一种快速虹膜定位算法的研究

Research on A Fast Iris Localization Algorithm

提出一种结合梯对检测和随机圆检测(RCD)的虹膜定位方法。通过对原始图像作滤波,提取图像梯度和边缘像素集合。按照梯度对满足条件搜索边缘图像中的梯度对,用梯度对计算圆的参数,通过统计计算来完成内圆定位。在处理后的边缘图中按规则随机取4点,根据RCD准则判断4点是否共圆,用公共圆判据完成虹膜外圆的分割。实验结果表明,该方法有效可行,可达到98.1%的成功定位率,并缩短了运算时间,减少了系统对内存的需求。

A fast algorithm for iris localization based on gradient pair detection and randomized circle detection （ROD） is presented. We employed edge detection to generate an edge map. The inner circle detection was to search the gradient pair vectors based on center symmetry rule and to obtain the circle parameters through the Statistic calculating. The outer circle detection used in the proposed ROD is that we first randomly select four edge pixels and define a distance criterion to determine whether there is a possible circle. We applied an evidence-collecting process to localize the outer circle of iris. The Experimental results show that the proposed method has an encouraging performance that shows 98.1% localization success and reduces the system operation time and memory demands.