US financial technology company NCR introduced various forms of new banking technologies, including an ATM that uses iris recognition to identify customers, to the South African financial industry last month.
“With telecommunications and information technology advancing at rapid speeds, it is important for financial institutions to employ the latest technology to ensure they remain competitive.
“The increased use of the Internet will move the management of banking transactions away from financial institutions to the customers themselves, and introducing systems which will serve clients better will determine future business levels,” reports NCR self-service marketing and strategy vice-president Chris Hughes.
“With more than 85 000 ATMs dealing with an estimated US$2,6-billion in cash every day around the world, it is evident that bank customers prefer self-service above teller transactions.
“We believe that technology, such as iris recognition, will encourage more people to use ATMs knowing that no one else can access their accounts.
“International research has also indicated that customers prefer a technology where they do not need to touch the screen, such as is the case with fingerprint recognition,” says Hughes.
He tells Engineering News that the company obtains the iris recognition technology employed in one of its new products, from Sensar, the first commercial spin-off of the Sarnoff Corporation in Princeton, New Jersey, US, an internationally acclaimed research and development facility.
Sensar is licensed by US company Iriscan,which holds the exclusive worldwide patents on iris recognition.
The technology to read and map the data in a person’s iris was developed by Dr John Daugman of Cambridge University, UK, and patented in 1994.
Iris identification is a highly accurate, easy-to-use and virtually fraud-proof means to verify a customer’s identity through what is known as biometrics.
Biometrics is a group of technologies that identify or verify individuals based on physiological characteristics.
These technologies include products that recognise faces, hands, fingers, signatures, voices, fingerprints and irises.
Computer technology is used in non-invasive ways to match the patterns of individuals in real time against enrolled records.
Sensar’s products use standard video cameras and state-of-the-art real-time image processing to acquire a picture of a person’s iris, digitally encode it, and compare it with one already on file.
Research shows the iris is the most unique data-rich physical structure in the human body – substantially more unique than a fingerprint.
The iris has 266 measurable characteristics while a fingerprint has rough ly 35.
Research shows the matching accuracy of iris recognition to be greater than that of DNA testing.
The probability that two irises could produce exactly the same code is about 1 in 1 078 (the population of the earth is around 1 010).
The random patterns of the iris are the equivalent of a complex human barcode, created by a tangled meshwork of connective tissue and other visible features.
The video cameras can read through all types of glasses, contact lenses and sunglasses (except mirror sunglasses) to obtain a picture of a customer’s eyes.
Enrolling a new client on the system is easy and takes about 30 seconds, explains NCR global sales support manager Roger Beecroft.
“A person is requested to look at a central point from where the cameras can take the best pictures.
“The iris is then scanned and the probability of a similar one calculated,” says Beecroft.
This information is then either stored on a central computer network or on a smart-card.
The iris record for a bank customer, for example, requires less than 512 bytes an eye for storage while other personal electronic identification products, such as fingerprint and voice print, require record sizes of 1 000 bytes a finger or more.
If the consumer uses smart-card technology, the customer profile could be saved on the card itself and held by the customer.
The technology would then cross-reference the picture of the iris with the data saved on the card, eliminating the need to use a network infrastructure to access a database on a host computer.
Hughes believes that these systems could have a substantial market in South Africa.
“However, initially it will be expensive for financial institutions to install these systems, but I believe that once they become more widely used, the users will experience great benefits,” says Hughes.
Iris recognition systems are used at the Nationwide Building Society in the UK as well as at the 1998 Winter Olympics in Nagano, Japan, to guarantee secure access for professional athletes.
Systems such as ATMs using iris recognition are developed at the company’s Advanced Solutions Concepts Laboratory in Dundee, Scotland.
Other concept systems which have been developed include ‘Stella’, an ATM using iris recognition and voice recognition, as well as a microwave oven which can be used to not only prepare supper, but undertake banking transactions.
“However, these are only concept products and it will probably take some time before they will be introduced commercially,” stresses Hughes.
First National Bank ATM deployment manager Craig Anderson tells Engineering News that the company is investigating the installation of ATMs using biometrics for identification.
“We will definitely use these types of ATMs in future to improve security, but first there are various hardware issues to be sorted out before we can implement these machines,” says Anderson.
ABSA assistant GM of self-service channels Jan de Villiers agrees with this, saying biometrics is a definite consideration for the future.
“However, we will have to investigate the cost of delivery since it is still too expensive to install.
“Another important issue is that the various banks in South Africa will h ave to get together and agree on one general system otherwise ATM transactions using a different bank’s ATM will not be possible.
“Different card standards, such as smart-cards will have to be investigated as well, since the data transfer over a network will probably slow down transactions,” says De Villiers.
He also noted that no study has been done to determine South Africans’ perception of biometric systems as yet and if customers will be willing to accept and use such ATMs.
Chris Hughes Roger Beecroft Iris recognition technology profile Probability of two irises producing the same code: 1 in 1 078 Independent variables extracted: 266 Iriscode size: 512 bytes Average recognition speed (database of 100 000 iriscodes): Identification – 2 seconds; verification – 3 seconds Database limitations: There are no operating limits on the size of the database Source: Sensar