Iris recognition using 2-D elliptical-support wavelet filter bank

Abstract

In this paper, a new identification method for iris recognition is presented. Among the four main steps of iris recognition, traditional segmentation and normalization steps are utilized in the proposed method. A non-traditional step for feature extraction is applied where a new bank of two-dimensional (2-D) elliptical-support wavelet Haar filter bank is used to capture the iris characteristics. The idea is based on a new geometrical image transform called 2-D elliptical-support wavelet transform (2-D ESWT). A five-level 2-D elliptical-support wavelet decomposition is needed to form a reduced fixed length quantized feature vector with improved performance. The efficient approach of Hamming distance is then applied as a final step for iris matching. Experimental results show that the proposed method is reliable with rapid recognition, since it achieves good recognition rate with reduced feature vector length. Thus, a less complex-implementation can be obtained for this identification method.