ICFHR 2010 Signature Verification Competition (4NSigComp2010)

From TC11
Jump to: navigation, search

Datasets -> Datasets List -> Current Page

Created: 2012-06-22
Last updated: 2012-006-25

Contact Author

Dr. phil. nat. Marcus Liwicki
DFKI - German Research Center for Artificial Intelligence
Trippstadter Str. 122
D-67663 Kaiserslautern, Germany
E-mail: liwicki@dfki.uni-kl.de
Phone +49 (0) 631 20575 1200
Fax +49 (0) 631 20575 1020

Current Version

1.0

Keywords

Forensic signature analysis, offline handwriting, signature, verification

Description

Sample signatures from the 4NSigComp2010 dataset.

The collection contains offline signature samples. The signatures were collected under the supervision of Bryan Found and Doug Rogers in the years 2002 and 2006, respectively. The images were scanned at 600dpi resolution and cropped at the Netherlands Forensic Institute for the purpose of 4NSigComp2010 signature verification competition.

Technical Details

The La Trobe signature collection for training contains 209 images. The signatures comprise 9 reference signatures by the same writer “A” and 200 questioned signatures. The 200 questioned signatures comprise 76 genuine signatures written by the reference writer in his/her normal signature style; 104 simulated/forged signatures (written by 27 forgers freehand copying the signature characteristics of the reference writer); 20 disguised signatures written by the reference writer. The disguise process comprises an attempt by the reference writer to purposefully alter his/her signature in order to avoid being identified or for him/her to deny writing the signature. The simulation/forgery process comprises an attempt by a writer to imitate the reference signature characteristics of a genuine authentic author.

The La Trobe signature collection for testing contains 125 signatures. The signatures comprise 25 reference signatures by the same writer “B” and 100 questioned signatures. The 100 questioned signatures comprise 3 genuine signatures written by the reference writer in his/her normal signature style; 90 simulated signatures (written by 34 forgers freehand copying the signature characteristics of the reference writer); 7 disguised signatures written by the reference writer. All writings were made using the same make of ballpoint pen and using the same make of paper.

Data Acquisition Details

Training Set

Collection of genuine signatures

The following signatures were supplied by the reference writer “A” for the training set:

  • 15 normal signatures per day over a seven day period; 9 signatures were chosen from this subset as reference set to which the questioned signatures are to be compared.
  • 6 disguised signatures per day over a seven day period. In addition to these signatures, the reference writer provided an additional 81 genuine signature samples (27 pages containing three signatures per page). Signatures from this supplementary pool were provided to the forgers as examples of the signature they were required to forge.

Collection of simulated/forged signatures

For the training set, 27 adult forgers were applied to forge the genuine signatures of writer “A”. The forgers were volunteers drawn from groups such as secondary school teachers and professional organizations. Each of the forgers was provided with 3 normal samples of the signature written by the reference writer. Forgers were instructed that they could use any or all of the supplied reference signatures as models for their simulations. Forgers were also instructed that their simulations must be unassisted (not tracings). Each forger was asked to complete the following task:

  • Inspect the genuine signature and, without practice, immediately attempt to forge it 3 times.
  • Practice simulating/forging the genuine signature 15 times then simulate/forge the signature an additional 3 times.

Test Set

Collection of genuine signatures

Similar to the training set data collection, another reference writer “B” provided a set of signatures over a five day period; 25 signatures were chosen from this subset as reference set to which the questioned signatures are to be compared. The test data contain 3 genuine and 7 disguised signatures.

Collection of simulated/forged signatures

The test data contain 90 simulated signatures. For the generation of simulated signatures, 34 adult forgers were applied. These individuals were volunteers. The forgers were either ’lay’ persons or calligraphers.

Folder Structure and File Naming

The signatures of the training set are arranged according to the following folder structure:

  • Disguised: Contains all the disguised signatures of specimen author ‘A’
  • Genuine: Contains all the genuine signatures of specimen author ‘A’
  • Reference: Contains the reference signatures of specimen author ‘A’ which can be used for training the classifiers for author ‘A’.
  • Simulated: Contains all the skilled forgeries for specimen author ‘A’

Note that the naming convention also reveals the type of signature for the training set. For example, D023: a disguised signatures, G003: a genuine signature, S001: a forged/simulated signature, and similarly R001: a reference signature. The signatures of the test set are arranged according to the following folder structure:

  • Reference: Contains the reference signatures of another specimen author ‘B’. These signatures are used to train the classifier for verifying authorship of the specimen author ‘B’.
  • Questioned: Contains all the signatures for which the task is to verify authorship of the specimen author ‘B’.
  • Correct Answer Key: The key containing details about the type of signatures for specimen author ‘B’, i.e., disguised, forged, or genuine.


References

  1. Marcus Liwicki, Elisa van den Heuvel, Bryan Found, Muhammad Imran Malik. "Forensic Signature Verification Competition 4NSigComp2010 – Detection of Simulated and Disguised Signatures", Proc. 12th Int. Conference on Frontiers in Handwriting Recognition, 2010

Disclaimer

The data of the 4NSigComp2010 may only be used for research purpose. Users are not allowed to add images of the signatures into their publications. Furthermore, any scientific work using these signatures should refer to the abovementioned Reference 1.

Submitted Files

Version 1.0

Files

Note: the password for opening the zip files is "I hereby accept the 4NSigComp2010 disclaimer." (without the double quotes).


This page is editable only by TC11 Officers .