PERFORMANCE EVALUATION OF SELECTED OPTIMIZATION TECHNIQUES FOR BI-INSTANCE BIOMETRIC SYSTEM

Abstract Identity verification remains a critical challenge, with standalone biometric systems often struggling with universality, distinctiveness, and performance limitations. To address these vulnerabilities, this project introduces an intelligently optimized bi-instance authentication system, integrating fingerprint and face recognition. Unlike conventional fixed-rule fusion approaches, our system harnesses the power of Swarm Intelligence (SI) algorithms Particle Swarm Optimization (PSO), Cuckoo Search Optimization (CSO), and Glowworm Swarm Optimization (GSO) to adaptively determine the most effective score-level fusion weights and decision thresholds. The objective function guiding these algorithms was the minimization of the Equal Error Rate (EER), a key metric for system accuracy. Our comprehensive evaluation demonstrates a dramatic enhancement in system performance: the optimized multimodal setup consistently achieved significantly lower EERs compared to individual modalities and a non-optimized fusion baseline. Notably, the optimized system achieved an EER of (Lowest EER Value)%, representing a (Percentage Improvement)% improvement over the baseline. The research concludes that intelligent optimization with SI algorithms is indispensable for maximizing the accuracy, robustness, and reliability of multimodal biometric systems, paving the way for more secure and efficient authentication solutions in diverse real-world scenario.