PILOT STUDY - A New Experiment on Signature Recognition
Keywords:signature matching, signature identification, signature similarity, feature ambiguity, feature similarity, handwriting recognition
The complexity in signature recognition problems lies in the fact that a signature usually comprises a small number of handwritten letters that may have limited identifying features and, at the same time, contain natural variations from one signature to the next. Even though it is a frequently encountered problem in forensic sciences, the document examiner’s common method of comparing a questioned signature with a group of control signatures depends upon human perceptual and cognitive processes that are often subjective. To reduce the subjective element in signature comparisons, the authors have experimented with an interactive signature recognition system called the Matching Index (MI) that allows a forensic document examiner (FDE) to utilize his/her expertise to select comparable characteristic features in both questioned and control signatures as well as introduce greater objectivity in the comparison process. An evaluation and assessment of the MI developed by the authors of the questioned signature with the group of control signatures has been implemented by considering numerically assessable features and quantitatively accounting for the information on natural variations manifested in the group of control signatures. Such a comparison provides a more objective expert opinion to present to the court. Successful results of a preliminary test, suggest that the present experiment is potentially promising to provide a more reliable and less subjective approach to signature identification.
How to Cite
©2018 Journal of Forensic Document Examination (JFDE). All rights reserved. Written permission must be obtained from the editor of the JFDE at email@example.com before copying, transmitting, storing, printing, or using for any other means. Authors may copy their article for educational and research purposes exclusively, retain the research data, and receive proper attribution and credit for their work.