Abstract
For visible wavelengths, a new method is introduced using truncated LoG filter with specific preprocessing to deal with challenges of eyebrows, illumination variation, eyelid masking, perspective, and movements. Here, the iris boundary is detected first and then pupil is marked with maximally stable extremal regions. The experiments conducted on challenging database UBIRIS v2 demonstrate the improved segmentation accuracy of the proposed methods. The comparison with other popular methods provides promise in terms of accuracy and speed.