摘要
为了解决虹膜识别过程中睫毛噪声干扰的问题,提出了一种基于形态学算子和小波变换相结合的睫毛抑制算法。该算法首先将虹膜图像中的睫毛区域模拟成背景图像中的"裂缝",通过形态学膨胀算子对裂缝区域进行像素填充,然后利用小波变换的多分辨率特性,对变换后的高频系数进行非线性小波阈值处理,低频部分进行反锐化掩膜,最后经小波逆变换重构虹膜图像。样本仿真实验表明:该算法可以使Daugman和Wildes定位算法的精确度分别提高2.1%和2.43%,定位时间相对减少24.3%和22.6%。
In order to improve the performance of eyelash interference suppressing algorithm in the iris recognition,a new eyelash suppression algorithm based on the combination between morphological operators and wavelet transform is proposed.In the proposed algorithm,the eyelash area of the iris image is firstly simulated into the ＇cracks＇ of the background image and the crack area is filled with pixels by morphological dilation operator.Then,the high frequen cy wavelet transform coefficients are processed as nonlinear wavelet threshold based on the multi-resolution characteristic of wavelet transform,and low-frequency parts are processed as the unsharp mask.Finally,the iris image is reconstructed by using inverse wavelet transform.The simulation results show that the Daugman and Wildes localization methods based on the proposed algorithm have an improvement of about 2.1％ and 2.43 ％ for location accuracy and a drop of about 24.3％ and 22.6％ for location time,respectively,compared with the Daugman and Wildes localization algorithms based on the Gaussian filtering algorithm.
出处
《数据采集与处理》
CSCD
北大核心
2013年第5期586-590,共5页
Journal of Data Acquisition and Processing
基金
江苏省高校科研成果产业化推进项目(JHB2012-9)资助项目
江苏省高校自然科学研究重大项目(13KJA510001)资助项目
关键词
虹膜定位
膨胀
小波变换
睫毛干扰
iris registration
dilation
wavelet transform
eyelash interference
