Improved Iris Segmentation Algorithm without Normalization Phase

Earlier state-of-art algorithms of iris recognition system works well for ideal-data, because of the perfect iris localization algorithms. The same ideal-algorithm suffers a lot under non-ideal data. Nonideal-data refers to eye image captured under unconstrained environments, such as non-uniform illumination, image captured at a distance, eye image with reflections, blurred image, off-axis eye images, majority of occlusions due to eyelashes and eyelids. So there arises a need for perfect iris localization and segmentation algorithm. This proposed method segments the iris region almost perfectly even in the nonideal data. Further, the fixed and consistent iris strip is obtained without the need for normalization phase. Moreover, this proposed method is tested on publicly available CUHK iris database. Experimental result shows that, this proposed iris segmentation algorithm results in better accuracy when compared to the earlier state-of-art-algorithms with reduced computational complexity.