Bio-Inspired Metaheuristic Feature Fusionmethod for Multi-Biometric Identification

Security for personal data has grown in importance in recent decades as a result of the internet's continued expansion. Various ways for recognising individuals are implemented to safeguard their personal information. The identification of users by their distinct physiological characteristics is a significant obstacle for biometric recognition systems. Therefore, in order to construct an effective multimodal recognition system, we used two modalities-the iris and the veins in a person's finger-to identify that individual. For feature extraction, the retrieved from imagesfinger vein and iris images are sent into the Adaptive Drone Squadron Optimization (ADSO) system. The characteristics obtained from the finger vein and iris are combined using a hybrid feature fusion method based on the suggested Enhanced sperm swarm optimization (ESSO) algorithm. Finally, Several-Phase Recurrent Neural Network(SP-RNN) is employed for personal recognition. Accuracy (0.97), True Negative Rate (TNR) (0.97), True Positive Rate (TPR) (0.96), Cumulative Match Curve (CMC) (0.907), Equal Error Rate (EER) (0.1001), Privacy (89.455), Rank-1 (88.0057), ROC (0.919), and security (88.820) were all improved by the suggested method.