Published Paper Details:

Eye-based biometric systems, leveraging retina and iris recognition, offer high reliability for human authentication. Iris recognition is favored for its stability and accuracy, even among identical twins, but faces challenges like eyelid and eyelash occlusions. This paper proposes a system using preprocessing, the MLRP algorithm, and image matching for efficient implementation. It refines iris recognition in less constrained imaging scenarios with an algorithm employing RANSAC and direct ellipse fitting for improved boundary localization. The system, validated using SDUMLA-HMT, CASIA-Iris-V3, and IITD databases, demonstrates high recognition rates, surpassing traditional methods. Additionally, IrisConvNet, a multimodal system using CNNs and Softmax classifiers, enhances performance and generalization. This approach, validated across three datasets, combines face and iris recognition for increased accuracy and security, overcoming traditional biometric challenges. The system's robustness and efficiency make it a leading solution in biometric authentication, promising broad application in security and identification systems.