Face Biometrics

Facial recognition technology has been in use for decades, and remains one of the most widely used biometrics. Facial recognition technology uses the layout of facial features and their distance from one another for identification against a “gallery” of faces with similar characteristics. These characteristics can be derived from either a still or video images. Using statistics, facial recognition algorithms can measure the differences between the face being searched and the enrolled faces in a gallery. The smaller the difference, the more likely those faces match.

Facial recognition technology is primarily used for three different types of applications:

First, facial recognition technology can anonymously characterize faces. This allows for counting unique faces presented to the sensor over a period of time (sometimes called a “people counter”). Other functions include estimating the age, gender, ethnic origin, and even body mass index of each unique face thus encountered, usually for marketing purposes.

Second, facial recognition technology can verify a face against a known image. For example, this would allow for confirmation that a face presented at a border checkpoint matches the digital face embedded in a document. It also allows for access control, such as at the entrance of a building with a known and restricted population. This function is typically called “verification”.

Third, facial recognition technology allows for identification of a face against a number of known faces within a database. For example, this allows for the technology to see if a criminal or terrorist in a surveillance video matches any mug shot photos of people previously arrested or convicted. This function is typically called “identification”.

Advantages of face as a biometric:

A mature technology which has seen significant recent improvements in terms of accuracy and speed of capture
Images can be captured in the background of an interaction, reducing the need for lengthy and intrusive capture procedures
Images can be derived from existing pictures or video footage – physical presence is not always required
Greater social acceptance than other biometrics

Common applications of face as a biometric modality: