Comparative study of different window sizes setting in median filter for off-angle iris recognition

Iris recognition is one of the most popular biometric recognition that has increased in the number of acceptance user gradually because of the reliability and accuracy provided by this system. However, this accuracy is highly correlated with the quality of iris image captured. Thus, a poor quality o...

Full description

Saved in:
Bibliographic Details
Main Authors: Hassan, R., Kasim, S., Esa, N. M., Zakaria, Z.
Format: Article
Language:English
Published: Insight Society 2017
Subjects:
Online Access:http://eprints.utm.my/id/eprint/81231/1/RohayantiHassan2017_ComparativeStudyofDifferentWindowSizes.pdf
http://eprints.utm.my/id/eprint/81231/
http://dx.doi.org/10.18517/ijaseit.7.5.3393
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Iris recognition is one of the most popular biometric recognition that has increased in the number of acceptance user gradually because of the reliability and accuracy provided by this system. However, this accuracy is highly correlated with the quality of iris image captured. Thus, a poor quality of the image captured required an enhancement technique. This study aims to identify the optimum window size for the median filter. Identifying the optimum window size setting required template matching value result of the off-angle iris recognition. The lowest value obtained showed that the window size applied was optimized. The result of this study demonstrated, for WVU-OA dataset for 15 degrees off-angle iris of right and left eyes, the window size of [5 5] and [7 7] respectively are optimum to maximize the median filter function. Meanwhile, for 30 degrees off-angle iris of right and left eyes data, the optimum windows size proposed are [7 7] and [5 5] respectively. On the other hand, analysis using UBIRIS dataset showed that the optimum window size for 30 degrees off-angle iris, both right and left eye is [7 7] which is able to maximize the performance of the median filter. In conclusion, the effective value to be applied to all dataset are [5 5] and [7 7] because in most cases it provides a better template matching compared to without applying the filtering method.