Recognizing any person by analyzing their facial features is the goal of a team of European scientists who have developed a system that identifies a“DNI of the face”from each person. The new systemfacial biometricsis based on the creation ofindividualized models, as explained by mathematician David Delgado Gómez, a researcher at the Carlos III University of Madrid (UC3M) and co-author of the study.
“Our goal is to create a model for each person that highlights themost discriminating characteristicsof each face? Delgado explains. Thus, to describe an individual, “characteristics that the rest do not possess are used in a natural way, such as thetall woman with blue eyesor thebald guy with a beard; we try to apply this idea to our algorithm, “he adds.
A facial biometrics system usually consists of three components: onecamerathat records an image, a program that determines theface geometryand a system that classifies all those elements to differentiate between some and other people.
For now, researchers have found that when the new facial ID system is used in a controlled environment it can reach the95 percent accurate, although its application is not without difficulties. For example, incorrect lighting when registering the image could change the color of the face and prevent identification. Another obstacle that this system still has to overcome is thepass of the timeand the changes that this brings to people’s faces, such as wrinkles or weight changes, which can mislead classifiers.
