Biometric Finger Vein Recognition Using Evolutionary Algorithm with Deep Learning

In recent years, the demand for biometric-based human recog-nition methods has drastically increased to meet the privacy and security requirements. Palm prints, palm veins, finger veins, fingerprints, hand veins and other anatomic and behavioral features are utilized in the development of different biometric recognition techniques. Amongst the available biometric recognition techniques, Finger Vein Recognition (FVR) is a general technique that analyzes the patterns of finger veins to authenticate the individuals. Deep Learning (DL)-based techniques have gained immense attention in the recent years, since it accomplishes excellent outcomes in various challenging domains such as computer vision, speech detection and Natural Language Processing (NLP). This technique is a natural fit to overcome the ever-increasing biomet-ric detection problems and cell phone authentication issues in airport security techniques. The current study presents an Automated Biometric Finger Vein Recognition using Evolutionary Algorithm with Deep Learning (ABFVR-EADL) model. The presented ABFVR-EADL model aims to accomplish bio-metric recognition using the patterns of the finger veins. Initially, the presented ABFVR-EADL model employs the histogram equalization technique to pre-process the input images. For feature extraction, the Salp Swarm Algorithm (SSA) with Densely-connected Networks (DenseNet-201) model is exploited, showing the proposed method's novelty. Finally, the Deep-Stacked Denoising Autoencoder (DSAE) is utilized for biometric recognition. The proposed ABFVR-EADL method was experimentally validated using the benchmark databases, and the outcomes confirmed the productive performance of the proposed ABFVR-EADL model over other DL models.