基于霍夫变换的可见光虹膜图像定位

Iris Localization for Visible-light Images Based on Hough Transform

虹膜定位是虹膜识别中的关键步骤,而非理想条件下的可见光虹膜图像定位算法则是近年来的研究热点。为了提高虹膜定位的精度,提出了一种基于霍夫变换的可见光虹膜图像定位算法,分析并确定了先定位虹膜外边界、再定位内边界的顺序,以提高边界定位准确度;针对外边界定位存在的问题,利用可见光虹膜图像颜色分布的先验信息,进行了边缘检测与筛选,提高了定位的速度与准确度;在此基础上,借助瞳孔和虹膜的位置关系指导虹膜内边界定位,进行了高光噪声检测与去除,采用基于距离约束的霍夫变换,排除噪声影响和减少计算时间。实验结果表明,该算法具有较高的定位准确度,并且能处理诸如严重遮挡、成像模糊、区域对比度低、虹膜纹理与隐形眼镜等干扰情况下的虹膜图像。

As a crucial step in iris recognition, the iris localization algorithms, especially those dealing with iris images taken in visible- light under non-ideal imaging conditions have received increasing attention recently. A localization algorithm for visible-light iris images has been proposed to enhance the accuracy of iris localization. The strategy of localizing limbic boundary has been determined which is carded out from out boundary to inner verge of iris to promote the accuracy of boundary location. To solve the problem in localizing lim- bic boundary, the prior of color distribution has been utilized to remove the noise in the edge map of the iris image for increase of the ac- curacy and speed of limbic boundary localization. Therefore, the relevance between center positions of iris and pupil has been utilized to instruct the pupil boundary localization and the reflections in pupil region have been detected and removed with Hough transform on dis- tance limitation to eliminate the noise effects and to reduce computation time via estimation of the pupil radius and center. Experimental results demonstrate that the localization accuracy of the proposed method is higher than traditional methods and has more capabilities such as dealing with iris images with occlusion, blur imaging, low contrast and interfering texture.