News > i3

The Labyrinth of Biometric Possibility


With iPhone X, Apple leads the charge toward using face recognition for primary smartphone security technology.

A phone designer can allow or deny access to a smartphone several ways including biometric approaches such as voice, fingerprint and face recognition. These methods are beginning to replace passwords because the technology is convenient and less time consuming. Safer, too — especially if you use a birthday or 1234 to protect important information. 
 
Four years ago Apple’s Touch ID fingerprint identity sensor popularized biometrics on mainstream smartphones. With the launch of iPhone X in September, Apple eliminated the fingerprint sensor from the device, instead creating a detailed 3D map of your face with its Face ID system, which uses a selfie-type setup called the “TrueDepth camera system.” 

TrueDepth is an engineering expedition into the labyrinth of biometric possibility. It employs a 7MP frontfacing camera and an infrared emitter that projects more than 30,000 dots onto the user’s face, visible only to an infrared camera. A proximity sensor allows the system to know when a user is close enough (it warns the user when he/she is out of range) and an ambient light sensor. The iPhone X also has an infrared (IR) illuminator so the system will work in the dark. The light projections and sensor data are then fed to iPhone X’s A11 chipset, a proprietary dualcore design that processes the biometric facial recognition data in real time.

 Secure and Accurate 
 

The facial images from iPhone X are securely stored on the phone’s Secure Enclave security chip, not in the cloud. Face ID uses machine learning to recognize your face. Any changes in your appearance are factored in to an updated image and the dot pattern via neural networks that create a mathematical model of your face. Apple reports it will recognize you even if you put on eyeglasses or a hat. TouchID only had a one in 50,000 chance of unlocking for the wrong fi ngerprint. But Face ID is even better: it has a one in 1,000,000 error rate, says Apple. The company even worked with special eff ects mask-makers in Hollywood to ensure it would be difficult to spoof.

Using facial recognition for unlocking the iPhone X (as well as activating Apple Pay) may seem vaguely similar to 2010’s Microsoft Kinect, which measured depth, position and motion using a color video camera and an infrared light sensor to log you onto your Xbox Live account. A facial recognition system was created with the Microsoft Kinect v2 using a facial mapping library distributed with the Kinect v2 SDK (Software developer’s kit) as a basis for the algorithm. The system extracted 31 points representing various facial landmarks, which created a reference data set. In 2013, Apple acquired PrimeSense, developer of the structured light sensor for the original Kinect.

EyeLock, a subsidiary of  VOXX International, is also improving its realtime iris and facial authentication, while the auto company Ford is seeking to put biometric sensors into cars. Intel also has delivered a depth-measuring RealSense system that, like Kinect, provides laptops equipped with Microsoft’s Windows Hello with facial recognition and gesture recognition capability. RealSense contains dualinfrared cameras for depth sensing (calculated based on parallax mapping — the same way the human eye works) and a high-resolution camera that can capture images in 1080p. Unfortunately, RealSense is too large and power hungry to be used in a small device like a smartphone. 

Qualcomm has expanded its Spectra Module Program with improved biometric authentication and high-resolution depth sensing. The company will include the new technology in its Snapdragon mobile processors and make the Spectra processor available for Android phones. 

Acuity Market Intelligence projects by 2022 global mobile biometric revenues will reach $50.6 billion annually. This includes 2.7 billion biometrically enabled smart mobile devices generating $3.1 billion in biometric sensor revenue annually. The firm projects by 2020, biometrics will be standard on 100 percent of the nearly 2.4 billion mobile devices sold each year, marking a new milestone in security. 

Murray Slovick

Tagged

Related