Isometric Efficient and Accurate Fourier-SIFT method in Iris Recognition System

Iris is the Optimum Biometric-trait present in Biometrics Security. Our emphasis on this paper is to obtain efficient, fast and robust algorithm set for iris detection. There are number of algorithms proposed for the efficient result but fails due to limitations. We tried in this paper to make, an efficient combination of the best schemes of normalization, corner detection, feature extraction and best matching algorithm available. We have used Modified-Trajkovic Operator (8-neighbours) to detect the corner, in which, the iris image sample is first optimized by sector based normalization into four sectors, this decreases the iris area but the modified 8-neighbours detect the corner accurately. Fourier-SIFT is then used to determine the keypixels with enhanced threshold cutoff and finally Modified-Hausdorff distance (or Gromov-Hausdorff distance) determines the matching algorithm and measures the distance between keypixels of enrolled and scanned iris during matching. The limitations in the above algorithms are rectified in this paper. This process is rigorously checked on CASIA-V3 and MMU iris images.