Feature Extraction of Iris Images using ICA for Person Authentication

This paper describes the Iris based authentication system using the Independent Component Analysis (ICA). In this paper, a customized iris extraction algorithm has been applied directly on the unfiltered grayscale image. The polar coordinate system has been used by the researchers for iris processing. Instead, an algorithm based on intrinsic coordinate system has been proposed to save on computation with satisfactory results. ICA has been used very successfully in face recognition. The proposed algorithm of ICA has been tested on "UBIRIS Iris Database of noisy and noise free images" of average iris visibility and the results showed the accuracy of identification more than 83% with excellent processing speed.