不均匀光照下的虹膜定位算法研究

虹膜定位是虹膜识别中的基础性环节。针对经典虹膜定位算法易受光照影响且速度慢的问题,分析不均匀光照对虹膜定位的影响,提出最小二乘法粗定位与微积分算法精定位相结合的虹膜定位方法：首先利用形态学灰度运算处理虹膜图像,对处理图像阈值化后提取并修复瞳孔区域,利用最小二乘法对瞳孔区域下部边界点进行内边缘粗略拟合;然后根据外边缘点存在区域的灰度梯度确定虹膜外边缘点,利用最小二乘法进行外边缘粗略拟合;最后利用微积分算法精确定位虹膜内外边缘。以中国科学院虹膜库CASIA（version 2.0）1 200幅虹膜图像做实验,平均耗费时间为4.38 s,定位成功率为98.3%。与经典的Daugman算法和Hough变换算法相比,所提出的算法对不同光照条件下的虹膜图像能更准确、快速地定位。

Iris Localization Algorithm under Asymmetry Quality of Illumination

Iris location is the basic step on iris identification.In allusion to the problem of being easily affected by illumination and low speed on classical iris localization algorithm,and on which the effect of different especially inadequate model of illumination is analyzed,the iris location method combining the least-square coarse method and calculus fine method is proposed.Firstly,morphological operation is applied to iris images,extracting and repairing the pupil area after applying threshold to processed image,carrying out a rough inner boundary fitting by utility of the least-square method on underpart boundary point of the pupil area.Then the outer boundary points of iris are determined according to the gray gradient of the region including edge points,using the least-square method to roughly fit the outer boundary.Finally,the calculus method is used to finely localize the inner and outer boundaries.The average time cost is 4.38 s,and 98.3% of 1 200 iris images from CASIA（version 2.0） are localized exactly.By comparison with Daugman＇s method and Hough method,the proposed algorithm can be applied for localization of iris images on different quality of illumination accurately and rapidly.