Breadcrumb navigation

High accuracy biometric authentication technology capable of simultaneous face and iris recognition

Featured Technologies

March 14, 2020

NEC, the world’s leader in biometric authentication technology, has developed a prototype proprietary authentication terminal capable of one-step face and iris recognition. The time required to complete authentication is only two seconds, making highly-accurate speedy authentication possible. We spoke with the researcher about the details of this device and technology.

The need for even greater accuracy to promote the widespread use of biometric authentication

Ryoma Oami
Principal Creator
Biometrics Research Laboratories

― What are the advantages of simultaneously performing face and iris recognition?

First, I’d like to start with the presupposition that NEC’s face and iris recognition technologies are already highly accurate on their own. Our technology has been shown to have over a 99% accuracy rate and was ranked No. 1 by a U.S. government agency (*). Despite this, being able to simultaneously conduct face and iris recognition is extremely significant.
One of the advantages is that it allows biometric authentication to respond to even more needs and can be used in an even greater variety of situations. For example, when face recognition is used by itself, authentication becomes difficult at night or in the dark as well as at medical facilities and food plants where a large portion of a person’s face will be covered. There are also individuals with psychological or other reasons for wishing to keep their faces covered. In such circumstances, authentication which relies solely on face recognition can be unsuitable. Meanwhile, authentication which relies solely on iris recognition may not work very well for people with glasses or vision impairments. Being able to simultaneously perform face and iris recognition makes it possible to provide authentication services even under such circumstances. Combining the two recognition technologies brings us one step closer to the realization of a universal authentication system.
Another advantage is accuracy. As I mentioned earlier, both face and iris recognition are already extremely precise. However, what happens when the number of users increases to the tens and hundreds of millions? At such levels, even a 0.01% error rate can lead to authentication errors. Combining the two recognition technologies is the equivalent of increasing the number of keys needed to enter your house to two or doubling the length of a password. In order for nations and large-scale corporations to feel secure when using authentication services, the existence of even more precise authentication technologies such as this one becomes essential. With this technology, it becomes theoretically possible to provide authentication services to several hundreds of millions of individuals without any issues. NEC has already delivered a national identification system to India which combines face, iris, and fingerprint recognition. It’s my belief that successfully designing a terminal capable of one-step face and iris recognition by combining the two highly compatible recognition technologies opens up new possible uses and ways of providing value.

Speedy authentication just by standing in front of a camera and without the need to adjust one’s posture

― What kinds of technologies are used?

The device uses the world’s No. 1 face and iris recognition engines as is. An extremely high degree of accuracy is already possible because of this. The main obstacle we encountered when developing this authentication device was actually finding a way to quickly capture images that could take advantage of the capabilities of these engines.
A camera with a wide-angle lens can capture a subject’s face relatively easily for face recognition, but the problem is iris recognition. High resolution images of both eyes need to be captured for iris recognition because a large amount of information becomes necessary for accurate authentication. This meant we needed a specialized camera just for this, and the angles and focus range would be extremely narrow because the image would be captured using a telephoto lens. As a result, the biggest technology issues we needed to overcome for developing this authentication terminal was finding a way to capture a subject’s iris smoothly.
Capturing the iris would be easy if we had users hunch over and move towards and away from the authentication terminal, but that would strain the user and be time consuming. For example, if you had to adjust your posture and move around when you’re making a payment, chances are you’d find it bothersome. It’s our belief that in order to have people truly realize the convenience of authentication systems using biometric authentication technology, the authentication process should be completed without any conscious effort or unnatural movements on the part of the user.
That’s why we ended up with a cylindrical authentication device. We designed the device to have a motor within the cylinder and cameras that can tilt up and down. By doing so, the cameras can detect and automatically track a user’s eyes. The technology we utilized here is NEC’s proprietary eye detection technology which has also been used for NEC’s remote gaze detection technology . Once the face recognition camera detects a subject’s eyes, the camera moves to an appropriate position and the iris recognition camera also detects the subject’s eyes. While the iris recognition camera is doing this, it’s also estimating the subject’s distance in order to determine the best focus and lighting. This is how we achieve a smooth authentication process in only two seconds.
Since a person’s face about 40 to 70 cm away from the camera can automatically be recognized, the device placed on a cashier table of average height can authenticate people who are anywhere between 110 and 190 cm tall. In most cases, this means a person won’t have to hunch over or make bothersome movements. The device has also been designed to authenticate people who are in wheelchairs without the need to stand up.
In the end, I think there are multiple possible approaches to how to integrate the results of simultaneous face and iris recognition to obtain a final authentication result. If someone wants a more universally applicable authentication system, then they can set it up so that the results of either the face or iris recognition will be sufficient. Someone who wants a more precise system can use a combination of the two. How the results are processed will differ according to the needs and the situation, so we’ll keep doing research to determine the best way to go about it.

Aiming to provide this terminal in 2021 for use in store payment systems, etc.

― Tell us about the prospects of this technology.

After conducting more tests in 2020, one of our goals is to begin offering this technology in 2021 for use in payment and access control systems.
We’re currently aiming to provide this technology to retail stores. Large-scale retailers with stores throughout a country deal with customers numbering in the tens of millions, so we believe that the ability of our technology to make the payment process more accurate by using face and iris recognition is of great value. Customers who are shopping will often be carrying bags in both hands or carrying children, so we see a face and iris recognition system which doesn’t require the use of hands to be a good fit. Imagine being able to shop without anything in your hands and doing so without a wallet or smartphone. The thought of such a future becoming a reality fills me with excitement.
Additionally, we believe this technology will be of use in a wide variety of situations such as managing access control to areas and locks for entering homes.
NEC has also successfully created face and iris recognition technology that works even with walking subjects (*). We’re also doing research into the possible development of multi-modal authentication systems that can be combined with the aforementioned technologies.

Successful authentication using face and iris recognition
Successful authentication using face and iris recognition

Relevant Laboratories