Face-Iris Multimodal Biometric System Based on Hybrid Level Fusion

Abstract

Multimodal biometric has become an interesting research topic over the past recent years. Its principle is the recognition of a person by fusing various behavioral and/or physical traits or fusing several unimodal biometric systems by using different levels of fusion. This paper proposes a face-iris multimodal biometric system based on hybrid level fusion. The proposed features are first extracted from the face and two iris (left and right) modalities and are based on 2D-log Gabor filter. SRKDA is then applied on the extracted features set in order to select the pertinent and discriminate ones. The proposed system is evaluated on CASIA Iris Distance database; and the experiment results obtained have shown its performance in the verification mode compared to existing state-of-the-art systems.