Gorazd Vrček (2008) Iris-based human verification system. EngD thesis.
Abstract
The biometric person autentification technique based on the pattern of the human iris is well suited to be applied to any access control system requiring a high level of security. A sy stem for personal verification based on iris patterns is presented in this thesis. In the introductory part, I presented the problem of biometric systems and the accuracy of their procedures, required knowledge of the anatomy of the human eye, what is verification based on the iris patterns and algorithm operating principle. In the middle part, I presented the verification algorithm, which consists of capturing images from the outside world, image processing to obtaining iris information, iris normalization and feature extraction, person classification or verification. Gabor filter was used for feature extraction from whtch iris bit template sequence is encoded. Hamming distance is calculated from iris template, which gives the estimate of the mach in the verification process. The last part focuses on the results presented by the verification algorithm. Results have been carried out on CASIA picture database, where a decision threshold for Hamming distance was set to 0,427, which gives 0-percent false acceptance rate and 11.584-percent ralse reject rate on tested group.
Item Type: | Thesis (EngD thesis) |
Keywords: | verification, biometric systems, eye anatomy, CASIA, iris, feature extraction, Gabor filter, Gabor wawelets, Hamming distance, implementation |
Number of Pages: | 69 |
Language of Content: | Slovenian |
Mentor / Comentors: | Name and Surname | ID | Function |
---|
doc. dr. Peter Peer | 294 | Mentor |
|
Link to COBISS: | http://www.cobiss.si/scripts/cobiss?command=search&base=50070&select=(ID=6863444) |
Institution: | University of Ljubljana |
Department: | Faculty of Computer and Information Science |
Item ID: | 793 |
Date Deposited: | 17 Dec 2008 12:13 |
Last Modified: | 13 Aug 2011 00:34 |
URI: | http://eprints.fri.uni-lj.si/id/eprint/793 |
---|
Actions (login required)