Deep learning model based on inceptionResnet-v2 for Finger vein recognition

Abstract

In recent years, The pattern of finger veins is widely recognized as an effective biometric for identifying a person. The traditional finger vein identification systems are based on handcrafted features. However, Finger vein systems has been switched toward automatic features extraction due to the emergence of deep neural networks that are capable of extracting deep features. In this paper, an inceptionResnet-v2 pre-trained deep convolution neural network model is proposed for finger vein identification. We tested the performance of the proposed model on the public Finger vein databases SDUMLA, MMCBNU and FV-USM. The obtained results indicate the effectiveness and security of the proposed network as it has achieved very small errors for all data sets