Fujitsu’s Biometric Liveness Detection Separates Forgeries from Real Faces

Fujitsu has unveiled a new passive liveness detection solution for facial recognition systems. The technology is designed to stop spoofing attacks carried out with a photo of the intended victim.

Unlike some other liveness detection systems, Fujitsu’s solution does not require a dedicated near-infrared camera, nor does it ask the user to complete any secondary (and inconvenient) head movements to establish their presence. Instead, the solution is able to use a standard camera to distinguish a forgery from a real face using only information present in the forgery the original image registered to the system.

To do so, the system takes advantage of the fact that any forgery will differ from a true face in a number of subtle ways. For example, a photo displayed on a screen may have unusual reflections, while a printed image may distort the face in certain areas.

Fujitsu’s solution isolates and extracts those specific variables. It then uses image processing to calculate the likelihood that the image is a fake, while accounting for the fact that even legitimate authentication images will differ based on the capture environment. The solution can determine whether a distortion is caused by sunlight or a nearby window rather than a screen, or that the image is blurry simply because the user’s head moved during capture.

According to Fujitsu, the new system will reduce costs for developers who will no longer need to include a second camera to enable liveness detection. It will also be more convenient for users, because it will eliminate the need for extra follow-up movements. The company is hoping that the solution will accelerate digital transformation and streamline the authentication process for remote workers that need to gain access to corporate materials.

Fujitsu is planning to launch the new system in March of 2021, shortly before the end of fiscal year 2020. When it does, it will be one of several companies with a fully passive liveness detection offering. ID R&D recently updated its popular IDLive Face solution, while IDmission achieved ISO compliance with its own passive liveness solution back in June.

Until then, Fujitsu will continue to improve the accuracy of its solution. The company has also introduced a new video analytics platform for smart vehicles.

–

September 18, 2020 – by Eric Weiss