Improvements in Video-based Automated System for Iris Recognition (VASIR)

Video-based Automated System for Iris Recognition (VASIR) performs two-eye detection, best quality image selection by adapting human vision and edge density methods, and iris verification for identifying a person. A new method of iris segmentation is implemented and evaluated that uses a combination of contour processing and Hough transform algorithms along with a new approach to eyelid detection. User-interaction is reduced by using automatic threshold selection to detect the pupil and by defining it to be a minimum boundary radius of the iris. VASIR's performance is evaluated with the MBGC datasets which were captured under unconstrained environments. The results show that the new method significantly improves the segmentation of the iris region and consequently the matching results. Our method also demonstrates that automated best image selection is nearly equivalent to human selection.