Iris Recognition: A Java based implementation

Abstract

Biometric authentication has become increasingly popular in security systems. Recently, the systems based on the human iris, which develops a unique pattern before birth, have produced very high rates of recognition. The iris image is first blurred using a Gaussian filter, and the edge is detected using the Canny edge detection technique. An algorithm, which uses the center of the image as a starting point, is proposed to isolate the pupillary region. The initial estimate of the location of the pupil is then refined, and the iris is located by using the integrodifferential operator. In order to detect the upper and the lower eyelids, we deploy the integrodifferential operator again; however, the path of contour integration is changed from circular to arcuate. A thresholding technique is then applied to locate the eyelashes. The annular iris region is unwrapped from a polar coordinate system to a rectangular canvas. The 2D Gabor wavelets are used to extract the discriminating features. Then, the phase information is extracted to produce an iris code of 2048 bit and a mask, which denotes the noisy regions, of the same length. The Hamming distance is applied for the matching purpose. We also design a graphical user interface (GUI) in Java which allows the comparison of two images, the verification that an image is that of a specific person, and to search through the previously scanned irises for an exact match. The proposed scheme is computationally effective as well as reliable in term of recognition rate of 99.21%.