Biometrics Fusion with Applications in Passenger Re-authentication for Automated Border Control Systems

Over 4.1 billion aircraft passengers flew in 2017. This number is expected to be nearly double by 2037. Due to the significant growth in airline services and passenger traffic, automated border control (ABC) systems have been installed to control the air traffic flow in an automatic manner while maintaining the security. However, there are multiple security checkpoints from airport entry to boarding the flight which requires passenger's re-authentication multiple times. We propose a deep representation based learning to combine face and soft-biometrics. The proposed model has applications in automated border control to further ease the traffic flow while maintaining high security at various checkpoints. Using a deep learning-based feature fusion framework, our method obtains 99.02% genuine match rate (GMR) at false match rate (FMR)=10
^-5
using ResNet-18 model.