一种改进的Log-Gabor滤波和SVM的虹膜识别方法

An Improved Iris Recognition Algorithm Based on Log-Gabor Filter and SVM

特征提取和分类是虹膜识别中的关键部分。由于小波分解后的低频子带包含了虹膜图像的主要信息,而Log-Gabor滤波能有效地提取出图像的纹理信息,将这两种方法结合是一个提取虹膜识别信息的有效途径。本文先对归一化的虹膜采用小波变换的方法细分图像,再用Log-Gabor滤波器对低频通道的子带图像进行更进一步的特征提取并量化,形成特征码本,最后采用支持向量机的分类器来进行分类。实验结果表明,分类器能很好地分离各类虹膜,识别率提高到了99.6%,等错率则降低为0.3%,比传统汉明距的分类方式有更优异的性能。

Feature extraction and classification is very important in the iris recognition. The low frequency sub-image of the wavelet transform contains the primary information of the iris, and the Log-Gabor filter can effectively extract the iris texture information. The combination of these two approaches is an effective way to extract the iris texture. This paper firstly decomposes the normalized iris image by the wavelet transformation to obtain the sub-images, and then uses a Log-Gabor filter to extract the features of the low frequency sub-image and generates the iris code. Finally support vector machines （SVM） is used to classify. The experiments results show the SVM can achieve good effect on the iris classification. The recognition rate is up to 99.6% and the equal error rate is reduced to 0. 3% . Compared with the hamming distance, the SVM has the better performance.