The correct sensor technology is key to a successful biometric implementation

It has been suggested recently that the South African biometrics industry is under threat from so-called ‘cowboys’ and fly-by-night operators. It is clear that service providers of this nature would pose an obvious threat to the development of the industry by fuelling negative perception of products and undermining the value of legitimate offerings.

The threat is certainly a significant one, says Nick Perkins, Divisional Director for Identity Management at Bytes Systems Integration. He suggests, however, that the real problem is that the biometrics industry as a whole tends to take the blame when projects fail, and this is often due to something as basic as incorrect sensor selection.

“This is usually caused by a customer not understanding the importance of a sensor that can provide good image capture capabilities and supreme matching performance in real world conditions. To understand the importance of such factors, one need only look to the financial services sector in this country. The banks in SA have selected Lumidigm multispectral imaging technology as the standard financial services industry sensor of choice for interacting with their customers biometrically,” says Perkins.

“These industries, along with telecommunications and retail, develop very large customer databases, and are therefore the sectors that will place the emphasis on highly effective technologies. Where Lumidigm is different from others is in the fact that it works in real world conditions.”

Lumidigm’s multispectral imaging, he explains, looks at and beyond the skin surface to the subsurface foundation of the fingerprint ridges. This subsurface information is both relevant to fingerprint capture and unaffected by surface wear and other environmental factors. The net result, adds Perkins, is that Lumidigm sidesteps the problems that more conventional technologies face when subjected to real world conditions. In addition, the subsurface capability allows the reader to discriminate between a real finger and a ‘spoof’ fingerprint.

Capturing high quality images is vital if one is to avoid what we in the industry call False Acceptance Rates (FAR) or False Rejection Rates (FRR) which occur when a sensor incorrectly accepts or rejects a live image that is being compared against a stored image within the system.”

He points out that in today’s market, a product is often pushed into a customer environment without considering the conditions under which the sensor is expected to perform. For example, the initial captured image in the database may have been taken in an office environment under controlled lighting, whereas the live scan image could be taken outside, with different lighting and complicated further by other factors, such as skin dryness or external moisture. The varying ages of users can also have an impact on a sensor’s ability to scan correctly. Lumidigm’s solution is able to effectively capture images across a broad age range, from children to the elderly.

“Even the de facto standards accepted by the industry are based on older fingerprint sensor technology, commonly called optical, where the sensor is imaging the fingerprint ridge impressions where they make contact with a piece of glass. Clearly, using a method that solely relies on information captured from the surface of the skin is one that can easily be affected by external environmental factors.”

“It is true that standards are often considered as a guide to choosing quality sensors,” continues Perkins, “but an increasing number of customers are now placing the emphasis on performance and waiving the standards that solely rely on surface information.”

“When it comes to high volume, performance-orientated environments, multispectral imaging sensors should always be considered in the selection process, as it is clear that these deliver the best performance in real world environments,” he concludes.