Vander Lugt Correlator based active contours for iris segmentation and tracking. (1st April 2017)

• Record Type: Journal Article
• Authors: Ouabida, Elhoussaine; Essadique, Abdelaziz; Bouzid, Abdenbi
• Is Part Of: Expert systems with applications. Volume 71(2017)
• Page Start: 383
• Page End: 395
• Publication Date: 2017-04-01
• Subjects: Biometrics -- Iris segmentation -- Active contours -- Optical correlation -- Eye tracking; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2016.12.001
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Fast an accurate iris segmentation system is presented. The Optical Correlation based Active Contour (OCAC) technique is described. A general study of the OCAC for iris segmentation and tracking is presented. Small computation time achieved by the algorithm is the advantage of the used method. The presented intelligent system perform a high accuracy in iris segmentation. Graphical abstract: Abstract: Iris segmentation using active contours approaches is receiving increasing attention. In this paper, a self-ruling active contour approach based on the optical correlation algorithm is proposed. The novelty of this research effort is to apply the Optical Correlation based Active Contours (OCAC) on iris segmentation and tracking and highlight the advantages of its small computation time and better accuracy performance. Optical correlation computed with a numerical simulation of the Vander Lugt correlator is used to detect iris and pupil areas which used as an initial contours. As a result, these initial contours assists the method to calculate terms in an energy expression. In the proposed method, several references images called filters of iris and pupil have been introduced. Images from four iris datasets as CASIA v4, WVU non-ideal, MMU2, UBIRIS v2, and a motion video were used in the experiments phase. To present an aggregate overview of the proposed method advantages, we computed several parameters as iris and pupil centers localization errors, iris and pupil rays errors, three performance metrics (as Jaccard coefficient, Dice coefficient, Hausdroff distance), average segmentation error, and average execution time. We compare these segmentation performance parameters with several leading techniques demonstrating significantly improved results with the proposed OCAC technique.