An Israeli company has developed a biometric process that identifies individuals based on their eye movements. It’s supposed to be more difficult to deceive than analyzing the retina or iris.
Each of us sees the world through different eyes. Leveraging this simple insight, the Israeli company ID-U Biometrics is now introducing a new biometric process: it identifies individuals not by iris or retina, but by the movements of the eyeball itself. Because these movements are also characteristic of each individual, the process aims to increase reliability in biometric identification and be even more difficult to deceive than other methods, promises ID-U Biometrics. «The interface is really quite simple,» explains Daphna Palti-Wasserman, CEO of the company, the principle. «The person looks at a symbol moving across the screen, while a camera records the eye movements.»
Even if the symbol moves in a different pattern each time, this analysis should reveal the viewer’s identity. ID-U Biometrics claims to have identified «kinetic peculiarities» unique to each eye.
Special hardware is not required for the process. Because only a screen and a camera are needed, it is cheaper and easier to install than other systems, says Palti-Wasserman. With just a standard video camera, the system achieves a 97 percent success rate. Many mobile phones and laptops already have suitable cameras built in, she adds. Therefore, ID-U’s process could be used on mobile devices in the future. The company is currently developing a corresponding app for the iPhone 4.
ID-U Biometrics aims to address weaknesses in existing biometric systems with its technology. Because fingerprint and retina patterns can be copied to deceive authentication systems. Eyeball movements, however, cannot be copied, assures Palti-Wasserman.
«We use a challenge-response sequence,» she adds. «With ‘challenge-response,’ an authentication process is employed where a task (challenge) is presented that only the person with specific knowledge can answer (response).» «The entire process depends only on what is shown on the screen,» says Palti-Wasserman.
A challenge-response approach is already used in some security processes that analyze keystrokes, handwriting, or voice, notes Kevin Bowyer, a biometrics expert at the University of Notre Dame in Chicago. «The key advantage is that you have even more control over the authentication situation to achieve better and more reliable results.»
In the ID-U system, the movement of the symbol is designed to query a dozen of the «kinetic peculiarities» of an eye. Its movements are captured 30 times per second. «When the target symbol jumps and the eye wants to follow, it uses a different motion mechanism than when it follows a slowly gliding symbol,» Palti-Wasserman vaguely describes the concept. She refuses to disclose details about the kinetic measurements. An analogy would be how two people drive along the same road: the trajectories they take are never identical.
To prove the reliability of the system, it must be tested on hundreds of subjects, and the tests should be conducted at significant intervals, Bowyer insists. «It’s common for biometric systems to perform great when multiple recordings are made of subjects in the same test period,» Bowyer explains his skepticism. «But when you take the same number of recordings at intervals of weeks or even months, the success rate drops drastically.»