基于支持向量机的虹膜识别方法

Iris recognition based on support vector machine

首先利用虹膜处理系统对采集到的虹膜图像预处理,得到条形图像;然后利用主元分析方法(即PCA方法)进行特征提取,以达到降维的目的,得到的一个训练样本对应一个40维的向量;最后利用支持向量机使用序列最小优化算法进行虹膜识别,平均识别率达到了94.3%。结果表明,文中的方法取得了较好的效果,降低了训练时间,提高了训练效率。

First the sampled iris images are preprocessed in the iris recognition system to get the stripe images. With the method of PCA, the features are extracted for reducing the dimension and a 40 dimention corresponded with a training sample. Then the iris recognition is carried out, with the Sequential Minimal Optimization（SMO）, by using the Support Vector Machine（SVM）. The mean recognition rate is about 94.3 %, which shows that the recognition method decrease the training time and improve the training efficiency.