Development of an iris authentication algorithm for personal identification /
Biometric systems differentiate people based on their uniquely characteristics manner. Among various biometric systems, iris recognition provides most reliable identification. In recent years, the development and practice of the field of iris recognition has expanded dramatically. Now it becomes a p...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Kuala Lumpur :
Kulliyyah of Engineering, International Islamic University Malaysia,
2015
|
| Subjects: | |
| Online Access: | http://studentrepo.iium.edu.my/handle/123456789/4581 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Biometric systems differentiate people based on their uniquely characteristics manner. Among various biometric systems, iris recognition provides most reliable identification. In recent years, the development and practice of the field of iris recognition has expanded dramatically. Now it becomes a practical area of science and technology. The developments of core algorithm increase its practical applications. The research regarding iris recognition is not only focusing on ideal image where camera uses infrared illumination but also focusing on non-ideal image which has been taken in presence of visible lighting. It takes lot of user cooperation to capture an ideal image which makes the system time consuming. To make the system more user friendly, the algorithm to handle non-ideal image is essential. The main aim of this research work is to develop an algorithm which can locate iris from both ideal image and non-ideal image. Three major steps of the iris recognition system are localization of iris, normalization of iris and feature extraction of iris. The Hough Transform and image thresholding technique has been applied to localize iris in a given eye image. The Hough Transform shows excellent performance to localize iris in an ideal image. However, Hough Transform fails to perform accurate localization for non-ideal image. On the other hand, image thresholding techniques show relatively good performance for both ideal and non-ideal image. The isolated iris region is then transformed from Cartesian to polar form by using Daugman intrego differential operator. Finally to encode the feature into a binary template 1D Log-Gabor filter has been used. A simple Boolean Exclusive-OR operator (XOR) function has been applied to check whether two binary templates are from same image or not. To validate the performance of the algorithm both ideal and non-ideal eye images have been used. Image from CASIA Iris Interval database has been used to validate the performance of algorithms for ideal image and image from UBIRIS database has been used to validate the performance of algorithms for non-ideal image. On a set of 138 different combinations, the algorithm shows 0% false acceptance rate. However, observation on 94 same class variations shows 4.25% false rejection rate. Therefore, the iris recognition algorithm proves to be a consistent and precise biometric technology. |
|---|---|
| Physical Description: | xvi, 90 leaves : ill. ; 30cm. |
| Bibliography: | Includes bibliographical references (leaves 81-89). |