基于自编码器和Fisher矢量编码的人脸防伪检测研究  

Research on face spoofing detection based on auto⁃encoder and Fisher vector encoding

作　　者：周珊珊 赵小超 李哲 ZHOU Shanshan;ZHAO Xiaochao;LI Zhe(School of Computer and Information Science,Hubei Engineering University,Xiaogan 432000,China;School of Computer Science and Information Engineering,Hubei University,Wuhan 430062,China)

机构地区：[1]湖北工程学院计算机与信息科学学院,湖北孝感432000 [2]湖北大学计算机与信息工程学院,湖北武汉430062

出　　处：《现代电子技术》2024年第13期1-6,共6页Modern Electronics Technique

基　　金：国家自然科学基金项目(72002067);湖北省自然科学基金项目(2020CFB497,2022CFB516)。

摘　　要：为了应对针对人脸识别系统的欺骗攻击,设计一种基于软件的方法来检测人脸欺骗攻击。首先,从多个颜色通道中随机提取局部视频块并将其向量化,随后训练并使用自编码器将提取的向量进一步编码为局部低维特征;接着,利用这些特征拟合一个高斯混合模型;然后,使用Fisher矢量编码将来自同一样本的全部局部低维特征聚合成一个全局特征向量;最后,通过支持向量机对这些全局特征进行分类。在三个标准人脸欺骗数据集上的实验结果证明了该方法的有效性。在IDIAP数据集上的半错误率为0%,在CASIA数据集和MSU数据集上的等错误率分别为3.33%和2.08%。该方法结合自编码器和Fisher矢量编码,在低维特征的情况下取得了较高的性能,可以有效地抵抗人脸识别中基于照片和视频的欺骗攻击。A software-based face spoofing detection method is designed in order to cope with spoofing attacks against face recognition systems.Local video blocks are randomly extracted from multiple color channels and then vectorized.An auto-encoder is trained and employed to encode the extracted vectors into local low-dimensional features.These features are used to fit a Gaussian mixture model.The Fisher vector encoding is applied to aggregate all local low-dimensional features from the same sample into a global feature vector.These global features are classified by support vector machine(SVM).The experiments on three benchmark face spoofing datasets demonstrate the effectiveness of the proposed method.Its half total error rate(HTER)on the IDIAP dataset is 0%,and its equal error rates(EERs)on the CASIA dataset and MSU dataset are 3.33%and 2.08%,respectively.The proposed method is based on auto-encoder and Fisher vector encoding,and achieves high performance with low-dimensional features.In conclusion,this method can effectively and efficiently counter photo-and video-based spoofing attacks in the face recognition.

关 键 词：人脸识别 欺骗攻击 人脸防伪检测 局部采样 自编码器 Fisher矢量编码 特征提取 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]