摘要
通过分析归一化虹膜图像中包含睫毛及眼睑的灰度特征,文中提出一种新的虹膜睫毛及眼睑检测方法,能够同时快速检测睫毛和眼睑,算法模型简单,复杂度低.首先对归一化虹膜图像进行中值滤波,然后用滤波后的图像与原图像做差,最后将做差的图像细化并去除伪目标点.通过对不同图库和一些特殊情况下采集到虹膜图像进行检测,证明该方法能够在很好检测睫毛及眼睑的同时,具有检测速度快,检测精度高的优点,克服了传统方法针对睫毛及眼睑建立不同的数学模型而导致复杂度增加,检测速度慢的缺点.
By analyzing the gray--scale characteristics of eyelashes and eyelids in normalized iris image, this paper presents a new eyelashes and eyelids detection method. This method can detect the eyelashes and eyelids simultane- ously with a simple mathematical model. Firstly, proposed the image by median filter. Secondly, to minus the new image with original one. At last, figure out the skeleton of the image. Experimental results show that the method can detect eyelashs and eyelids quickly and effectively. Overcome the traditional methods for the eyelashes and eye- lids to create different mathematical models which led to increase in complexity, shortcomings of testing slow.
出处
《微电子学与计算机》
CSCD
北大核心
2011年第4期122-126,130,共6页
Microelectronics & Computer
基金
国家自然科学基金项目(60672078)
关键词
虹膜识别
睫毛检测
眼睑检测
中值滤波
骨架化
iris recognition
eyelash detection
eyelid detection
median filter
skeletonization