欢迎访问《图学学报》 分享到:

图学学报

• 视觉与图像 • 上一篇    下一篇

一种快速有效的虹膜图像预处理方法

  

  • 出版日期:2012-08-31 发布日期:2015-07-28

A fast and effective method of iris image preprocessing

  • Online:2012-08-31 Published:2015-07-28

摘要: 为解决虹膜识别中的虹膜定位时间较长、虹膜内边界变形、眼睑和睫毛普
遍存在的问题,提出了一种新的虹膜图像预处理方法,该方法对虹膜图像采用数学形态学算
子对图像进行处理,合理有效地减少了边缘图像中非虹膜的边缘点;然后对图像直方图采用
双直线拟合直方图求取检测眼睑和睫毛的阈值。对中科院自动化所CASIA-IrisV3 虹膜数据
库的实验结果显示,提出的虹膜图像预处理方法准确快速,能合理有效去除眼睑和睫毛。

关键词: 图像预处理, 数学形态学, 虹膜定位, 眼睑检测, 眼毛检测

Abstract: To solve the problems that the iris localization is too long, the inner edge of iris is
not ideal circular, and eyelids or eyelashes commonly exist, a novel method of iris image
preprocessing is presented. The false edge points of iris are greatly reducing by mathematical
morphology operator to process the iris image. Finally, eyelashes and eyelids are detected by double
line fitting. Testing on the CASIA-IrisV3 database, experiment results show that the proposed
algorithm has a high performance on speed and precision, and eyelids and eyelashes detection
algorithm can remove eyelid and eyelashes reasonable and effectively.

Key words: image preprocessing, mathematical morphology, iris location, eyelid detection;
eyelash detection