基于改进几何特征的虹膜定位算法

针对随机Hough变换造成的大量无效采样和无效累积问题，提出了基于改进几何特征的虹膜定位算法；首先通过先验知识对传统的去短枝算法进行了改进，然后对仍然留在图像上的噪声点和高频细节用‘投影’算法进行了进一步的去除，这样就降低了无效采样的概率；在虹膜粗定位的过程中，文中采用的是圆的内接直角三角形；任意取两点，比取三点的方法降低了无效采样的概率，提高了虹膜定位的效率。

Iris Location Algorithm Based on Improved Geometric Features

To solve the problems of a large amount of invalid samplings and useless accumulations in the randomized Hough transform, an iris location algorithm based on improved geometric features is presented. At first, the paper improves traditional algorithm of eliminating short twig by transcendental information, then ulteriorly eliminates the noises and high-frequency details that leave in the images by using projective algorithm, thus it reduces the probability of invalid sampling. In the processing of iris location, orbicular inner right-angled triangle is used. Because of using only two points, it reduces the probability of invalid sampling for second time compared with taking three points, and advances the efficiency of iris location.