Affine Invariant Iris Human Identification Using Fuzzy System

The present study offers a new method to humanidentification by the iris images. The proposed system consistsof two main parts: feature extraction and decision-making. Inthe former part, elimination of pupil and other redundant partsare done and then sketch of the iris images are extracted.Afterwards, two feature vectors are obtained based on angularand radial partitioning. In the latter one, however, Manhattandistance function is used to determine the similarity of images.To find the ultimate answer, there is a fuzzy decision makingwhose inputs are the values derived from Manhattan distance offeature vectors and the output is the level of similarity. Theperformance of the proposed method is examined on NISTdatabase and the accuracy value is ۹۹.۲۱%. Beside accuracy,resistance to all attacks and changes such as rotation, sizechanges and noise and low cost of calculation are otherbenefits, fitting the system for real time applications