What challenges does iris recognition face?

Iris recognition technology is an important way to identify unique individuals in the field of biometrics. We are generally more familiar with fingerprint recognition and face recognition. Currently, these two more mature technologies are used to unlock smartphones (such as fingerprint unlocking that Apple started to try on iPhone5s, and face unlocking by Nubia, etc.). However, Samsung recently officially announced that it will launch its first iris recognition tablet, which is 10,000 times more accurate than a fingerprint scanner. At the same time, Fujitsu's "unlock with a glance" eye pattern recognition phone at MWC was amazing. Microsoft also said that WIN10 will fully support facial, fingerprint and iris recognition, integrating the mainstream biometric technologies on the market, as long as the hardware can collect standard data. For a time, mobile iris payment came to the forefront and sparked widespread heated discussion.

Phenomenon

1. Development opportunities of iris recognition

1. Iris recognition is popular among manufacturers

As we all know, iris recognition technology has long been used in personnel entry and exit identification management systems in military armories, bank vaults, airport internal management, and government departments with important security levels. However, with the popularization and promotion of iris technology, the cost of technology has decreased, and more and more iris equipment manufacturers have begun to enter the civilian market. EyeStart is one of them. In cooperation with Intel, Minsheng Bank and Lenovo, it launched EyeSmart, the world's first iris mobile phone with payment function. In addition, Shenzhen companies Puchuang Tianxin and Fujitsu have successively launched smartphones with iris recognition technology. For a time, mobile phone iris payment came to the forefront and sparked widespread heated discussions. It is also based on this background that some people in the mobile phone industry even believe that iris recognition technology may be the next new point to detonate the smartphone market.

With the serious homogeneity of smartphones, new technical highlights are needed for upgrading. This is also the focus of mobile phone manufacturers. They must stimulate users' purchasing desire through unique functions and experiences. In addition, the current methods of unlocking mobile phones and apps are mainly four-digit password recognition (if fingerprint recognition fails, password recognition is required before access) and gesture recognition, which are extremely low in security. With the increasing privacy of mobile phone information and the connection between mobile phone applications and transactions, payments, and bank securities products, account security, information security, and capital security will require mobile phone manufacturers to provide a higher level of security "fence" to shield the insecurity that may occur at any time.

"Mobile payment requires high security, and the security of fingerprints as identity authentication is not even one ten-thousandth of that of irises. Moreover, in terms of payment authentication methods, users not only want security but also simplicity and convenience. The last and most important step of mobile transactions is to enter the payment account password. Iris recognition payment can save users from the embarrassment of forgetting their passwords. At the same time, irises can completely replace fingerprints to turn on the phone. It is more convenient to unlock the phone and software directly with irises, which meets the needs of lazy people."

(II) Iris recognition has obvious advantages

Iris recognition technology is based on the iris in the eye for identity recognition, and is used in security equipment (such as access control, etc.) and places with high confidentiality requirements. The iris is composed of quite complex fibrous tissue, containing many interlaced details such as spots, filaments, crowns, stripes, crypts, etc. These features are determined in a random combination from the eighth month of the baby's embryo, and once formed, they remain unchanged throughout life. It can be said that the iris is currently the most accurate biometric recognition technology in the world, with excellent anti-counterfeiting properties.

In addition, compared with fingerprint recognition, iris recognition is more accurate and reliable. The false acceptance rate (FAR) and the false rejection rate (FRR) are much lower than those of fingerprints. The feature points of fingerprints are about 50, while the feature points of irises can reach 244 independent free dimensions, which greatly improves the accuracy. In addition, the iris is located between the pupil and the sclera (white of the eye). It is the visible part of the human eye, but it is the only internal organ of the human body that is exposed to the outside. It will not change as much as the skin and facial muscles with age. "

It can be seen from this that iris recognition technology is indeed more accurate than fingerprint recognition technology and face recognition technology.

