Iris Identification using Multiclass Support Vector Machine based on Five Regions for Iris Segmentation

Abstract

Personal identification using Iris has been stabilized as efficient technology. In this paper, a new framework of iris segmentation based on five regions has been employed. A set of features will be deriving from these regions after treatment them by different filters. Finally, the derived features will be used to identify each iris using multiclass support vector machine. The proposed system used only 50% of Iris information to extract the features of it. The system performs with an identification rate 97% for training phase and 91% for test phase on 350 images (50 person, 7 images for each person) from CASIA V 1.0 database.