July 19, 2006 — A scanner that reads a fingerprint as well as the unique pattern of tissue and blood content beneath the skin could offer higher reliability for biometric security.
The paper-thin sensor, being developed by Nanoident Technologies, based in Linz, Austria, could be on the market in one to two years as a safe and secure way for accessing sensitive data on smart cards, cell phones, and other electronic devices.
According to Klaus Schroeter, CEO and founder of Nanoident, conventional finger scanners typically read just the pattern of the surface print. That has two big disadvantages: the recognition accuracy is only about 97 percent and the scanners can be fooled by fake fingers posing as the real deal.
But a scanner that captures both the surface and subsurface structures improves the recognition accuracy to 99 percent and is nearly impossible to fool.
"If you can measure the hemoglobin content, you can say this is a live finger," said Schroeter.
The flexible sensors — just two millimeters wide and 12 to 15 mm long — are produced at a cost of one to two dollars apiece using inkjet technology. A semi-conducting, ink-like liquid is printed onto a polymer or glass substrate to form light emitters, light sensors, and the circuitry necessary to process the electronic signals.
Similar components made with rigid silicon semiconductors can run between three and ten dollars apiece.
The light emitters on Nanoident's sensor illuminate the finger with different wavelengths of light: blue for the surface print and infrared to penetrate three to four mm deep into the finger's substructure.
The light sensors measure the amount of light reflected back and a proprietary software algorithm produces the patterns unique to the finger.
When incorporated into a smart card or electronic device, such as a personal digital assistant, the user would first record their unique print and then store it into the device's memory.
Later, when wanting to access sensitive information, the user would place or swipe their finger over the sensor to confirm their identity. If the print did not match what was saved in the system, the person would not gain access.
"It's a very advantageous approach," said Arun Ross, assistant professor of computer science and electrical engineering at West Virginia University in Morgantown and a researcher at the Center for Identification Technology Research.
But, he said, gathering more information puts greater demands on the system to cancel out unwanted "noise" — such as physical flaws in the finger or environmental variables including humidity — and combine the data into a meaningful pattern.
"More information does not translate into accuracy unless there is fusion," he said.
Schroeter said the company will have its first working prototypes by the middle of next year and is currently working a new sensor that would be incorporated into the screen of an electronic device.
