Microsoft patent filing shows new 3D facial biometric technology with multi-spectral camera array
Microsoft has developed a 3D facial recognition technology that could be integrated into smartphones and other mobile devices, based on a convolutional neural network and enabling facial biometrics in both short and longer-range distances, Patently Apple reports.
The patent filing, published by the U.S. Patent and Trademark Office and titled ‘Depth+Multi-Spectral Camera,’ describes a plurality of spectral illuminators and sensors making up a sensor array that generate each pixel with a depth value and spectral value for each spectral light sub-band. A previously trained biometric engine outputs a confidence value indicating the likelihood that the image includes a face, and attempts to match it to the enrolled data.
Patently Apple points out that the patent was originally filed in May 2018, prior to the announcement that Microsoft will launch the Surface Neo and Surface Duo smartphones, which are expected to launch in Q4 2020 with folding dual displays. The convolutional neural network is covered within the patent, and could potentially identify facial features such as freckles and eye color.
Components of the system include an optical filter, infrared illuminator and multiple spectral illuminators, along with controllers and sensors for various aspects of the system.
Apple is also rumoured to be working on a TrueDepth 3D facial recognition system with more range, for applications like augmented reality, and the next version of the TrueDepth camera may be integrated into the rear of forthcoming iPhone versions, according to Patently Apple. The iPhone maker was recently granted a patent for a 3D facial biometric system with eye tracking.
Microsoft’s technology may never appear in any device, or could alternately be destined for use in future versions of HoloLens or another product.
The global market for 3D sensors is expected to grow at a highly robust 22 percent CAGR through 2024, according to recent market analysis.