轮廓初始点检测对眼睛虹膜区域反射光的去除

针对传统人眼反射光模型单一的缺陷，提出了一种自适应的对不规则形状人眼反射光的检测并去除算法。对含有人眼反射光区域，应用区域生长算法对人眼图像进行分割，并利用垂直边缘检测算子提取虹膜左右轮廓边界，拟合虹膜轮廓，对虹膜所在轮廓区域，采用提出的反射光轮廓初始点检测算法，反射光区域判定算法，对其轮廓的位置进行自动定位并提取，利用虹膜及瞳孔的纹理信息对人眼反射光覆盖区域进行修补。通过对不同人眼图像进行验证，结果表明该方法对检测不同噪声条件下的虹膜反射光具有鲁棒性，并且能够更加快速有效地对虹膜反射光进行去除。

Eye corneal reflection removal based on boundary initial position detection

Corneal reflection removal is a critical step for an iris recognition system because it directly affects the gaze detection and tracking. Due to the different luminous environment, the model of the corneal reflection is not a regular circle. So an irregular shape of corneal reflection should be considered to detection and removal. In this paper, it focuses on reflection boundary which involves in iris region and it proposes a new method to locate the corneal reflection based on intensity gradient and texture gradient features. Also, in order to narrow the detection scope, it proposes to extract the iris region via detecting the iris center which based on color image segmentation, vertical edge operation and curve fitting. The result shows that the technique performs well, especially, to detect the salient reflection boundary. The proposed method is robust and applicable on various kinds of eye image with different eye center location and noise properties.