A Comparative Study of Feature Extraction Approaches for an Efficient Iris Recognition System

A wide variety of biometrics based tools are under development to meet the challenges in security in the existing complex scenario. Among these, iris pattern based identification is the most promising for its stability, reliability, uniqueness, noninvasiveness and immunity from duplication. Hence the iris identification technique has become hot research point in the past several years. This paper compares recognition rates, speed and other efficiency parameters resulting from three iris feature extraction algorithms that use statistical measures, lifting wavelet transform (LWT), and Gray-Level Co-occurrence Matrix (GLCM) respectively. Experimental results show that while LWT provides higher recognition rate, GLCM approach offers reduction in computation time with a small compromise in recognition rate. It also demonstrates that statistical measures is the most economical when recognition requirement is crucial.