基于特定感兴趣区采样的虹膜定位改进算法

从虹膜图像特点出发,以瞳孔形心为辅助点实现对虹膜图像特定感兴趣区采样.并且根据虹膜内外圆半径的生理特点设定Hough变换半径参量,依据虹膜内外圆近似同心圆来筛选与虹膜外圆最匹配的Hough变换结果.该算法在主动避开眼皮,睫毛干扰的同时又显著降低了Hough变换的计算量.通过对中科院自动化所CASIA虹膜数据库108组图像的虹膜定位测试结果表明,该方法平均定位时间0.83 s,定位准确率98.9%.

An Improved Iris Location Algorithm Based on Sampling to Special Regions of Interesting(ROI)

According to the characteristics of the iris image,a new algorithm,which takes the center of the pupil as a reference point to sample the special region of interest (ROI) of iris image is proposed.This algorithm also designs the radius parameters of Hough transform based on the physiological character of iris,and it chooses the best matching result that was achieved from Hough transform,for the inner and outer circles of an iris are approximately concentric.This algorithm is helpful to avoid the interference of eyelid and eyelash and reduce the calculation of Hough transform remarkably.The 108 groups iris location test by using the images of CASIA was finished.The result shows that the average location time is 0.83 s,and the location accuracy rate is as high as 98.9%.