Biometrics Sensor Fusion

Abstract

Performance of any biometric system entirely depends on the information that is acquired from biometrics characteristics (Jain et. al., 2004). Several biometrics systems are developed over the years in the last two decades, which are mostly considered as viable biometric tools used for human identification and verification. However, due to some negative constraints that are often associated with the biometrics templates are generally degraded the overall performance and accuracy of the biometric systems. In spite of that, many biometrics systems are developed and implemented over the years and deployed successfully for user authentication. Modality based categorization of the biometric systems are made on the basis of biometric traits are used. While single biometric systems are used for verification or identification of acquired biometrics characteristics/attributes, it is called uni-biometrics authentication systems and when more than one biometric technology are used in fused form for identification or verification, it is called multimodal biometrics. It has been seen that, depending on the application context, mono-modal or multimodal biometrics systems can be used for authentication. In biometric, human identity verification systems seek considerable improvement in reliability and accuracy. Several biometric authentication traits are offering ‘up-to-the-mark’ and negotiable performance in respect of recognizing and identifying users. However, none of the biometrics is giving cent percent accuracy. Multibiometric systems remove some of the drawbacks of the uni-modal biometric systems by acquiring multiple sources of information together in an augmented group, which has richer details. Utilization of these biometric systems depends on more than one physiological or behavioral characteristic for enrollment and verification/ identification. There exist multimodal biometrics (Jain et. al., 2004) with various levels of fusion, namely, sensor level, feature level, matching score level and decision level. Further, fusion at low level / sensor level by biometric image fusion is an emerging area of research for biometric authentication. A multisensor multimodal biometric system fuses information at low level or sensor level of processing is expected to produce more accurate results than the systems that integrate 17