1. KAMEL GHANEM GHALEM - Complex System Laboratory, Higher School of Electrical and Energetic Engineering, Oran, Algeria.
2. ALI BENOUAR - Complex System Laboratory, Higher School of Electrical and Energetic Engineering, Oran, Algeria.
3. KAMILA ALIANE - Complex System Laboratory, Higher School of Electrical and Energetic Engineering, Oran, Algeria.
4. YASMINA ZINE - University of Science and Technology of Oran Mohamed Boudiaf, USTO-MB.
5. MOHAMED REDA AHMED BACHA - Complex System Laboratory, Higher School of Electrical and Energetic Engineering, Oran, Algeria.
This work introduces a new method to improve iris contour identification, especially for iris images taken in a constrained environment, by combining the Otsu method with hysteresis local thresholding. The iris segmentation is processed on the whole iris disc. A competitive method called the Dezert-Smarandache hypothesis is used to apply the fusion of the left and right iris at the score level. Using the challenging CASIA-IrisV4 Interval database, the proposed approach demonstrated encouraging results with an accuracy of 94.06%, a False Acceptance Rate (FAR) of 5.83%, a False Rejection Rate (FRR) of 20.41%, and an Equal Error Rate (EER) of 20.7%.
Biometric, Iris, Thresholding Method, Dezert Smarandache Theory.