CBAS: A CNN-based biometrics authentication system

Abstract

Biometric identification is an important security application that requires non-intrusive capture and real-time processing. Security systems based on fingerprints and retina patterns have been widely developed, but can be easily falsified. Recently, identification by vein patterns has been practiced as a promising alternative. The paper presents a CNN-based system, realized on an FPGA that is able to perform all three recognition methods. The focus is on the vein recognition algorithm whose detection quality is sufficient for person identification but scalable for authentication.