在远距离、运动中采集到的虹膜图像库中包含了闭眼虹膜图像,一种检测闭眼虹膜图像的方法被提出用于提取这些虹膜图像.该方法使用圆的积分微分算子确定闭眼虹膜图像的候选图像,使用直方图分析将候选图像进行二值化,采用边缘检测和连接、图像膨胀和最小二乘拟合等技术拟合抛物线,通过抛物线开口方向判断该图像是否为闭眼虹膜图像.在UBIRIS.v2虹膜数据库中进行了实验,结果表明,检测闭眼虹膜图像方法的使用提高了虹膜分割算法的效果.%In the recognition-at-distance and recognition-on-move system, iris images captured include some close-eye iris images. A methodology of judging closed-eye iris images in the iris database is used to extract the close-eye iris images. This methodology first uses integro-differential operator to obtain the candidate close-eye iris images. After obtaining a binary map by analyzing the histogram of the iris image, the parabola is fitted based on image dilation, edge detection, edge link and least square fitting. Then the iris images with the parabola opening to the top are considered as the closed-eye iris image. The experiments are performed in the UBIRIS.v2 iris training database, which shows that proposed methodology can judge the close-eye iris images effectively.
展开▼
