对于不完美的个人身份鉴别,高精度的虹膜区域分割有助于后面的特征提取和分类识别。不同光照下虹膜的边界定位适于采用微积分方法,但是这种方法容易受到光源像点的影响,使得定位失败,并存在实时性差的问题。本文采用亮点梯度检测方法消除光源像点干扰问题,通过瞳孔位置估计提高微积分定位快速性。仿真结果表明,本文算法对于不同光照下虹膜的边界定位是有效的。
For personal identification based on imperfect irises,accurate iris segmentation is helpful to the feature extraction,classification and identification. The integral-differential method can be used to locate the iris boundaries in eye-images under different illumination cenditions. Iris boundary localization based on integral-differential method is modified to improve the localization success rates and the localization speed by eliminating light spots interference and using pupil the position estimation. Simulation results show that the modified algorithm is effective for iris boundary localization under different illumination conditions.