查看文章

This study provides a comprehensive comparison of various iris recognition algorithms, including Hamming distance, feed-forward neural network, and support vector machine (SVM) methods. The study aims to identify the most accurate and efficient algorithm for iris recognition in biometric authentication systems. The dataset from CASIA and uniform preprocessing techniques ensure fair comparisons. The Hamming distance algorithm achieves 79% recognition accuracy but suffers from high false accept and false reject rates due to its threshold-based nature. The feed-forward neural network algorithm achieves an improved accuracy of 87.5% and handles complex classification tasks effectively. However, it is computationally intensive and requires manual feature selection. To address these limitations, the SVM algorithm is explored using linear, polynomial, and quadratic kernels with techniques like SMO, QP, and …