Palm Vein Recognition Using Convolution Neural Network Based on Feature Fusion with HOG Feature

With the advancement of science and the development of the times, the role of identity authentication technology in living society has become increasingly prominent. Palm vein recognition has become a popular research in biometric recognition due to its unforgeable characteristics. Aiming at the complex design of palm vein recognition based on traditional feature engineering methods, this paper proposes to use the deep features of convolutional neural network fused with hog features to identify palm vein. Experiments show that this method has achieved both higher speed and accuracy on two different databases, which achieves the recognition accuracy of 99.25% on CASIA and 99.90% on PolyU respectively.