Abstract: A novel iris segmentation method is proposed to remove noises such as eyelashes and eyelids occlusion, and the pupil, which would possibly exist in iris regions and result in poor recognition performance. The accuracy boundary of the pupil is achieved by an active counter model technique. After computing the wavelet transform of the intensity signal of the normalized iris images along horizon direction, the position of eyelid occlusion edge points can be obtained by evaluating the behavior of the wavelet modulus maxima over scales, and the eyelid occlusion regions can be determined by polynomial-fitting. Combined with the eyelashes detecting results obtained by the 1-D Gabor filter, efficient iris segmentation is achieved. Compared with the existing iris segmentation algorithms, the proposed method not only illuminates the aberration influence on the normalized iris image caused by the traditional circular model of the pupil, but debases the implement complexity by avoiding the issues of determining the searching ranges of 4-parameters spaces involved in traditional eyelid boundary detecting. Experimental results on the CASIA iris database images show that the proposed segmentation method can decrease the Equal Error Rate of the recognition system from 8% to 4.4%.

Key words: iris recognition, iris segmentation, active contour model, wavelet modulus maxima