基于特征向量场重构的虹膜识别方法

受人脑工作机制的启发,提出基于特征向量场重构的虹膜识别方法.该方法首先对两幅归一化后的矩形虹膜图像,取其各自的行、列一阶差分并映射成特征向量场;其次,根据重构思想从一幅虹膜图像中取出行一阶差分,从另一幅图像中取出列一阶差分,利用二个差分重构出一个新的特征向量场.通过计算两个特征向量场的重构距离判断两幅虹膜图像的异同.与其他方法相比,该方法的优点在于不需要对虹膜图像进行去噪与增强,节省了计算时间;计算中没有使用积分形式的滤波器,使得该方法适于硬件实现.该方法经历了285 290次匹配实验,正确识别率达到98.81%.方法的有效性得到了充分验证.

Iris Recognition by Restructuring Characteristic Vector Field

Relating the iris recognition to the working mechanism of human brain, a method of restructuring the characteristic vector field is recommended. It calculates the lst-order differences separately from the rows of one normalized iris image and columns of another one, then the two differences are mapped to form a vector field. Based on the restructuring idea, the lst-order differences are further calculated from the rows of an iris image and from the columns of another iris image for restructuring a new characteristic vector field. Then, whether the two iris images are resulting from the same iris is judged by computing the restructuring distance between the two fields. Comparing the method with some other existing ones, it shows the advantage that its preprocessing procedure does not require denoising and image enhancing but reduces the time for computation. Moreover, because there is no filter in integral form used during computation, the method is adaptable to hardware implementation. With the match test done 285,290 times, it was revealed that the rate of correct recognition is up to 98.81% by use of this method, thus verifying its validity.