Technologists have been working on facial recognition technology for some time. Now it seems to be coming to practical use, especially in retail. The theory of facial recognition software is that each face is different enough from all others, that precise measures of such things as distances between the parts--eyes, nose, mouth, forehead features, hairline, ears and the like can serve as differential features just as well as fingerprints. In criminal investigations, facial recognition is used in concert with other identifiers like fingerprints to improve the precision of its investigatory system.
In September, 2014, the FBI announced that it has achieved "full operational capability" of its biometric database that would replace its old fingerprinting system. The system incorporates and links multiple kinds of biometric data, voice features, palm prints, DNA profiles and (perhaps most controversially) a state-of-the-art facial recognition technology. The facial recognition system allows criminal investigators to identify suspects by precise comparisons of facial images with a giant database of mug shots.
Meanwhile, a system of binocular cameras has been under development by the National Institute of Justice (NIJ). This system would enable facial recognition in real-time as people went about their business under the glare of the cameras. Lighting of facial images affects the accuracy of ordinary face identification. However, the NIJ is working on systems to overcome these problems, such as 3D modelling. According to the National Institute of Standards and Technology (NIST), the incidence of false positives (claiming an identification erroneously) has been halving every two years since 1993. They claim the rate of false positives is now around .003 percent.
Many people find the notion of the involuntary use of facial recognition to identify them as a threat to basic privacy rights. Technically it could allow the government to relate your face to medical, social, and financial records. Fingerprints and DNA identifiers are less controversial because they are not available to be just grabbed by a camera. There is something sacrosanct about a face. No one can go about their business in civilization without exposing their face to public view. People pause to consider these social and ethical matters then move on.
Facial recognition systems automate the old process of identifying mug shots by eye. The systems turn the photographs into a system of codified measurement numbers so that each person has a "face print." Most current facial recognition systems identify 80 "nodal points" or measurements on the human face and convert these measures into a standard series. The standard numeric series for the subject can easily be correlated with other standard face print series from other sources. A match is declared when correlations reach a standard level of strength.
Lighting of facial recognition cameras is an area of research focus because the use of imperfect visible light degrades the measurement capability of the cameras. Infrared light is under experimental use for making facial recognition images in real-time. The present study emphasizes the use of three-dimensional cubic data from a "Multi/Hyperspectral" imaging system. Such systems would eliminate some of the limits of current facial recognition. Such a system would include skin properties in the face print, for instance.
Facial recognition is commonly used for security purposes. However, the systems are increasingly being used for other purposes. Microsoft's Kinect gaming system uses facial recognition to differentiate between players in its games. Some mobile payment systems use facial recognition algorithms to securely identify and authenticate users. Systems employing facial recognition are being studied and deployed in airport security as well.
Smartphone facial recognition is under concentrated development. Smartphone facial recognition capabilities include image tagging and other social networking functions as well as personalized marketing functions. A research team at Carnegie Mellon University has developed an iPhone app that can take a picture of an individual and, within seconds, return the individual's name, date of birth and social security number. Facial recognition systems built into billboards have been developed with integrated software that identifies the gender, ethnicity and approximate age of passersby and alters itself to deliver targeted advertising.
Apple engineers decided to make the screen cover as much like the front as possible, which eliminated the fingerprint sensor used to unlock the phone. They opted to use 3D facial recognition sensing developed by an Israeli company, Touch ID, one of Apple's recent acquisitions. The system appears to function as well as the fingerprint system.
Taking optical measures of relative sizes and distances between facial features has proven as useful for identification as fingerprints. Technology is quickly being developed to make these measurement using digital cameras and smartphones. Applications range from commercial customer recognition, to security identification, and as keys for unlocking smartphones and computers.