DIGITAL

Scientific publication

Photogrammetric Fingerprint Unwrapping

Publication from Digital

Gerhard Paar, Maria del Pilar Caballo Perucha, Arnold Bauer, Bernhard Nauschnegg

8th ISPRS Conference on Optical 3-D Measurement Techniques, Vol.II, , July 9-12, 2007, Zurich, Switzerland. , 2007

Abstract:

Fingerprints are among the most important biometric cues for forensics,
 access control, health care, security and other authentication applications.
 Many fingerprint sensors acquire only a sub-region of the relevant
 part of the finger, in most cases the finger undergoes a deformation
 due to necessary touching of the sensor. Using a contactless strategy
 with more than one overlapping camera images for photogrammetric
 acquisition of the proximal phalanx (front-most finger segment) has
 a set of advantages and enables to perform objective fingerprint
 unwrapping for new and more robust fingerprint matching algorithms
 and further exploitation of the 3d surface of the finger. We report
 on a photogrammetric workflow for nail-to-nail fingerprint reconstruction
 for the purpose of virtual unwrapping. A calibrated sensor setup
 consisting of 5 cameras and dedicated illumination provides a data
 base with 4 stereo pairs of adjacent camera images. The finger silhouette
 images are used for carving a rough 3d finger model used as prediction
 for stereo-matching based refinement. The final proximal phalanx
 reconstruction consists of a principle finger axis and a digital
 surface grid containing perpendicular distances from the axis. For
 model texture a weighted combination of shading corrected images
 from the two closest cameras are used for each model grid pixel.
 Due to the knowledge about 3d position of each texture point the
 implementation of different unwrapping methods is possible. Further
 exploitation of the reconstruction will be more robust fingerprint
 detection methods, both by making use of the 3d data, and utilizing the comprehensive

Keywords: Photogrammetry, Biometrics, Fingerprint Unwrapping