Latent fingerprint and Iris fusion for enhancement of performance of human identification system

Abstract

The biometric recognition system is a highly precise human identification system as per Federal Bureau of Investigation (FBI). Latent fingerprint biometrics are an efficient human identification system for criminals based on available crime evidence shreds. However, biometric trait limitations such as noise in sensed data, lack of individuality, and intra-class variation result in a low matching score, which has a negative impact on the identification as well as investigation process. The increasing demand for biometric systems for accurate identification has resulted in the development of unimodal biometric systems to multimodal biometric systems, particularly when the biometric inputs are considered as evidence of any societal crime. This paper focuses on showing the advantage of using modal based approach over the threshold based and development of a multimodal biometric framework for achieving high accuracy in human identification. The comparison of the identification rate/recognition probability of modal-based approach (87.3%) is high in compared to threshold-based approach (80%). Latent fingerprint biometric is fused with iris biometric using score level fusion (sum rule and product rule) rules. With the experimental results, it has been found that highest accuracy rate (91.15%) in proposed identification framework can be achieved while using sum rule. The proposed framework is useful in a societal crime scenario, as assumed in this paper.