一种基于圆周差分法的新颖虹膜定位算法

A new algorithm on iris image localization based on circular integrodifferential operators

虹膜定位是虹膜识别中非常重要的一个环节。Daugman提出的圆周差分定位法鲁棒性和精确性较好,但当在整幅图像中搜索虹膜时,将非常的耗时,同时也会因为受到睫毛等的干扰而降低定位的精确性。本文在圆周差分方法的基础上,利用行、列的灰度信息和直方图来实现瞳孔的粗定位,然后采用一种新颖的快速迭代搜索方法来实现瞳孔和虹膜的精定位。采用此种粗定位与精定位相结合的方法极大的缩短了虹膜定位的时间,提高了定位的准确性。通过对大量虹膜图像的定位实验表明,新方法取得了非常好的结果。

Iris localization is a very important part of iris recognition. The circular integrodifferential operators of John Daugman is robust and accurate. However it costs more time when used in the whole iris image to search the iris and reduces the accuracy because of the disturbance of eyelash. Based on the circular integrodifferential operators, a new idea is used to achieve the approximate position of pupil and iris in this article, Then a fast iterative searching method is used to realize the accurate iris localization. By using this algorithm, the time cost in the iris localization is sharply shortened, and the precision is improved a lot. After doing a great deal of experiments, the result show that the proposed method has an encouraging performance.