Do you know how accurate biometric face recognition technology is?
Author: huifan Time: 2021-06-03
How accurate is face recognition technology?
According to the British "Security Action" magazine website as early as May 2011, London Stansted Airport has begun testing security gates equipped with facial recognition software, and if the results are good, it will be promoted in airports and checkpoints throughout the UK. This will be the first attempt to involve facial recognition technology in security checks on a large scale for important occasions. So, what is facial recognition technology?
- Facial recognition technology helps police catch suspects
According to an information report in China Youth Daily, a duty sentry of the People's Liberation Army (PLA) in Chongqing was attacked and killed, and the attacker took away a semi-automatic rifle. This is the "3-19" sentry attack case in Chongqing that shocked the whole country. Due to the low light when the attack occurred, the assailant's face was always covered under a light-colored wire cap, and the surveillance video failed to capture his front. But a week later, local police released a computer mockup based on an unclear video screenshot. How did the blurred video screenshot become a clear image?
The incredible thing about biometric facial recognition technology is right here. The principle is to extract information containing various facial parameters from a frame or a video, draw a facial portrait, and compare it with the stored images in the database, which involves multi-dimensional search, optimization calculation, computer imaging and other technologies. It has a wide range of applications, including law enforcement agencies to obtain evidence, security systems to pass credentials and personal identification.
- There are countless "key points" in the face recognition
When we look at a person, we usually recognize him or her without paying much attention to the details of the face, and the subtleties of the face are overlooked. In fact, there are many features of the human face that cannot be changed: the ratio of the inner and outer orbits of the eyes, the ratio of the length of the bridge of the nose to the width of the nose ...... These proportional relationships, which are determined by the facial bones, make up countless "key points" that do not change because of fat, thin, expression, or age. These proportional relationships are determined by the facial skeleton. So, based on these key points, statistical calculations are made to build a data model, and the human face is analyzed into a "matrix" composed of numbers and turned into a "language" that the computer can read. In this way, the face is converted into numbers by the computer and calculated little by little, finally extracting a set of features that can express the whole face.
May 13, Spain, Catalonia Technical University, Professor of Computer Science Emilino Acosta, in the British "air and business travel network" wrote an article that the human body has a natural identification system, including smell, hearing, vision, "smell to know women The most notable examples of olfactory and auditory identification are "smelling a woman" and "hearing her before seeing her". The visual recognition system is even more commonplace. We can know who someone is just by looking at them, a process that involves a lot of information processing done by the human brain on the fly. But the computer is not as smart as the human brain, it does not have the innate intelligence of the system, only by the programmed software to complete.
In many cases, facial recognition technology can come in handy when eyewitnesses cannot provide enough information about the criminal. Bangs, masks, hats and glasses are the most common obscurants in face images, which have a great impact on the accuracy of recognition and are also the usual blinders used by criminals. Criminals are certainly cunning, but there is no way to prevent their personal information from being recorded by the camera. The technology can be identified by a number of eyes, mouth, nose, eyebrows, etc., effectively avoiding the impact of age, posture changes and obscurants on the capture of facial information, so it is not difficult to "remove" the criminal's glasses and mask.
- An old technology with a new face
The actions of the British airport have led many to believe that "facial recognition technology" is a recent development. Between 1964 and 1965, Woody Bledsoe, a distinguished American mathematician and one of the pioneers of artificial intelligence technology, realized the importance of this technology and, together with two other scientists, successfully applied it to the intelligence service he was working for. Only due to the confidential nature of his work, the technology was not more widely known.