Capture methods for fingerprint biometrics continue to evolve, and 3D contactless fingerprint images measured with structured light show potential to address some of the field’s lingering challenges, attendees heard last lunch of the European Biometrics Association (EAB).
IDloop Research Manager Dr. Tom Michalsky presented the presentation on “3D imaging of biometric characteristics of human skin”. IDloop manufactures contactless biometric devices to enable faster workflows and lower maintenance costs.
Michalsky reviewed the standard method of biometric contact fingerprint collection through frustrated total internal reflection. This is the usual scanner architecture, based on a light source, a prism, a lens and a camera. TFT sensors are also becoming more familiar and have the advantage of providing the correct scale and avoiding prisms, but according to Michalsky they retain other disadvantages of contact scanners, such as leaving latent prints and the possibility of bacterial transmission.
The development of the 2D contactless fingerprint has followed and shows better usability compared to contact scanners. Recent research explores the problem of scale ambiguity as a function of the distance from the hand to the camera lens. Despite attempts to compensate for this scaling issue, smaller fingers tend to offer a lower match rate with non-contact images. 2D non-contact fingerprints can also be affected by perspective distortions, Michalsky says, with side views changing the distance between hairlines and details in ways that can’t be corrected without more information.
Michalsky also discussed what he calls “2.5D imaging” for fingerprints, which is based on an approximate 3D structure of the finger, which solves some but not all of the challenges of 2D imaging.
IDloop’s non-contact optical technology, on the other hand, uses absolute scale without perspective distortion. Images are created with structured light, projected onto the subject. This approach provides sub-10 micrometer resolution, he says.
Other common 3D measurement techniques such as stereo vision and time of flight were also described in the presentation. Stereo vision cannot be used for 3D fingerprint imaging, due to its mismatch with periodic structures, while time-of-flight is limited to a resolution of about 1 cm depth.
Structured Light Fingerprint and 3D Impression Biometrics
The development of structured light dates back to the 1960s, but recent advances in applying the technology to biometrics have addressed specific challenges. An example involves artifacts caused by the minute movement of a slightly shaking finger as its image is captured. Reconstruction algorithms are used to perform motion correction in the IDloop system.
Structured light systems must also be calibrated very finely, and the researchers found the standard calibration technique lacking, with unacceptable levels of distortion.
Michalsky dates only this year the real 3D fingerprints without contact and without movement artifacts. Pores, a third level feature, can be measured, in addition to the more easily visible biometric features.
The researcher also shared data on the structure of the ridge and valley of the fingerprints of a two-year-old subject, which shows the continued difficulty of biometric verification of young children, but also suggests that clearer images may be easier to capture with non-contact 3D imaging than other methods.
According to Michalsky, the true 3D contactless fingerprint allows local curvature to be used as an additional data point for better matching accuracy.
Michalsky suggests some advantages of the method in application. For the identification of small children, fingerprints taken with the same method could possibly be used even to reunite kidnapped babies with their parents. It could also be used in presentation attack detection, where it would easily identify printed spoofing artifacts.
3D | biometric matching | biometrics | biometric research | contactless | eBike | European Association for Biometrics | fingerprints | LoopID