Interoperability a challenge for touchless fingerprint biometrics, researcher tells EAB
Touchless fingerprint biometrics have reached good levels of useability and user acceptance, but challenges relating to performance and interoperability with touch-based fingerprint systems require further research, according to Jannis Priesnitz of the National Research Center for Applied Cybersecurity. The European Biometrics Association (EAB) presented Priesnitz’ talk based on his paper providing ‘An Overview of Touchless 2D Fingerprint Recognition’.
A review of the literature on touchless fingerprint biometrics shows that interest in the technology has increased steadily, Priesnitz notes.
Touchless fingerprint technology has a significantly different workflow from traditional fingerprint biometrics, starting with image enhancement to enable the characteristics of the fingerprint to be extracted. Fingerprint detection, alignment and quality assessments are also often carried out before feature extraction and biometric comparison.
Acquisition devices are divided into dedicated hardware systems, which offer some distinct advantages, and typical image capturing devices like webcams and smartphone cameras, where processing can sometimes be carried out locally, but challenges such as related to lighting are more likely to arise.
From there, Priesnitz explains measures taken to clarify features and exclude unclear image areas.
The NIST Fingerprint Image Quality (NFIQ) 2.0 algorithm can be used to assess contactless fingerprint images, according to a recent paper by Priesnitz. Legacy feature extractors are typically also capable of working with contactless fingerprints, but tend to deliver worse performance than dedicated touchless fingerprint extractors. Likewise, dedicated touchless comparison algorithms tend to outperform general fingerprint biometric comparison algorithms.
One challenge in touchless fingerprinting is providing the positioning guidance for the sensor area, which in touch-based smartphone systems is fairly effective and convenient for users.
Various image quality concerns were discussed, along with sensor interoperability and different camera qualities.
Presentation attack detection (PAD) is more difficult with touchless systems, Priesnitz says, less because of the wider possible range of spoof attacks, including photographs and videos, than because the lack of touch against a surface prevents the use of some effective fingerprint spoofing detection methods.
Several questions related to measuring fingerprints captured with smartphone cameras and aligning them with the 500 dpi standard for touch fingerprints.
On the business side, Veridium recently struck a partnership with Jumio to integrate touchless fingerprints and face biometrics for remote identity verification in April.