CAJ | 학술논문

针对虹膜识别中的睫毛干扰抑制问题,本文提出了一种基于形态学算子的睫毛干扰抑制算法.该算法将虹膜图像中的睫毛对象看作是背景图像中存在的随机的,不规则的细微裂缝,通过膨胀算子对这些裂缝进行像素填补,实现了对睫毛干扰的有效抑制.与常见的高斯滤波法相比,所提算法具有更强的睫毛干扰抑制能力.通过CASIA虹膜库的仿真实验表明,所提算法比高斯滤波法消除的睫毛像素点要多40%,可以使Daugman和Wildes定位算法的定位精度分别提高1.7%和2%,定位时间分别减少27.9%和24.2%.
침대홍막식별중적첩모간우억제문제,본문제출료일충기우형태학산자적첩모간우억제산법.해산법장홍막도상중적첩모대상간작시배경도상중존재적수궤적,불규칙적세미렬봉,통과팽창산자대저사렬봉진행상소전보,실현료대첩모간우적유효억제.여상견적고사려파법상비,소제산법구유경강적첩모간우억제능력.통과CASIA홍막고적방진실험표명,소제산법비고사려파법소제적첩모상소점요다40%,가이사Daugman화Wildes정위산법적정위정도분별제고1.7%화2%,정위시간분별감소27.9%화24.2%.
In order to improve the performance of eyelash interference suppressing algorithm in iris recognition method,a new algorithm based on Morphological operators was developed.Considering the eyelash as the free nondirectional silt in the background of eye image,the suppression of eyelash interference can be taken as a process to fill up these fine slits by using dilation operator.Compared with Gaussian filter algorithm,the proposed algorithm can suppress more eyelash interference,and has better performance.The experiment results from the CASIA iris database show that the proposed method can reduce more 40% eyelash pixels than that of the Gaussian filter algorithm.Using the proposed method,the accuracy of the Daugman's algorithm and Wildes' algorithm,which are traditional iris location algorithm,are improved 1.7% and 2% respectively,and the location time are decreased by 27.9% and 24.2% respectively.