Iris Recognition System Evaluation Experiments Using CASIA Version3
Samar Al-Saqqa, Mohammad Al‐Rawi, King Abdullah · 2013
In this paper, we provide new evaluation experiments on iris recognition system, these experiments are based on different evaluation metrics which investigate the iris recognition system accuracy using a new version of the well-known public database CASIA version3. Two datasets from CASIA V3-Interval database are used, the first set is set100 contains 100 images and the second set is set2421 contains 2421 images, to get the best recognition the optimum values of the 1-D log Gabor filter parameters are recorded to each set. The registered False Accept Rate (FAR) and False Reject Rate (FRR) using Set2421 are 0.07%, 0.12% respectively when the separation threshold value is 0.4, while FAR and FRR are zeros using Set100 when the separation threshold value is 0.39. The relation between FAR and FRR for different values of the threshold is represented by Receiver Operating Curve (ROC). The recommended template size to use is (20x200) and the number of shifts is eight. (Samar Al-Saqqa ,Mohammad Al-Rawi, Moh'd Belal Al- Zoubi. Iris Recognition System Evaluation Experiments Using CASIA Version3. J Am Sci 2013;9(9):117-127) (ISSN: 1545-1003) http://www.jofamericanscience.org. 17