Cnn Based Multimodel Biometric Authentication System Using Face And Fingerprint

Abstract

 Multimodal biometric authentication methods have been introduced to provide accurate and safe solutions, integrating both soft and hard biometric schemes. This paper suggests a new hybrid strategy that guarantees the user's loyalty to the device, as well as checking whether the user has passed the biometric test as a regular or spoofed one. The suggested scheme is two modules: Module I combines fingerprint, and face recognition to suit the relevant databases, while module II incorporates models based on fingerprint, and face anti-spoofing convolutional neural networks (CNN) to identify spoof. The hash of a fingerprint is correlated with the fingerprint database at beginning level. Following a successful fingerprint match, it is checked on a fingerprint model based on CNN to check whether it is a fake or actual. A same process is repeated for the face recognition, and the system allows users to log in to the system based on cumulative proof. In the system introduced, utilizing Squeeze Net to provide accurate and reliable test results align with previous systems, challenging the restrictions of usual authentication and spoofing activities.