基于面部生命特征的3D假面欺骗攻击检测方法

Facial Vital Sign Based Countermeasure Against 3D Mask Attacks

现有的人脸认证系统大都易于遭受欺骗攻击,传统的攻击方式主要包含影印照片和视频回放。随着3D打印技术的不断发展和成熟,使用3D面具进行欺骗攻击逐渐成为新的威胁。针对3D面具欺骗攻击提出了一种新的特征用于攻击检测。该特征基于人体心脏搏动会导致真实用户的面部血流量发生周期性变化这一生理现象,提取了面部皮肤颜色信号的频谱。在公开的3D面具欺骗攻击数据库上的试验表明,联合使用该生理特征和面部纹理特征的抗攻击方法相比于以往单独使用纹理特征的方法准确率得到了显著提升,可以更好地抵抗3D面具的欺骗攻击。

Vulnerability to spoofing attacks is the main drawback for current face authentication systems. Traditional spoofing attacks include displaying printed photos and replaying recorded videos. With the development of 3D printing technology, the 3D mask spoofing attack has been becoming the new threat. A novel anti-spoofing feature was proposed to 3D mask attacks. A liveness feature from the power spectrum of facial color signal was extracted based on a physiological phenomenon that the color of real human face changed periodically due to the blood circulation. The performance of countermeasure jointly using the liveness feature and facial texture feature was evaluated on a public database named 3D Mask Attack Database（3DMAD） and achieved a higher accuracy comparing to previous methods that only considered texture features.