Multi-perspective Finger Vein Scanner

The PLUSVein Finger rotation data set (PLUSVein-FR) has been acquired using a custom designed multi-perspective finger vein scanner depicted in Fig.1. It provides finger vein images all around the finger (360°) with a resolution of 1°. The finger is placed in the center of the scanner (axis of rotation), whereas the NIR camera (right side) and the NIR illumination unit (left side) are placed on opposite sides of the finger (light transmission). While the camera and the illumination module rotate around the finger, the different projections of the finger are acquired.

The illumination module on the left side consists of 5 NIR laser diodes (808 nm) placed on a strip. The images are captured by a NIR enhanced industrial camera (IDS Imaging UI-1240ML-NIR, max. resolution 1280x1024 pixels) with a 9 mm wide-angle-lens (Fujifilm, HF9HA-1b, 9mm, 2/3"). An additional NIR longpass filter (Midopt LP780, useful range: 800-1100mm) mounted on the lens blocks visible light up to a wavelength 780 nm. The rotation is accomplished using a stepping motor (SY42STH47-1684A). The stepper and the rotor are connected via self printed cogwheels having a gear ratio of 1:5/3 (motor to rotor). One step corresponds to 0.0675°, thus it is possible to capture a maximum of 5.333 different projections. The sensor has a size of 25.8 x 32.5 x 45.5 cm (width x height x depth). The rotor has a length of 38 cm.

For more details on the the scanner hardware, the interested reader is referred to:

1. [Prommegger18a      ] Multi-Perspective Finger-Vein Biometrics Bernhard Prommegger, Christof Kauba, Andreas Uhl In 2018 IEEE 9th International Conference on Biometrics Theory, Applications and Systems (BTAS), pp. 1-9, Los Angeles, California, USA, October 22 - October 25

PLUSVein Finger Rotatioin Data Set (PLUSVein-FR)

The data set contains finger images captured from 63 different subjects, 4 fingers (right and left index and middle finger, respectively) per subject, which sums up to a total of 252 unique fingers. As the images are captured all around the finger in steps 1°, there are 361 different perspectives (361 as we captured one frame for 0° and 360°). Every finger was acquired 5 times. This results in 252 * 361 * 5 = 454.860 images in total. One projections consists of 252 * 5 = 1260 images. Some sample images together with their extracted MC features are depicted in Fig.2.

Publications

For more details on the PLUSVein-FR data set and the work we did using this data set, the interested reader is referred to:

1. [Prommegger18a      ] Multi-Perspective Finger-Vein Biometrics Bernhard Prommegger, Christof Kauba, Andreas Uhl In 2018 IEEE 9th International Conference on Biometrics Theory, Applications and Systems (BTAS), pp. 1-9, Los Angeles, California, USA, October 22 - October 25
2. [Prommegger18b     ] Longitudinal Finger Rotation - Problems and Effects in Finger-Vein Recognition Bernhard Prommegger, Christof Kauba, Andreas Uhl In Proceedings of the International Conference of the Biometrics Special Interest Group (BIOSIG'18), pp. 1-11, Darmstadt, Germany, September 27 - 28
3. [Prommegger19a     ] Longitudinal Finger Rotation - Deformation Detection and Correction Bernhard Prommegger, Christof Kauba, Michael Linortner, Andreas Uhl IEEE Transactions on Biometrics, Behavior, and Identity Science 1:2, pp. 123-138, 2019
4. [Prommegger19b   ] Different Views on the Finger — Score-Level Fusion in Multi-Perspective Finger Vein Recognition Bernhard Prommegger, Christof Kauba, Andreas Uhl In Andreas Uhl, Christoph Busch, Sebastien Marcel, Raymond Veldhuis, editors, Handbook of Vascular Biometrics, pp. 261-305, Cham, Switzerland, Springer Nature Switzerland AG, 2019
5. [Prommegger19d     ] Rotation Invariant Finger Vein Recognition Bernhard Prommegger, Andreas Uhl In 2019 IEEE 10th International Conference on Biometrics Theory, Applications and Systems (BTAS), pp. 1-9, Tampa, Florida, USA, September 23 - September 26
6. [Prommegger19e    ] Perspective Multiplication for Multi-Perspective Enrolment in Finger Vein Recognition Bernhard Prommegger, Andreas Uhl In Proceedings of the International Conference of the Biometrics Special Interest Group (BIOSIG'19), pp. 107-117, Darmstadt, Germany, September 18 - 20
7. [Prommegger20a    ] Advanced Multi-Perspective Enrolment in Finger Vein Recognition Bernhard Prommegger, Andreas Uhl In Proceedings of the 8th International Workshop on Biometrics and Forensics (IWBF'20), pp. 1-6, Porto, Portugal, April 29 - April 30
8. [Prommegger20b   ] A Fully Rotation Invariant Multi-Camera Finger Vein Recognition System Bernhard Prommegger, Andreas Uhl IET Biometrics 10:3, pp. 275-289, 2021
9. [Prommegger20c     ] Rotation Detection in Finger Vein Biometrics using CNNs Bernhard Prommegger, Georg Wimmer, Andreas Uhl In Proceedings of the 25th International Conference on Pattern Recognition (ICPR), pp. 6531-6537, 2020
10. [Prommegger21a    ] Rotation Tolerant Finger Vein Recognition using CNNs Bernhard Prommegger, Georg Wimmer, Andreas Uhl In Proceedings of the IEEE 20th International Conference of the Biometrics Special Interest Group (BIOSIG 2021), pp. 1-8, Darmstadt, Germany, September 16 - September 18

