1 Iris Recognition Algorithms. . . . . . . . . . . . . . . . . . . . . 7 1.7.2 Two-Channel One-Dimensional Filter Banks . . . . . . . . . 13 1.7.3 Two-Dimensional Filter Banks. . . . . . . . . . . . . . . . . . . 15 1.8 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 1.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2 Features Based on Triplet Half-Band Wavelet Filter-Banks. . . . . . 23 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.2 Review of the Related Filter Banks. . . . . . . . . . . . . . . . . . . . . 25 2.2.1 Triplet Halfband Filter Bank . . . . . . . . . . . . . . . . . . . . 27 2.2.2 Factorization Based on a Generalized Half-Band Polynomial. . . . . . . . . . . . . . . . . . . . . . . . . 28 2.3 Design of New Class of THFB. . . . . . . . . . . . . . . . . . . . . . . . 29 2.3.1 Design Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.3.2 Properties of the Designed THFB Desirable for Iris Feature Extraction . . . . . . . . . . . . . . . . . . . . . . 32 2.4 Iris Recognition Algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.4.1 Feature Extraction Using a New Class of THFB. . . . . . . 34 2.4.2 Design of k-out-of-n:A Post-classifier for Iris Recognition. . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.5 Experimental Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 xi xii Contents 3 Combined Directional Wavelet Filter-Banks Based Features . . . . . 45 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 3.2 Review of the Related Directional Filter Bank . . . . . . . . . . . . . 45 3.3 Construction of the Directional Filter Bank . . . . . . . . . . . . . . . 46 3.3.1 Design of 1-D Biorthogonal Wavelet FB Using Factorization of an HBP. . . . . . . . . . . . . . . . . . . 46 3.3.2 Construction of 2-D Separable Filter Bank. . . . . . . . . . . 48 3.3.3 Construction of Fan Shaped Filter Bank . . . . . . . . . . . . 49 3.3.4 Construction of Directional Wavelet Filter Bank. . . . . . . 50 3.3.5 Construction of Rotated Directional Wavelet Filter Bank. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 3.4 Feature Extraction Using DWFB and RDWFB . . . . . . . . . . . . . 53 3.5 Experimental Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4 Iris Representation by Combined Hybrid Directional Wavelet Filter-Banks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 4.2 Review of the Related Filter Banks. . . . . . . . . . . . . . . . . . . . . 60 4.3 Design of Combined Hybrid Directional Wavelet FB. . . . . . . . . 60 4.3.1 Construction of 2-D Separable Filter Bank. . . . . . . . . . . 60 4.3.2 Design of the Triplet Halfband Fan Shaped Filter Bank. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.3.3 1-D to 2-D Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.4 Iris Feature Extraction Using CHDWFB. . . . . . . . . . . . . . . . . . 64 4.5 Experimental Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 5 Ordinal Measures Based on Directional Ordinal Wavelet Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 5.2 Review of the Related Filter Banks. . . . . . . . . . . . . . . . . . . . . 70 5.3 Design of Triplet Halfband Checkerboard Shaped Filter Bank (THCSFB). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 5.3.1 Design of Triplet 1-D Framework from Generalized HBPs. . . . . . . . . . . . . . . . . . . . . . . . 71 5.3.2 1-D to 2-D Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . 72 5.3.3 Properties of the 1-D to 2-D Mapping. . . . . . . . . . . . . . 73 5.3.4 Directional Extension of Wavelet Filter Bank. . . . . . . . . 73