COMPARATIVE STUDY OF SUPPORT VECTOR MACHINE AND HAMMING DISTANCE USED FOR IRIS RECOGNITION

Abstract

This paper presents a comparative study of two well-known classification techniques of iris patterns, along with detailed description of some preprocessing steps. In preprocessing stage, Circular Hough Transform and Canny Edge Detector are employed for iris segmentation, while for iris normalization and feature extraction, the Rubber Sheet Model and one-dimensional (1-D) Log-Gabor Filter are used respectively. Finally for classification/matching of iris patterns, Hamming Distance and Support Vector Machine (SVM) are applied. The evaluation results on CASIA V.1 dataset show that Hamming distance algorithm is more suitable for the classification (with average accuracy of 93.85 %) of iris patterns.