一种新颖的虹膜图像噪声检测方法*

为提高虹膜识别的正确率，针对虹膜图像中存在着眼睫毛和眼睑这两类较难检测的遮挡噪声，在分析现有检测虹膜噪声算法的优缺点后,提出了一套新颖的虹膜图像噪声检测方法：基于Gabor滤波变换的灰度均值法检测睫毛和利用最小二乘法检测眼睑。实验表明，该算法能有效地检测两种遮挡噪声，准确率分别达到95.10%和96.51%，且等错率（EER）指标与已有算法相比最优，提高了虹膜识别系统的整体性能。

New method for iris image noise detecting

In order to improve the precision of iris recognition, aiming at eyelid and eyelash occlusions in iris image, this paper presented a new approach to detect iris image noise after analyzing advantages and disadvantages of the current occlusion detecting algorithms: the method of detecting eyelash based on average gray value Gabor filter transforming and the method of detecting eyelid based on least square method. The experimental results indicate that the proposed algorithms have better effectiveness on detecting these two kinds of occlusion noises, the accuracy reach 95.10% and 96.51%, ERR is more excellent than existing methods, improving the whole performance of iris recognition system.