Enhanced Iris Segmentation Algorithm by extricating Normalization Phase

Abstract

The objective of the proposed method is to segment the uniform iris region(s) from the input eye image without the normalization phase. Four successive major operations are carried out to localize the pupil region. They are downscaling the acquired eye image, reverse function, threshold operation and the neighborhood pixel method. Therefore, pupil’s boundary is detected exactly irrespective of its shape, either a circle or an ellipse. The traditional approach identifies the annular iris boundary and then proceeds to normalization. In the proposed method, a special wedge is framed to extract the portion of the iris region directly from the left side of the pupil region. Thereby, normalization phase is exempted in the iris recognition process. On completion of the above phases, detection of pupil boundary and segmentation of a part of the iris region is done. Experimental results show that this proposed method has the superior accuracy in pupil localization and less computational time in segmenting the iris region.