基于轮廓匹配和多项式拟合的虹膜分割算法  

Iris segmentation algorithm based on contour matching and polynomial fitting

作　　者：张薇 李菁 吕晓琪[1,2] ZHANG Wei;LI Jing;LÜ Xiaoqi(School of Information Engineering,Inner Mongolia University of Science and Technology,Baotou 014010,China;School of Information Engineering,Inner Mongolia University of Technology,Hohhot 010051,China)

机构地区：[1]内蒙古科技大学信息工程学院,内蒙古包头014010 [2]内蒙古工业大学信息工程学院,内蒙古呼和浩特010051

出　　处：《中国医学物理学杂志》2021年第1期47-50,共4页Chinese Journal of Medical Physics

基　　金：国家自然科学基金(61771266,61179019);包头市科技计划项目(2015C2006-14)。

摘　　要：虹膜图像在采集过程中由于个体差异、采集环境等因素的影响,存在遮挡、偏角等噪声。针对该问题,提出一种基于轮廓匹配和多项式拟合的虹膜分割方法。首先将虹膜图像分为理想型虹膜图像与非理想型虹膜图像;其次针对理想型虹膜图像,根据圆的标准方程检测虹膜边界实现虹膜分割,针对非理想型虹膜图像,通过轮廓匹配算法获得带遮挡部分的虹膜后,采用多项式拟合的方法去除眼睑等遮挡部分,实现虹膜的准确分割。本文算法实现虹膜分割的错误率nice1为1.5%,nice2为1.9%,且F1值达到93.65%。结果表明,该方法能够有效地去除虹膜图像中的遮挡噪声,具有较高的虹膜分割准确率。In the process of iris image acquisition,there are noises such as occlusion and declination due to individual differences,acquisition environment and other factors.Aiming at this problem,an iris segmentation method based on contour matching and polynomial fitting is proposed.First,the iris images are divided into ideal iris images and non-ideal iris images.Then,for ideal iris images,the iris segmentation is realized by detecting the iris boundary according to the standard equation of circle,and for non-ideal iris images,after obtaining the irises with occluded part through the contour matching algorithm,the polynomial fitting method is used to remove the occluded part such as eyelid to realize the accurate segmentation of irises.In this paper,the error rate of this iris segmentation is 1.5%for nice1 and 1.9%for nice2,and the F1 value reaches 93.65%.The results show that this method can effectively remove the occlusion noise in iris images and has a high accuracy of iris segmentation.

关 键 词：虹膜 图像分割 轮廓匹配 多项式拟合 

分 类 号：R318[医药卫生—生物医学工程] TP391.4[医药卫生—基础医学]