An Improvement of Iris Pattern Identification Using Radon Transform

Article Sidebar

PDF
DOI: https://doi.org/10.37936/ecti-cit.200731.54212
Keywords:
Iris Feature extraction Radon transform

Main Article Content

Pattraporn Ariyapreechakul
Department of Computer Engineering, Faculty of Engineering Chulalongkorn University, Bangkok 10330, Thailand
Nongluk Covavisaruch
Department of Computer Engineering, Faculty of Engineering Chulalongkorn University, Bangkok 10330, Thailand

Abstract

This research proposes an improvement of iris personal identification system using Radon transform which is used to extract feature stored as templates in a database. In our research, the system has been tested to find the best system performance with 3 distance functions: regional correlation, Euclidean distance and absolute distance. The test images are from CASIA iris database. Our experimental results reveal that the improved approach greatly increases the system performance for all distance functions. Among the 3 distance functions, the absolute distance gives the best results which reduce EER from 18% (previous approach) to 3.695%.

Article Details

How to Cite
[1]
P. Ariyapreechakul and N. Covavisaruch, “An Improvement of Iris Pattern Identification Using Radon Transform”, ECTI-CIT Transactions, vol. 3, no. 1, pp. 45–50, Apr. 2016.
Issue
Vol. 3 No. 1 (2007): ECTI Transaction on CIT (May 2007)
Section
Artificial Intelligence and Machine Learning (AI)