一种高效的基于曲线拟合的虹膜区域分割方法

摘要

虹膜识别由于一系列优于其它生物特征识别方法越来越受到欢迎.虹膜识别系统的高可靠性正是得益于其独特的、不变纹理虹膜区域.根据大量的研究表明,使用不同的先进特征提取算法的平均识别率在98％以上.与此同时,如何找到一种强有效的方法来对虹膜区域进行准确而高效的分割成为了一个瓶颈.传统的分割方法需要大量的阈值和边缘检测,或需要在很大的参数空间表决,这些方法不仅难以操作而且耗时严重.文中提出了一种强有力的基于曲线拟合的虹膜区域分割方法,通过基于虹膜数据库THU-V1和CASIA-V1的测试证明:即使在严重的睫毛遮挡情况下,该方法也是非常有效和准确的.%Iris recognition is gaining popularity due to a series of advantages over other biometric identification methods.The high reliability of an iris recognition system results from the unique and invariable texture in the iris region.According to lots of researches,the average recognition rate is above 98％ using different state-of-the-art feature extraction algorithms.At the same time,the bottleneck is to find a robust method to segment the iris region accurately and efficiently.The traditional segmentation methods involve a lot of thresholding and edge detection,or searching in large parameter spaces,which is really difficult to manipulate and time consuming.A robust iris segmentation method based on curve fitting is proposed in this paper,the tests over iris database THU-V1 and CASIA-V1 prove that it is quite efficient and accurate even in severe eyelashes occlusion case.