Abstract: This website describes the project fingerprint conformance ground truth database (GTD), which has been initiated by members of the ISO SC37 and the German Federal Criminal Police Office (BKA). The conformance ground truth database shall serve for developing and conducting semantic conformance tests as well as biometric sample quality assessment.
Fingerprint Conformance Ground Truth DatabaseSemantic Conformance Testing: The deficiencies of current fingerprint minutiae extractors and the impact on system interoperability have been identified in the research report (DeliverableD62.pdf) of the MTIT project. The Working Group 3 of ISO/IEC JTC1 SC37 is now working on methodologies for semantic conformance testing of standardized biometric interchange records. In this scope the project ISO/IEC 29109-2 AMD1 was started that is focused on semantic conformance testing of fingerprint minutiae data as defined in ISO/IEC 19794-2:2005. While conformance testing on a syntactical level has been formulated in the modality specific assertion test with ISO/IEC 29109-2 , there is yet no established testing methodology that could attest that a compact minutiae interchange record (minutiae template) is indeed a faithful representation for the input signal. Thus the development of a testing methodology can benefit from ground truth data being collected by forensic experts. An overview of the GTD project and its intentions is given in the concept paper.
GTD Database Segments: In order to satisfy the needs for semantic conformance testing and sample quality assessment different database segments are processed. The first segment is a composite of the NIST sepcial database SD14 and SD29 and consists of approx. 5000 images. The second segment shall be a segment of latent prints from crime scenes with low image quality. The third segement shall consist of liver prints and be based on the Bologna fingerprint data FVC2000 (Db1, Db2, Db3) FVC2002 (Db1, Db2, Db3) and FVC2004 (Db1, Db2, Db3).
Minutiae Mark-Up Process: For the first database segment approx. 1000 images have been marked-up by three dactyloscopic experts as of February 2010. Further experts opinions are seeked and forensic institutes are invited to contribute and to join the GTD project. The mark-up work is conducted with a dedicated user interface (GTD-GUI) collecting all relevant meta data such as minutia orientation, type and quality. The GTD-GUI and the GUI-Tutorial can be downloaded from this website. The GUI software is provided as is and no guarantee is given for this software.
Publications and further Information:
- T. Mansfield, A. Bazin: Research report on minutiae interoperability tests, MTIT project, Deliverable D6.2, (2007)
- C. Busch: Generating Ground Truth Data for Fingerprint Minutia Conformance Testing, GTD concept paper, (2009)
- C. Busch, D. Lodrova, E. Tabassi, W. Krodel: Semantic Conformance Testing for Finger Minutiae Data, in Proceedings of the IEEE IWSCN, pages 17-23, (2009)
- D. Lodrova, C. Busch, E. Tabassi, W. Krodel, M. Drahansky: Semantic Conformance Testing for Finger Minutiae Data, in Proceedings BIOSIG 2009, pages 31-42, (2009)
- S. Abt, C. Busch, C. Nickel: Applikation des DBSCAN Clustering-Verfahrens zur Generierung von Ground-Truth Fingerabdruck-Minutien, in Proceedings BIOSIG 2010, pages 95-106, (2010)
- S. Abt, C. Busch, H. Baier: A Quality Score Honoring Approach to Semantic Conformance Assessment of Minutiae-Based Feature Extractors, in Proceedings BIOSIG 2011, pages 21-32, (2011)
Forensic Institutions involved in the Markup of Fingerprint Images: The GTD project is kindly supported by the following institutions:
Academic Institutionn involved in GTD: The GTD project is kindly support by the following institutions:
- Fraunhofer Institute for Computer Graphics Research (IGD)
- Norwegian Information Security Laboratory (NISlab)
- Brno University of Technology (BUT)
Further question: Please contact email: christoph.busch igd.fraunhofer.de