Smoothening of iris images and pupil segmentation using fractional derivative and wavelet transform

Abstract

Pupil localization is the most significant preprocessing step in recognizing the iris. Iris images are often degraded by low resolution, specular reflections; occlusion by eyelids, contact lenses etc. In this paper, a novel approach, which combines smoothing of iris images and segmenting the pupil, is proposed. First, a fractional derivative mask is used for smoothening the iris images, which acts as a preprocessing step for improving the accuracy of segmentation. Subsequently, the pupil is segmented from the smoothened iris images, using wavelet transform. From the experimental results, it is clearly evident that the proposed method is not only efficient in pupil segmentation, irrespective of its shape but also capable of handling low contrast images or images with noise. Public Iris databases such as CASIA Version 1.0 and UTIRIS database are used for performance evaluation.