Abstract: Multimodal biometric system fuses the data presented by multiple biometric sources, hence offering better performance than unimodal biometric system. In multimodal biometric systems, failure of any one trait may not seriously affect the person authentication as other trait can successfully work. Fusion of multiple modalities can take place at different levels i.e feature level, score level or decision level. But the data acquired from different modalities may be heterogeneous. So, there is a need to normalize the data into common domain. This paper presents the overview of various fusion and normalization techniques used in the biometric systems. Also, analyze the performance of different normalization techniques with various fusion techniques.
Keywords: Fusion techniques, multimodal biometric, normalization techniques.