Smart Sensors Limited, a spin-out from the University of Bath, UK, has introduced version 2 of its MIRLIN software toolkit, enabling security system developers to create iris recognition systems capable of two million matches per second, four times faster than other current technologies. Functions include image acquisition, enrollment, and verification or determination of identity (ID) from human irises.
Analyzing a digital image of the iris using techniques similar to those employed in video encoding, the Smart Sensors patented solution requires less processing power than alternative methods and the software code can operate in only around 300 Kilobytes of memory on a silicon chip, according to the developers. As a result, systems can be based on small, low-cost microcontrollers with less processing power than an ordinary personal computer. This makes the technology suitable it’s claimed for use in fixed or battery-operated, handheld equipment.
Smart Sensors’ MIRLIN software development kits (SDK) enable systems to be optimized for the best combination of speed and accuracy. Images can be scanned in segments from 18 degrees to 0.72 degrees wide, depending on the accuracy required. The technology works with iris characteristics that generally conform to ISO/IEC standards but it is capable of delivering accurate results with images that are significantly poorer than those defined by the standard.
The SDK can work with a variety of image sources, including several manufacturers’ iris cameras. Versions are available to support various processors and operating systems. To make licensing more attractive for customers, the company only charges a fee for each product or each server on a network, not for every user or enrollment.
Customers are supported through service level agreements that provide bug fixes and product enhancements. Smart Sensors, which is based in the SETsquared business incubator at the University of Bath, has recently been awarded the Frost & Sullivan "Global Iris Biometric Systems Technology Innovation of the Year Award" for 2009.
