Iris identification based on log Gabor filtering

An accurate biometric identification system is a critical requirement in a variety of applications. Iris-based identification has been receiving a lot of attention since its introduction. However, some techniques have limitations in identifying persons accurately and efficiently. In this work, a procedure that captures local and global characteristics of the iris, using a band of log-Gabor filters is proposed. We call the vector that stores these characteristics a feature descriptor or simply the iris descriptor. Our experimental results demonstrate that log-Gabor filters identification capabilities outperform other procedures based on the basic Gabor filters.