Iris recognition using localized Zernike's feature and SVM

Iris recognition is an approach that identifies people based on unique patterns within the region surrounding the pupil of the eye. Rotation, scale and translation invariant, are very important in image recognition. Some approaches of rotation invariant features have been introduced. Zernike Moments (ZMs) are the most widely used family of orthogonal moments due to their extra property of being invariant to an arbitrary rotation of the images. These moment invariants have been successfully used in the pattern recognition. For designing a high accuracy recognition system, a new and accurate way for feature extraction is inevitable. In order to have an accurate algorithm, after image segmentation, ZMs were used for feature extraction. After feature extraction, a classifier is needed; Support Vector Machine (SVM) can serve as a good classifier. For the N-class problem in iris classification, SVM applies N two-class machines. Indeed, in this type of validation, data are divided into K subsets. At any given moment, one is for testing and the other one is exclusively for validation. This method is called K-fold cross validation (Leave one out) and each subset is considered as an original series. Simulation stage was accomplished with IIT database and the comparison between of this method and some other methods, shows a high recognition rate of 98.61% on this database.