Comparison and combination of iris matchers for reliable personal identification

The biometric identification approaches using iris images are receiving increasing attention in the literature. Several methods for the automated iris identification have been presented in the literature and those based on the phase encoding of texture information are suggested to be the most promising. However, there has not been any attempt to combine these phase preserving approaches to achieve further improvement in the performance. This paper presents a comparative study of the performance from the iris identification using log-Gabor, Haar wavelet, DCT and FFT based features. Our experimental results suggest that the performance from the Haar wavelet and log Gabor filter based phase encoding is the most promising among all the four approaches considered in this work. Therefore the combination of these two matchers is most promising, both in terms of performance and the computational complexity. Our experimental results from the all 411 users (CASIA v3) and 224 users (IITD v1) database illustrate significant improvement in the performance that is not possible with either of these approaches individually.