Fusion of Sclera and Periocular Features for Biometric System

This paper proposes the biometric verification system based on ocular features. We form the multimodal biometric system considering two recent biometric traits in ocular region- sclera region and periocular region. For feature extraction of sclera part we use simple technique which eliminates the expensive image enhancement process i.e Local Binary Pattern (LBP) and the matching scores are generated. For feature extraction of periocular region we use structured random projections and matching score are generated. From these matching scores the score level fusion is done with Extreme Learning Machine (ELM). This method has shown 94.40% of accuracy.