The Effects of Gender Factor and Diabetes Mellitus on the Iris Recognition System’s Accuracy and Reliability

Abstract

We present a performance analysis of iris recognition system for healthy and diabetes affected irises, separately for female and male users. The database consists of 546 pictures from 162 healthy irises (62% female users, 38% male users) and 772 iris images from 181 diabetic eyes with a clearly visible iris pattern (80% female users, 20% male users), Adaptive weighted Hough ellipsopolar transform technique was used for iris segmentation, and then three popular iris encoding algorithms were implemented. Bhattacharyya distance was used for comparison of diabetic and healthy irises of men and women. We founf that for male users, diabetic effects on the performance of the system are more intense that for female users.