【摘要】 主要实现一种不同于经典人脸识别安全防御系统的安全解锁方法。更具体地讲,不同于普通的应用手机前置摄像头进行拍照并预留图片进行人脸识别的方式,通过增加后置摄像头的图像信息,实时验证。为了防止伪造信息对系统攻击,加入了带有特定信息的眨眼反馈机制。通过完成系统要求的随机生成的眨眼频率,有效防止了图片、视频攻击。对于面具攻击,加入了表情识别功能。认为这项功能的使用是一种全新的安全验证机制。目前大部分生物验证机制都有其缺陷和局限性。人脸识别作为性价比最高的验活方式,成为安全验证的首选。但是仍然没办法有效防范视频图像攻击,就算是提高硬件性能进行双目识别方式的检验也无法防御3D面罩的攻击。目前提出的若干解决方案仍不能很好地应用于现实生活场景中,对该安全机制进行了设计并实现,并和其他现有的机制进行比较说明该方法的实用性和创新性。更多还原

【Abstract】 This article mainly implements a secure unlocking method that is not used in classic face recognition security defense systems. More specifically, it is not used for ordinary application mobile phone ’s front camera to take pictures and reserve pictures for face recognition. We add real-time verification by adding image information of the rear camera. In order to prevent forged information from attacking the system, we have added a blink feedback mechanism with specific information. The experimenter completes the randomly generated blink frequency required by the system, Effectively prevent image and video attacks. For mask attacks, this article adds expression recognition capabilities. We consider the use of this feature to be a new type of secure authentication mechanism. Most of the current biological verification mechanisms have their shortcomings and limitations. Face recognition, as the most cost-effective biopsy method, has become the first choice for security verification. However, it is still unable to effectively prevent video image attacks, and even if the performance of the hardware is improved and the binocular recognition method is tested, it cannot prevent the 3D mask attack. Several currently proposed solutions cannot be applied well in real life scenarios. This paper designs and implements the security mechanism. The comparison with other existing mechanisms shows the practicability and innovation of the method.更多还原