But this is not all. In terms of scene adaptability, iris recognition technology also shows great advantages. Recently, the Biometrics Center of Carnegie Mellon University has developed a new iris scanning technology that can identify people 40 feet (about 12 meters) away by scanning the iris of the human eye. In the demonstration video of this technology, Marios Savvides, director of the Biometrics Center of Carnegie Mellon University, was sitting in the driver's seat of a small truck. Behind him, a car was a long distance away. His colleague simply used a camera to point at the rearview mirror of the car to identify his identity. The specific operation is that the computer system associated with the camera will compare the picture of Savvides' iris with its data, and then accurately display Savvides' name and other identity information.

It is understood that at present, US law enforcement agencies are now increasingly interested in the application of biometric technology. Seattle recently announced that they will become the largest city in the United States to adopt facial recognition software. Last summer, the US FBI also adopted a similar system.

Finally, regarding the application cost of iris recognition technology, an industry insider said, "Unlike retinal scanning technology, iris recognition only requires a camera to capture the iris part of the eyeball, which is much simpler on a technical level." Therefore, the cost of iris recognition is close to that of fingerprints or even lower.

2. Difficulties and Challenges in the Development of Iris Recognition

After the iris recognition technology is completed and the technology is transferred, it will be applied to prison management, airport security, family planning, anti-terrorism, and foreign ATMs. Iris security phones will also be released. Although iris has achieved good recognition results after years of development, it does not necessarily completely solve the problem of iris recognition, because in complex scenes, especially at long distances, multiple targets, and communication volume, how to accurately identify a person's identity still requires a lot of research.

The future development direction of this system includes users in different postures and speeds, different lighting environments, etc. Especially under network conditions, new opportunities for iris recognition are everywhere, such as realizing iris collection and recognition from three to five meters or even ten meters away, including challenges on the Internet and mobile Internet on mobile phones. There are three challenges: the first is the challenge of convenience. The second is the challenge of robustness. The third is the challenge of security. That is:

How to conveniently obtain iris images of people moving over long distances?

How to robustly recognize iris images from multiple sources?

How to ensure the security of iris recognition systems in ubiquitous network environments?

Although the inherent security of irises is much safer than fingerprints and faces, it is possible that someone can use iris textures or glass eyeballs, or wear invisible glasses to enter the iris recognition system. Therefore, irises need to be imaged from close to long distances, from passive imaging to active imaging, and from static imaging to dynamic imaging. In the future, no matter what the situation is, the irises of multiple people can be collected at the same time, including from long distance to close distance, and even people can present 3D faces, including imaging acquisition and user cooperation. Iris still needs to be transformed and innovated in biometrics, because the objects and distances of iris collection are ever-changing. Therefore, it is difficult to guarantee that high-quality iris images can be collected every time by using a fixed model to deal with changing scenes.

In addition, because the application scope of iris recognition was high-tech but involved very few people, there was no large database for iris algorithm self-learning, that is, a large iris information base. Globally, the largest iris data collection is in India, where about 600 million people have participated in the iris collection of ID cards, accumulating a large amount of iris data and samples in the country, which can better optimize the iris recognition algorithm. However, my country started late in accumulating data in this area and has not yet entered large-scale commercial applications.

"In developed European and American countries, due to human rights considerations and the protection of citizens' privacy information, whether iris recognition can be used on a large scale like in India remains to be discussed. At the same time, Chinese government departments have been using fingerprint recognition in passport approval and other aspects, and it will take some time for users to understand and accept iris recognition.

In addition, it should be noted that the differences in irises of different races also require comprehensive consideration by mobile phone manufacturers, especially those with global strategies. According to a reporter's investigation, Westerners' irises contain very little melanin, making it relatively easy to obtain images, while most Orientals' eyes are black, and some are black with brown. Ordinary cameras do not capture obvious texture features of Orientals. These issues need to be segmented and customized for users in different regions from the hardware equipment to the algorithm level.