Comparative analysis of selected feature extraction techniques for iris recognition system

Abstract

The complexity of features extraction has a great influence on the program and processing speed of iris recognition system. Inappropriate feature extraction methods use in the representation of iris images may affect the classification accuracy of iris recognition system. It is important in iris recognition technology to look into different feature extraction techniques. This paper addresses the challenge of how to deduce the effectiveness of some feature extraction methods by carrying out comparative analysis on selected feature extraction techniques: Gabor Wavelet Transform, Scale Invariant Feature Transform (SIFT) and Haar Wavelet Transform. The experimental evaluation of the feature extraction techniques based on CASIA iris dataset gave different results of each technique. Finally, it was showed that the Gabor Wavelet Transform technique outperformed the other two techniques in term of recognition accuracy, training and testing time