Improved Iris Recognition using Discrete Fourier Transform
Jaydeep Kale, Nilesh G. Pardeshi, Vikas N. Nirgude · 2013
This paper presents efficient algorithm for iris recognition using Two dimensional (2D) Discrete Fourier Transform (DFT) and illustrate how increased iris region improves performance. Phase components present in 2D DFTs of given images are used to determine similarity between two images. Algorithm is evaluated with CASIA iris image databases (version 1.0) and it clearly demonstrates that the use of phase components of iris images help to achieve highly accurate iris recognition. By experimentation, it is observed that instead of considering only lower half part of iris, in addition if we consider the portion present in upper half part of iris then it decreases False Match Rate (FMR) significantly as effective region available for iris comparison is increased. Decrease in FMR results in decrease in error rate (EER). General Terms Pattern Recognition, Security.