Iris Feature Extraction for Personal Identification Using Lifting Wavelet Transform

Iris recognition, as an emerging biometric recognition approach has become a major research topic with practical applications in recent years as it promises nearly perfect recognition rates. In this paper, a novel, efficient approach for iris recognition is presented. The goal is to develop a lifting (integer) wavelet based algorithm that enhances iris images, reduces noise to the maximum extent possible, and extracts the important features from the image. The similarity between test and training iris images is estimated using some standard distance measures and comparison of threshold. The proposed technique is computationally effective with recognition rate of 99.97 % on the standard CASIA iris database.