Windows Hello Face authentication

Microsoft Face authentication in Windows 10 is an enterprise-class authentication mechanism, This mechanism is integrated into the Windows Biometric Framework (WBF) as a core Microsoft Windows component called Windows Hello. Windows Hello Face Authentication utilizes a camera specifically designed for near-infrared (IR) imaging configurations to authenticate and unlock Windows devices and Microsoft Passport.

1, the main advantages and functions of Windows Hello face authentication

Here are the main advantages of using Windows Hello face authentication:

Facial recognition across all Windows 10-based devices and platforms with compatible hardware (near infrared sensors).

User-friendly interface that provides authentication in the form of a single signature to unlock Microsoft Passport.

Enterprise-level authentication and access to content supported by Microsoft Passport Pro, including web resources, websites, and payment methods.

The ability to use infrared to provide consistent images under different lighting conditions (while allowing for subtle changes in appearance, including facial hair, makeup, etc.).

2. Scheme

The two main application scenarios for Windows Hello face authentication in Windows 10 are login or unlock with authentication, and re-authentication to prove that you are still in the same place.

3. Working principle

The Windows Hello face recognition engine consists of four different steps that enable Windows to understand who is in front of the sensor:

Find faces and find signs:

In the first step, the algorithm detects the user's face in the camera stream and then locates the facial marker points (also known as alignment points), which correspond to the eyes, nose, mouth, and so on.

Head direction:

To ensure that the algorithm has enough field of view against the face to make authentication decisions, it ensures that the user is facing the device +/ -15 degrees.

Representation vector:

The algorithm uses marker locations as anchor points to build a representation by taking thousands of samples from different areas of the face. The most basic form of representation is a histogram that represents the difference between light and dark around a particular point. No face images are stored - it's just representations.

Decision Engine:

Once the sensor has a representation of the user in front of it, it is compared to the registered user on the physical device. The representation must exceed the machine learning threshold for the algorithm to accept it as a correct match. If multiple users are registered on the system, this threshold is increased to help ensure that security is not compromised.

4. Registration

Registration is the step of generating your own item or set of representations (for example, if you wear glasses, you may need to register with and without glasses) and storing them in the system for future comparison. This collection of representations is called registered profiles. Microsoft never stores the actual image, and your registration data is never sent to a website or app for authentication.

Most users will probably need to register with each device once. Additional registration is required for users who meet the following criteria:

Occasionally wear certain types of glasses

A significant change in the shape or texture of the face

Move to an environment where near-infrared light is strong (for example, if taking the device outdoors in the sun)