Efficient Iris Recognition System Using Relational Measures

This paper proposes an efficient iris based authentication system. The segmented iris is unwrapped, normalized and enhanced using the proposed local enhancement technique. Occlusion mask determination is performed to detect eyelid, eyelashes and reflections using morphological and filtering operations. Features are extracted and matched from enhanced image using relative intensities of regions and encoding them into a binary template. The proposed recognition approach has obtained a CRR of 99.07% on CASIA-4.0 Interval, 98.7% on CASIA-4.0 Lamp and 98.66% on IITK database. It has also achieved an EER of 1.82% on CASIA-4.0 Interval, 4.2% on CASIA-4.0 Lamp and 2.12% on IITK database.