Abstract

Iris-based biometric systems have gained importance for secure access control. However, the need for improved accuracy and efficiency remains a challenge. This research addresses these challenges by leveraging the Cultural Chicken Swarm Optimization technique (CCSO), which integrates a belief space for enhanced feature selection, optimizing both accuracy and computational efficiency. A total of 240 students from Ladoke Akintola University of Technology participated in the data collection process, with left and right iris images captured using a CMITech iris camera. The data underwent preprocessing, followed by feature extraction using the Haar Wavelet-Based Technique. CCSO was applied for feature selection, optimizing the discriminative power of the features. The optimized features from both irises were fused, and matching scores were computed using Mahalanobis distance to classify users as genuine or impostors. The experimental results demonstrate that the CCSO technique outperforms the standard Chicken Swarm Optimization (CSO) in both accuracy and computational efficiency. For the left iris, CCSO achieved a 23.33% FAR, 9.44% FRR, and 87.08% accuracy, while for the right iris, it achieved a 21.67% FAR, 8.89% FRR, and 87.92% accuracy, significantly improving upon CSO. For the multiinstance dataset, CCSO further improved accuracy to 96.25%, reducing the FAR and FRR to 5.00% and 3.33%, respectively, while cutting computation time by nearly 35.00%. CCSO also reduced the Equal Error Rate (EER) to 4.17%, as opposed to CSO’s 7.50%. These results highlight the potential of CCSO in real-time biometric systems, and future research will explore its application to other biometric modalities and larger datasets.