Metalenz uses polarization information for secure face biometrics

Deploying secure facial recognition hardware camera sensors in smartphones has always been costly. Although Apple has achieved this, most Android devices still need help in this area. With face unlock becoming increasingly popular, many manufacturers are implementing face biometrics vulnerable to deepfakes.

Metalenz has developed a hardware system to tackle these challenges by consolidating individual plastic lenses into a single flat surface. This approach not only simplifies the design but also lowers the cost of camera systems, allowing OEMs to integrate sophisticated sensing devices into consumer electronics.

Metalenz’s PolarID is the company’s facial recognition system, which leverages proprietary metalens technology that is capable of using polarization information. In an interview with DEMO, Metalenz CEO Rob Devlin explains, “If you can extract the polarization information, you can even analyze the underlying biology or chemistry of certain objects that you’re observing.”

In many cases, cameras capable of capturing polarization information require bulky and expensive equipment. Metalenz has miniaturized these capabilities to integrate into smartphones and other applications for access control.

Devlin also mentions, “You could do a scan of someone’s face with polarization and see whether a skin growth is cancerous or not. So there’s a future set of applications we see building off of the initial face unlock, but that’s the real first target.”

The potential applications of this camera system extend into medical diagnostics and automotive security, offering a face-based authentication mechanism for unlocking vehicles. With PolarID, the system allows users to also make digital payments and access sensitive information.

Last year, the company collaborated with Qualcomm to demonstrate PolarID on their latest Snapdragon mobile platform. The objective is to introduce a secure face unlock feature across the entire Android ecosystem, including approximately “1.5 billion devices.”

During the interview, Devlin presents the spoof detection capabilities, which were being run on an Android device powered by a Qualcomm Snapdragon chip. The system was tested against various spoofing methods, including 3D-printed masks and high-resolution photos.

These simulated facial features can deceive many current systems, particularly those relying solely on RGB cameras. However, PolarID’s utilization of polarization data sets it apart by separating genuine human skin and synthetic materials.

The company recently appointed AI and Imaging expert Gaurav Aggarwal and Mari-Anne Gagliardi, a fabless semiconductor operations veteran.

Article Topics

biometric liveness detection  |  biometrics  |  face biometrics  |  metalens  |  Metalenz  |  Polar ID  |  research and development  |  smartphones