NEXT Biometrics CEO says sensor size does matter

June 10, 2015 - 

NEXT Biometrics recently commissioned the largest, commercial fingerprint sensor test ever conducted, which found that fingerprint sensor accuracy directly correlates to the size of sensors used.

The report, conducted at the University Carlos III of Madrid noted that the smaller the image collected by a fingerprint sensor, the larger the rejection during acquisition, and the higher the error rates generated by the device.

The study conducted by Professor Raul Sanchez-Reillo, Director of GUTI, known as the University Group for Identification Technology at University Carlos III of Madrid, undertook a detailed analysis of image quality and biometric performance of small fingerprint sensors. The so-called “Madrid Study” is the largest known mass market test of fingerprint sensors ever conducted. The study, conducted over four months, tested a randomized sequence of sensors using over 580 people and over 180,000 prints, along with ISO-based best practices. The origin of the sensors were kept confidential to both operators and subjects and all test subjects made two visits, separated by at least two weeks.

“This university study supports what biometric experts around the world have said for years,” stated NEXT Biometrics CEO Tore Etholm-Idsoe in an exclusive interview with BiometricUpdate.com. “Size is a key factor in fingerprint sensor accuracy.”

With the consumer electronics market focused on miniaturization, especially involving mobile devices, such as smartphones and tablets, the biometrics industry has become fixated on designing and producing smaller and smaller sensors. The university study however conclusively and persuasively demonstrates the paradox that smaller sensors lead to less accurate results.

“A reliable fingerprint sensor system cannot compromise on sensor area,” notes Etholm-Idsoe. “Laws of nature applies in fingerprinting. A small sensor will capture a small image containing a small number of time constant data only.”

According to Etholm-Idsoe, the study shows that if the acceptable rate for false acceptance is set at one in 10,000, a sensor size of 200 square millimeters will result in less than one percent false rejections when used with a high-end algorithm. Comparatively, if the sensor’s active area is reduced in half to 100 square millimeters, then the average false rejection rates increase significantly to more than five percent. The study also shows that considerably more users will be rejected during enrollment when using a small area sensor.

“In simple terms, smaller isn’t better,” said the NEXT Biometrics CEO. “Often, smaller sensors are close to useless and less than five percent of sensors sold are actually used by end customers.”

The “Madrid Study” also found that sensor accuracy further declines if dissuasive factors are added in. “Adding real life complexities like wet, dirty or worn fingers further adds to everyday challenges resulting in poor user experience and low security levels,” says Etholm-Idsoe.

The NEXT Biometrics CEO therefore argues that smaller sensors are insufficient to secure more complex applications. “Our industry is trending from the simple unlocking of phones towards more serious payment, cloud and corporate access applications. As this happens, the importance of sensor system reliability will grow further.”

The main study recommendation is thus to avoid using small size sensors. In case such sensors are used, then two recommendations should be followed. The first one is to improve the training of the user in interacting with the sensor, as to reduce biometric failure rates, known as failure to enroll (FTE) and failure to acquire (FTA). The second recommendation is to complement the recognition process with other mechanisms, in order to improve accuracy during operational recognition processes. Another recommendation is that the industry should move towards standardization.

“It is important that industry works together to have biometrically sounds products with applicable and appropriate industrial design,” Etholm-Idsoe stated. “It is our intention to work to promote greater standardization in order to get end-users on board with end specifications.”

“Going forward, NEXT will support specification standardization and other such initiatives. Standards and third-party verification of sensor supplier marketing claims are important for the long term success of our industry.”

Leave a Comment

comments

About Rawlson King

Rawlson O’Neil King is a contributing editor at BiometricUpdate.com and is an experienced communications professional, management consultant, trade journalist and author who recently published a book about control and electronic networks and who has written numerous articles in trade publications and academic journals about smart home and building technologies. Follow him @rawlsonking2.