Filename and Directory Structure

Directory Structure:
[Dataset]/[Perspective]/[SubjectID]/

Filename:
[SessionID]_[SubjectID]_[FingerID]_[ImageID]_[Perspective].png

Placeholder:

• [Dataset]: Name of the data set (PLUSVein-FR, PLUSVein-FR-ED, PLUSVein-FR-ND)
• [SessionID]: The ID of the session as a 2-digit numer with leading zeros.
• [SubjectID]: The ID of the subject as a 3-digit numer with leading zeros, starting with 001.
• [FingerID]: The ID of the finger as a 2-digit numer with leading zeros.
  • 02: left index finger
  • 03: left middle finger
  • 07: right index finger
  • 08: right middle finger
• [ImageID]: The number of the sample as a 2-digit numer with leading zeros, starting with 01.
• [Perspective]: The angle of the perspective in degree as a 3-digit numer with leading zeros in the range of 000 to 360.

Example:
The 2nd sample of the right index finger of subject #3 at 40° acquired in the 1st session of the PLUSVein-FR data set:
PLUSVein-FR/040/003/01_003_07_02_040.png

Obtaining the Database

• Subset: ±45° around the palmar view

  This subset consist of the perspectives of ±45° around the palmar view in steps of 1°. It contains finger images captured from 60 different subjects, 4 fingers per subject, which sums up to a total of 240 unique fingers. Each finger is acquired 5 times. This results in 91 perspecitves, each containing 1.200. In total that sums up to 109.200 images. The data set is publicly available for research purposes.

  Download Information:

  We are currently reviewing the distribution of the data set in view of GDPR compliance. In the meantime, we cannot give access to the PLUSVein Finger Rotation data set to requests from data controllers or processors established outside the European Union. We will update this page once new procedures have been established.

  For the user from the EU:
  To obtain the subset, you have to agree to our license agreement:
  PLUSVein-FR: ±45° around the palmar view - Consent Form
  

  Please download, fill in and sign the license agreement and send it to A. Uhl or via mail to our department. After checking the license agreement you will be provided with a download link for the data set.

  Requested Citation Acknowledgment

  [Prommegger19a     ] Longitudinal Finger Rotation - Deformation Detection and Correction Bernhard Prommegger, Christof Kauba, Michael Linortner, Andreas Uhl IEEE Transactions on Biometrics, Behavior, and Identity Science 1:2, pp. 123-138, 2019

• Subset: Longitudinal finger rotation validation data set

  In order to be able to test the robustness of a recognition scheme against longitudinal finger rotation, data sets that depict realistic scenarios regarding finger rotation are needed. Such data sets must satisfy the following characteristics:

  1. The data set needs to provide finger vein images from perspectives spread over the desired range.
  2. The distribution of the rotation angles must follow the characteristics of the desired scenario.
  3. It needs to contain enough longitudinal rotation in order that a rotation compensation is useful.
  4. Ideally, also the rotation angles of the different samples are known.

  Therefore, two data sets are generated from the publicly available subset (±45° around the palmar view) of the PLUSVein-FR. The first data set, PLUSVein-FR-ED, contains vein images whose rotation angles are equally distributed over the entire range of ±45°. It corresponds to the unconstrained placement of the finger in a contact-less acquisition system. The rotation angles of the second data set, PLUSVein-FR-ND, are normally distributed. This data set models a realistic real world scenario of a classical unsupervised single perspective acquisition system. Prommegger \etal estimated the rotation angles of different finger vein data sets in [Prommegger19c]. The SDUMLA-HMT (external link) exhibited the highest degree of finger rotation with rotation angles up to ±45° (σ = 10.6°). This standard deviation was used for the generation of the PLUSVein-FR-ND. The distributions of the rotation angles of the two subsets are depicted in Fig.2, some statistical data in Tab.1.

  

  Tab.1 - Statistical data of PLUSVein-FR-ED and PLUSVein-FR-ND.

  Dataset	PLUSVein-FR-ED	PLUSVein-FR-ND
  Subjects	60	60
  Fingers	4	4
  Samples	5	5
  Images	1200	1200
  Images	1200	1200
  Distribution	equal	normal μ = 0.03°, σ = 10.6°
  Min rotation angle	0°	0°
  Max rotation angle	45°	40°
  Min dist b/w samples	0°	0°
  Avg dist b/w samples	29.33°	12.5°
  Max dist b/w samples	89°	55°

  Download Information:

  We are currently reviewing the distribution of the data set in view of GDPR compliance. In the meantime, we cannot give access to the PLUSVein Finger Rotation data set to requests from data controllers or processors established outside the European Union. We will update this page once new procedures have been established.

  For the user from the EU:
  To obtain the subset, you have to agree to our license agreement:
  PLUSVein-FR: Longitudinal finger rotation validation data set - Consent Form
  

  Please download, fill in and sign the license agreement and send it to A. Uhl or via mail to our department. After checking the license agreement you will be provided with a download link for the data set.

  Requested Citation Acknowledgment

  [Prommegger19d     ] Rotation Invariant Finger Vein Recognition Bernhard Prommegger, Andreas Uhl In 2019 IEEE 10th International Conference on Biometrics Theory, Applications and Systems (BTAS), pp. 1-9, Tampa, Florida, USA, September 23 - September 26