Features-Level Fusion of Face and Handwritten Signature in Multimodal Biometric Identification System

摘要

Biometrics is an emerging tool used to identify humans by their physical and/or behavioral characteristics. This article presents a novel neural network-based approach for features-level fusion in a multimodal biometric identification system by combining both physical (human face) and behavioral (handwritten signature) traits. A single biometrics system has the weakness of providing neither 100% identification nor a 0%. false accept rate (FAR)/false reject rate (FRR). One solution to this is to combine different biometrics together to get a multimodal biometric identification system. Moreover, a multimodal system is also robust in providing security against spoof attacks. Images of 32 x 32 pixels are used to eliminate bulk storage and processing requirements.