An Unified Trio-Biometric Authentication Systems: Optimized NeuroFusionNet (OENF-Net) And HMVW Optimizer

This research introduces the deep learning-based methodology of the multimodal biometric system (Trio-biometric recognition) utilizing ears, cuticles and eye-sockets, as the distinct identifiers. This methodology contains the validation phase and the enrolment phase. While on enrolment, the high-quality images for ears, cuticles, and eye-sockets, were acquired and pre-processed to clear visibility and consistent illumination for features. By using an inception V3 model, the features extracted, and then they are stored with the database (cloud). In a validation phase, the images for ears, cuticles, or eye-sockets are acquired. Moreover, Authentication are performed by using a new optimized NeuroFusionNet (ONF-Net), whose weights are optimized using the new Multi-Verse Grasshopper Whale Optimizer (HMVW). An OENF-Net determines whether an individual is authenticated or not. This proposed approach enhances the robustness and accuracy with biometric authentication. On combination of ears, cuticles, and eye-sockets as multimodal identifiers and using the deep learning techniques and optimized RBFNs, the efficient and comprehensive authentication framework (with accuracy of 94.3%) are achieved.