John Daugman presents four advances in iris recognition: 1) improved methods for detecting and modeling iris boundaries using active contours, enabling flexible coordinate systems; 2) Fourier-based methods for handling off-axis gaze by detecting it and rotating the eye into orthographic perspective; 3) statistical inference methods for detecting and excluding eyelashes; and 4) exploration of score normalization based on the amount of iris data and database search scale. Statistical results from 200 billion iris comparisons in the UAE database show the impact of normalization on receiver operating characteristic (ROC) curves. The paper discusses the challenges of iris recognition, including the need for high accuracy in detecting and matching irises, especially in difficult conditions. It introduces active contours for precise segmentation of iris boundaries, allowing for noncircular shapes and flexible coordinate systems. Fourier-based trigonometry is used to estimate gaze deviation and correct for off-axis imaging. Statistical inference methods are used to exclude eyelashes from influencing the IrisCode. Score normalization is explored to address the issue of varying numbers of iris data points, ensuring consistent decision confidence levels. The SQRT normalization rule is proposed, which rescales raw Hamming distances based on the number of bits compared. This normalization improves performance by reducing false matches and maintaining confidence levels, especially in large-scale databases. The paper also discusses the importance of score normalization in different regions of the ROC curve, where it significantly reduces false match rates (FMR) in large-scale comparisons. The UAE database of 632,500 IrisCodes is used to demonstrate the effectiveness of normalization, showing a dramatic reduction in FMR when using the SQRT rule. The results highlight the necessity of normalization for accurate and reliable iris recognition in large-scale applications.John Daugman presents four advances in iris recognition: 1) improved methods for detecting and modeling iris boundaries using active contours, enabling flexible coordinate systems; 2) Fourier-based methods for handling off-axis gaze by detecting it and rotating the eye into orthographic perspective; 3) statistical inference methods for detecting and excluding eyelashes; and 4) exploration of score normalization based on the amount of iris data and database search scale. Statistical results from 200 billion iris comparisons in the UAE database show the impact of normalization on receiver operating characteristic (ROC) curves. The paper discusses the challenges of iris recognition, including the need for high accuracy in detecting and matching irises, especially in difficult conditions. It introduces active contours for precise segmentation of iris boundaries, allowing for noncircular shapes and flexible coordinate systems. Fourier-based trigonometry is used to estimate gaze deviation and correct for off-axis imaging. Statistical inference methods are used to exclude eyelashes from influencing the IrisCode. Score normalization is explored to address the issue of varying numbers of iris data points, ensuring consistent decision confidence levels. The SQRT normalization rule is proposed, which rescales raw Hamming distances based on the number of bits compared. This normalization improves performance by reducing false matches and maintaining confidence levels, especially in large-scale databases. The paper also discusses the importance of score normalization in different regions of the ROC curve, where it significantly reduces false match rates (FMR) in large-scale comparisons. The UAE database of 632,500 IrisCodes is used to demonstrate the effectiveness of normalization, showing a dramatic reduction in FMR when using the SQRT rule. The results highlight the necessity of normalization for accurate and reliable iris recognition in large-scale applications.