i A NEW APPROACH FOR HUMAN IDENTIFICATION USING THE EYE A Thesis Submitted to the Faculty of Purdue University by N. Luke Thomas In Partial Fulfillment of the Requirements for the Degree of Master of Science in Electrical and Computer Engineering May 2010 Purdue University Indianapolis, Indiana ii ACKNOWLEDGMENTS First and foremost, I would like to gratefully acknowledge my thesis advisor, Dr. Yingzi (Eliza) Du, for her guidance, supervision, and knowledge during my course of study, research, and thesis work. I would like to express my gratitude to my thesis committee members, Dr. Maher Rizkalla and Dr. Brian King, for their time and helpful comments. I thank Ms. Valerie Lim Diemer, Ms. Jane Simpson, and Ms. Sherrie Tucker for their constructive suggestions and help formatting this thesis. I also would like to sincerely thank my lab mates Mr. Craig Belcher, Mr. Zhi Zhou, Mr. Tron Artavatkun, and Mr. Kai Yang for the technical discussions, their help and support in collecting data for this thesis research, and friendship. I would like to express my appreciation to those subjects who contributed their biometric information for use in the IUPUI multi-wavelength database. I would also like to acknowledge the Department of Computer Science at the University of Beira Interior for providing the UBIRIS database. This work was partially funded by the ONR Young Investigator Award #: N00014-07-1-0788. I would like to thank the ONR for their generous financial support. Finally, and most importantly, I would like to thank my wife for her support, dedication, and love. She has graciously accepted the difficulties and problems that have accompanied my graduate studies. Without her support and inspiration, this work would not have happened. iii TABLE OF CONTENTS Page LIST OF TABLES ............................................................................................................. vi LIST OF FIGURES .......................................................................................................... vii ABSTRACT… ................................................................................................................... xi 1. INTRODUCTION AND LITERATURE REVIEW .....................................................1 1.1 Biometrics ..............................................................................................................1 1.2 Iris Recognition ......................................................................................................2 1.2.1 Cooperative Iris Recognition .......................................................................2 1.2.2 Non-Cooperative Iris Recognition ...............................................................4 1.2.3 Visible Wavelength Iris Recognition ...........................................................5 1.2.4 Multiple Wavelength Iris Recognition.........................................................6 1.2.5 Limitations of Iris Recognition ....................................................................6 1.3 Retina Recognition.................................................................................................7 1.4 Sclera Recognition .................................................................................................7 1.4.1 The Sclera and Conjunctival Vasculature ....................................................7 1.4.2 Advantages of Sclera Recognition ...............................................................8 1.4.3 Previous Works on Recognition in the Sclera Region .................................9 1.4.4 Sclera Recognition Databases ....................................................................11 1.4.5 Sclera Recognition Challenges ..................................................................17 2. OVERVIEW OF PROPOSED SYSTEM ....................................................................20 2.1 Background of Vein Recognition ........................................................................20 2.2 Vein Recognition vs. Sclera Recognition ............................................................20 2.3 Proposed Method .................................................................................................21 3. SCLERA SEGMENTATION ......................................................................................23 3.1 Review of Iris Image Segmentation Methods ......................................................23 3.2 Review of Previous Sclera Segmentation Methods .............................................24 3.3 Proposed Sclera Segmentation Method ...............................................................26 3.3.1 Downsampling the Image ..........................................................................27 3.3.2 Conversion to HSV Colorspace .................................................................27 3.3.3 Estimation of Sclera Area ..........................................................................28 3.3.4 Convex Hull of Estimated Sclera ...............................................................29 iv Page 3.3.5 Color-Based Sclera Region Estimation Scheme ........................................29 3.3.6 Iris Segmentation .......................................................................................30 3.3.7 Iris and Eyelid Refinement ........................................................................32 3.3.8 Creation of the Sclera Mask and Upsampling the Sclera Mask ...........................................................................................................37 3.4 Sclera Segmentation Results ................................................................................38 3.4.1 Challenging Cases for Sclera Segmentation ..............................................40 3.5 Summary ..............................................................................................................42 4. FEATURE ENHANCEMENT AND EXTRACTION ................................................44 4.1 Review of Image Enhancement Techniques ........................................................44 4.1.1 Contrast Stretching.....................................................................................44 4.1.2 Histogram Equalization .............................................................................45 4.2 Review of Previous Sclera Feature Enhancement and Extraction Techniques ............................................................................................................46 4.3 Proposed Sclera Feature Enhancement and Extraction Method ..........................47 4.3.1 Gabor Filter Vein Enhancement ................................................................47 4.3.2 Vascular Pattern Extraction: Line Segment Descriptor .............................51 4.4 Sclera Feature Enhancement and Extraction Results...........................................53 4.4.1 Challenging Cases in Sclera Feature Extraction and Enhancement… ..........................................................................................53 4.5 Summary ..............................................................................................................56 5. SCLERA MATCHING 5.1 Review of General Matching Schemes ................................................................57 5.1.1 Hamming Distance.....................................................................................57 5.1.2 Euclidean Distance.....................................................................................57 5.1.3 Spectral Angle Measure .............................................................................58 5.1.4 Information Distance .................................................................................58 5.2 Review of Previous Sclera Matching Techniques ...............................................59 5.3 Proposed Sclera Matching Technique .................................................................60 5.3.1 Sclera Template Registration .....................................................................60 5.3.2 Sclera Template Matching .........................................................................62 5.4 Sclera Matching Results ......................................................................................64 5.4.1 Example Individual Matching Result ........................................................64 5.4.2 UBIRIS Database Matching Results ..........................................................69 5.4.3 Matching Results for IUPUI Multiple Wavelength Database .....................................................................................................77 5.4.4 Challenging Cases for Sclera Matching .....................................................79 5.5 Summary ..............................................................................................................85 6. COMPARISON TO IRIS RECOGNITION IN THE VISIBLE WAVELENGTHS .......................................................................................................87 6.1 Comparison of Results .........................................................................................87 v Page 6.1.1 Comparison with UBIRIS Database ..........................................................87 6.1.2 Comparison with IUPUI Multiple Wavelength Database .........................88 6.2 Summary ..............................................................................................................89 7. CONCLUSION ............................................................................................................90 7.1 Future