Study on eyelid localization considering image focus for iris recognition
Study on eyelid localization considering image focus for iris recognition
复制标题
DOI：
10.1016/j.patrec.2008.05.001
复制
发表时间：
2008-08-01
影响因子：
5.1
通讯作者：
Park, Kang Ryoung
中科院分区：
文献类型：
--
作者：
Jang, Young Kyoon;Kang, Byung Jun;Park, Kang Ryoung
本文提出了一种鲁棒检测算法，可用于虹膜识别中眼睑区域的检测。与以往的研究相比，本研究有以下四点优势和贡献。首先，我们对检测到的睫毛和镜面反射进行插值，使我们能够确定更准确的眼睑搜索区域。其次，我们能够通过找到眼睑和虹膜外边界之间的交叉位置来定义一个有限的眼睑搜索区域。第三，通过考虑对焦值的眼睑检测掩模，可以在虹膜图像散焦的情况下提高眼睑检测性能。第四，将旋转项应用到拟合检测到的眼睑候选点的抛物线霍夫变换中，即使在眼睛旋转的情况下也能检测到准确的眼睑位置。实验结果表明，对上眼睑和下眼睑的检测准确率分别为91.33%和98.45%。(c) 2008 Elsevier B.V.版权所有
This paper proposes a robust detection algorithm that can be used to detect eyelid region for iris recognition. This research has the following four advantages and contributions compared to the previous works. First, we interpolate the detected eyelashes and specular reflections, which enable us to determine a more accurate searching area of eyelid. Second, we are able to define a limited eyelid searching area by finding the cross position between the eyelids and the outer boundary of the irises. Third, by using eyelid detection mask considering focus value, we can enhance the eyelid detection performance even in the case of defocused iris image. Fourth, by applying the rotation term into the parabolic Hough transform which fits the detected eyelid candidate points, we can detect the accurate eyelid position even in the case of rotated eye. As the experimental results show, the detection accuracy rates were 91.33% and 98.45% when detecting the upper and lower eyelids, respectively. (c) 2008 Elsevier B.V. All rights reserved.