一种基于小波包分解的虹膜防伪检测方法

Method for Fake Iris Detection Based on Wavelet Packet Decomposition

虹膜识别系统易受到伪虹膜的攻击,包括利用打印在隐形眼镜、玻璃眼或其他材料上的假虹膜纹理进行的攻击,所以检测出假虹膜的防伪算法至关重要。为此,提出一种利用小波包分解进行特征提取的虹膜防伪检测方法。首先对目标图像进行二级小波包分解,然后计算各子带图像的熵,最后将各子带图像的熵值之和作为提取的特征值。实验在包含500幅虹膜图像的数据库中进行,结果表明此方法可以有效地将真假虹膜区分开来。

Iris recognition systems are vulnerable to spoofing by fake copies which include a fake iris printed onto a contact lens, glass eye, or other artifice. The anti-spoofing algorithm that picks out fake iris data is highly desirable. An application of wavelet packet decomposition to the feature extraction part of a fake iris detection algorithm is represented. First, 2-layer wavelet packet decomposition is performed to the target image. Then, the entropy value associated with each sub-image is calculated. Finally, the sum of the entropy values is used as a feature value for discriminating fake iris from real ones. Experiments, conducted on a database of 500 iris images indicate the method to be very effective.