Abstract: This paper presents a fast iris localization algorithm based on morphology concerning problems that iris pattern which is disturbed by eyelid, eyelash and faucal, localization with low-speed and low accuracy, etc. It employs gray projection algorithm and morphology to realize coarse localization of the pupil and carries out fine localization to both inner and outer iris circles within a small range by using both Hough algorithm and circumference differential algorithm. It adopts improved template algorithm and circumference gradient algorithm to locate eyelid. It uses gray threshold algorithm combined with morphology to detect eyelash. The experimental results show that the proposed method achieves excellent results in terms of both speed and accuracy for localization.

Key words: iris localization, Hough algorithm, circumference differential algorithm, template detect, gray projection

摘要： 针对有眼睑、睫毛、光斑干扰的虹膜图像的定位速度慢且不精确等问题，提出了一种快速精确的虹膜定位算法。利用灰度投影法和形态学实现瞳孔的粗定位，再分别利用Hough算法和圆周差分算法在小范围内对虹膜内外边缘进行精定位，采用改进后的模板检测法和圆周梯度法定位眼睑，运用灰度阈值法结合形态学检测睫毛。实验结果表明，该算法在定位速度和准确率上都达到了非常好的效果。

关键词: 虹膜定位, Hough算法, 圆周差分算法, 模板检测, 灰度投影