నైరూప్య

PERFORMANCE ANALYSIS OF IRIS AUTHENTICATION SYSTEM IN WAVELET DOMAIN

P. S. Revankar1, Anisa Anjum

Biometric technologies are considered more secure with respect to other authentication techniques based on password and ID card. Most reliable biometric technology in terms of identification and verification is based on iris. Iris recognition nevertheless is computationally very costly and complicated system. Feature extraction is a significant task in the overall processing of iris recognition system. Most discerning information present in an iris must be extorted to provide accurate recognition of individuals. A feature extraction technique is believed valuable if it decreases the size of the feature template and computational speed as well as provides high accuracy. This paper evaluates performance of basis wavelets for iris feature and optimal wavelet transform is determined among Haar, Daubechies, Symlet and Reverse Biorthogonal wavelets. Optimal performance is acquired with db10 wavelet. With proposed iris authentication system FAR, FRR and size of feature template is reduced up to 0%, 0.62%, and 1216 bits respectively.