Real-time iris recognition and anti-spoofing: a fusion of GLCM Features and EAR metrics

Real-time performance of iris recognition systems faces significant technical hurdles when combined with anti-spoofing measures because this widely used biometric operates precisely by detecting accurate and unique features. An enhanced iris recognition system is proposed which utilizes Gray-Level Co-Occurrence Matrix (GLCM) for feature extraction alongside Eye Aspect Ratio (EAR) for performing live detection. GLCM analyzes iris textures expertly, yet EAR operates as a real-time spoof detector to verify user presence. A Voting Classifier ensemble consisting of Support Vector Machine (SVM) and Random Forest and Extra Trees Classifiers were implemented for classification purposes to advance system performance. The system demonstrated remarkable outcomes during benchmark iris datasets testing since it proved capable of achieving high precision and recall along with accurate live and spoofed iris differentiation. The proposed system demonstrated an accuracy rate of 99.5% which exceeded previous model performances in terms of speed and security measures. The system demonstrates its real-time recognition capabilities while maintaining security standards which makes it an ideal solution for access control and mobile security applications as well as sensitive transaction systems.