Researchers from the US National Institute of Standards and Technology (NIST) conducted a comparative study of 11 applications for human recognition by the iris of the eye and found that some programs are capable of almost lightning-fast reliable identification of a person with a single image capture from the camera.
This means that very soon a new method of personal identification may be introduced everywhere. It will be much faster than comparing fingerprints and can be applied to anyone and everyone.
Iris identification has been known since the 1980s; the method has a number of advantages over the classical (great-grandfather's) finger rolling technology. The fact is that, contrary to the popular belief about the invariability of prints, in many people engaged in hard physical labor, prints can change beyond recognition, while the iris after 1.5 years of age "freezes" in its invariability. It is possible to obtain an image of the iris from a distance of 10 cm to 4 m, and the recording camera can be invisible to the object, because it operates in the near-infrared range (750 nm). This differs sharply from identification by the retina, when the eye must be pressed against a special scanner, and even more so from DNA analysis or fingerprinting.
In other words, iris identification is fast, accurate and can be used anywhere and everywhere without legal problems, including instant identification at an airport or train station. The Iris Exchange (IREX) III survey recently conducted by the National Institute of Standards and Technology was perhaps the first to open our eyes to all this.
To compare 92 algorithms for iris recognition provided by nine private companies and two university laboratories, 2.2 million people were selected, at one time included in government databases of the iris of the eye. A variety of people were subjected to experimental identification - both those who were in the specified database and random subjects.
As it turned out, among the algorithms were not the best quality: the accuracy ranged from 90 to 99.2%. The number of errors for different software differed by more than an order of magnitude. At the same time, some identification algorithms were able to iterate through the entire database in less than a second - which means that if it included all US residents, identification would take no more than 10 seconds.
It is curious that the main reason for inaccurate identification for the best algorithms is the inaccuracy of camera operators who direct them at too large angles to the iris of the person being identified. The creation of stereoscopic scanning cameras operating in the near infrared range would radically solve the problem. Although such a pair of cameras would cost twice as much as current devices, the increased accuracy would allow the technique to be used today as a key means of effective and near-instantaneous identification.
In conclusion, we note that the plans of India, Mexico, Indonesia and, of course, the USA and Canada include the adoption in the near future of framework laws providing for the massive introduction of the described means of identification, including in public places and in transport.
Promotional video: