Improved Multimodal Biometric Security: OAWG-MSVM for Optimized Feature-level Fusion and Human Authentication

Nowadays, the increasing demand for high-safety and dependable authentication systems, has led to the growth of the unimodal biometric system (UMBS) so that the multimodal biometric system (MBS) has developed. The MBS will utilize more than a single biometric trait of an individual for authentication and safety purposes. Now onwards, Fusion is playing a main role in the MBS, and various fusion methods are utilized in BSs. Feature-level fusion is a famous approach as compared to others. In this fusion procedure, features are extracted from each biometric trait, and these features are extracted and merged into a final vector of high dimension (HD). This research article implements a novel method to perform fusion at the feature-level by optimized feature-level fusion. Now, reliable feature vectors are chosen to utilize optimization methods, such as ant lion and grey wolf optimization. The proposed work has implemented a novel OAWG-MSVM model for choosing optimized feature sets. However, the proposed work recommended the authentication or recognition method and it utilized the multi-class support vector machine (MSVM) with a kernel function. Lastly, the evaluation of the researched model is calculated by some calculations, such as accuracy, SN, and SP. It is considered that the implemented method attained 98.0% accuracy when comparison has been done with existing methods (CNN and OGWO). The research model has been designed in the MATLAB platform. Keywords: Biometric System (BS), Multimodal Biometric System (MBS), Optimized Ant Wolf Grey-Multiclass Support Vector Machine (OAWG-MSVM), Convolutional Neural Network (CNN), Oppositional Grey Wolf Optimization (OGWO).