基于自适应阈值的虹膜分割算法研究  

Research on algorithm for iris segmentation based on adaptive threshold

作　　者：李鹏[1] 曹兵[1] 

机构地区：[1]南京理工大学机械工程学院,江苏南京210094

出　　处：《电子设计工程》2018年第3期184-188,共5页Electronic Design Engineering

摘　　要：虹膜分割是虹膜识别必不可少的过程,目前虹膜分割算法一般都是采用固定阈值对图像进行二值化,然后通过Hough变换检测圆的方法。虹膜颜色相对较浅,这样的方法不仅需要合理的输入参数,而且无法通过二值化方法分割出虹膜区域,鲁棒性很差。为此,提出了一种自适应的虹膜分割算法:用自适应阈值对图像进行二值化,通过一阶中心矩计算瞳孔中心点的位置,进而分割出瞳孔;对原图像进行直方图均衡化,以瞳孔中心为圆心,用不同角度和不同半径的扇形对图像进行扫描,求出不同半径扇形圆弧区灰度值和之差的最大值所对应的半径,根据半径和瞳孔中心找出若干点,再通过这些点进行圆的拟合。实验结果表明,该算法能够较好地分割出虹膜区域,鲁棒性强。Iris segmentation is an essential process of iris recognition,the algorithms for iris segmentation are generally used by fixed threshold for image binarization and then through the Hough transform to detect circle.Iris' s color is relatively shallow,so the method not only needs reasonable input parameters,but it can't segment the iris region through binary image.The robustness is poor.Therefore,an adaptive algorithm for iris segmentation was proposed:The image binarization with adaptive threshold,calculating the pupil center position by the first order central moment,then getting the region of pupil;Histogram equalization is performed on the original image,the pupil center as the center of circle,the sector with different angles and radius to scan the image,calculating the maximum difference between the intensity of different radius of fan-shaped arc area and getting appropriate radius.some points were found out according to the radius and the center of pupil.The circle was fitted by the points.Experimental results show that the proposed algorithm can segment the iris region well,and its robustness is strong.

关 键 词：虹膜分割 自适应 二值化 灰度值 拟合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]