Finger Vein Biometric System Based on Convolutional Neural Network

This paper presents a finger vein biometric system based on convolutional neural networks. Biometric systems are critical for use in forensic and military applications. Finger vein biometrics are more secure than other biometric features because of their uniqueness. The finger vein images are preprocessed using the modified speeded-up adaptive contrast enhancement (MSUACE) method. The histogram of oriented gradients (HOG) and modified difference of Gaussian (MDOG) algorithms are used to extract the finger vein features. Using the Advanced Encryption Standard (AES), we improved the security of finger vein features and protected them from a variety of threats. The convolutional neural network (CNN) is used to classify the features. The accuracy and EER of the presented biometric system are 96.65% and 0.35, respectively.