Ultrasound-Based Liveness Verification

基于超声波的活体验证

• 批准号: 80953
• 金额: $ 17.44万
• 依托单位: EYN LIMITED
• 依托单位国家: 英国
• 项目类别: Collaborative R&D
• 财政年份: 2020
• 资助国家: 英国
• 起止时间: 2020 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=80953
• 关键词: Ultrasound; Based; Liveness; Verification

Face aliveness or simply “liveness” detection is a challenging problem during (remote) authentication and identity verification. Such techniques are often used for Know Your Customer (KYC) during opening a bank account, applying for a mortgage, log into HMRC, apply for universal credit, to name a few.In this context, by liveness detection we refer to the genuine presence of a live human’s biometrics such as their face. Liveness detection confirms that such presence is not coming from a replica, that is, a pre-captured photo, a pre-recorded video or a mask. EYN developed a prototype of a novel passive liveness detection system. Our proposed solution is innovative for three main reasons. First, we conduct liveness detection passively without asking the user to do something and unlike existing “active” approaches it protects the “how” and “when” of the challenge-response that is being done, therefore, limiting the scope of an adversary to counter-attack. In other words, we leverage short ultrasound pulses to detect genuine presence of the human face and leverage corresponding video to cross-check results. Second, the solution does not require any custom hardware, but leverages existing hardware, that is, the camera, loudspeaker and microphone which are available in most portable devices such as a smartphone or a tablet. Third, it is a 100% software-only solution that can be deployed in billions of portable devices available in the market, at scale.

人脸活体检测是远程身份认证和身份验证中的一个具有挑战性的问题。这些技术通常用于在开立银行账户、申请抵押贷款、登录HMRC、申请通用信用卡等过程中了解客户（KYC）。在这种情况下，通过活体检测，我们指的是真实存在的真人生物特征，例如他们的脸。活性检测确认这种存在不是来自复制品，即，预先捕获的照片、预先录制的视频或掩模。EYN开发了一种新型被动活性检测系统的原型。我们提出的解决方案具有创新性，主要有三个原因。首先，我们被动地进行活性检测，而不要求用户做什么，与现有的“主动”方法不同，它保护了正在进行的挑战-响应的“如何”和“何时”，因此，限制了对手反击的范围。换句话说，我们利用短的超声波脉冲来检测人脸的真实存在，并利用相应的视频来交叉检查结果。其次，该解决方案不需要任何定制硬件，而是利用现有硬件，即在大多数便携式设备（例如智能手机或平板电脑）中可用的摄像头、扬声器和麦克风。第三，它是一个100%纯软件解决方案，可以大规模部署在市场上数十亿台便携式设备中。