Taraneh Haeri - Department of computer engineering Mashhad Branch, Islamic Azad university Mashhad,
Hamidreza Pourreza - Department of computer engineering Ferdowsi university of Mashhad

in this paper, a novel iris recognition method withmulti-class support vector data description (SVDD) classifier ispresented. This method detects and recognizes a new class by thedescription of known classes. In SVDD, the width parameter sand the penalty parameter C influence the learning results. TheN-fold M times cross-validation method is well-known andpopular scheme to calculate the best (C, s) values. The method weutilized for iris localization includes simple filtering, edgedetection, and Hough transform. For feature extraction, 2D Log-Gabor wavelets are used and output is a one-dimensional vectorof statistical features for each iris area.

Hough Transform; Iris biometric; 2D log-Gabor wavelet; Statistical features; SVDD classifier