快速去除睫毛及高效眼睑检测方法

Rapid Removal of Eyelashes and Efficient Eyelid Detection Method

根据虹膜分割得到的图像确定上下眼睑的检测区域,针对上下眼睑的分布特征,分别设计了等级滤波器及方向滤波器并结合膨胀的方法去除上下眼睑周围的睫毛;进而设计了基于gabor滤波的快速眼睑检测方法,能够在减少睫毛干扰的眼图中高效的找到上下眼睑的有效点,并对这些点使用二次多项式分别拟合出了上下眼睑的曲线,从而计算虹膜的有效区域,作为判断该眼图是否能够进行特征提取,提高了虹膜识别的效率,为虹膜识别提供了辅助作用;算法模型简单,复杂度低,运行速度快,自适应性强,上下眼睑检测时间均在18mm左右,能够在各种尺寸的眼图以及灰度分布不均的眼图中精确定位眼睑位置。因此该算法可应用于芯片,以及日常虹膜识别功能的软件开发中。

The detection area of upper and lower eyelids is determined according to the image obtained from iris segmentation.According to the distribution characteristics of upper and lower eyelids,hierarchical filters and directional filters are designed respectively, and eyelashes around upper and lower eyelids are removed by expansion method.Furthermore,a fast eyelid detection method based on Gabor filter is designed,which can effectively find the effective points of upper and lower eyelids in the eye diagram to reduce eyelash interference.The curves of upper and lower eyelids are fitted by quadratic polynomials to calculate the effective area of iris. Finally,we can judge whether the eye image can extract features.This method not only improves the efficiency of iris recognition,but also provides an auxiliary role for iris recognition.The algorithm has the advantages of simple model,low complexity,fast running speed and strong adaptability.The detection time of upper and lower eyelids is about 18mm.The algorithm can accurately locate the eyelid position in eye images of various sizes and uneven gray distribution.Therefore,the algorithm can be applied to the development of chips and daily iris recognition software.