Sample Selection for Optimising Signature Enrolment

Guest, R.M. and Fairhurst, M.C. (2006) Sample Selection for Optimising Signature Enrolment. In: Proc. 10th International Workshop on Frontiers in Handwriting Recognition (IWFRH), 2006 October, La Baule, France. (The full text of this publication is not available from this repository)

The full text of this publication is not available from this repository. (Contact us about this Publication)

Abstract

Accurate and representative template construction is critical for the successful and robust operation of dynamic signature verification systems. This paper presents a comparison of enrolment sample selection strategies using a large database of donated signature samples. Enrolment strategies evaluated include several that are based on sample variance within standard signature performance features, together with a series of established reference methods used for purposes of “baseline” comparison. Sample selection enrolment strategies are tested on two representative commercial signature verification systems by assessing the rate of successful enrolment and verification. The results show that using a prediction method based on the measurement of variance within samples for selection increases the ability to successfully enrol. The investigation is thus important in helping to develop possible best practice for implementing effective enrolment procedures.

Item Type: Conference or workshop item (Paper)
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800 Electronics (see also: telecommunications) > TK7880 Applications of electronics (inc industrial & domestic) > TK7882.P3 Pattern Recognition
Divisions: Faculties > Science Technology and Medical Studies > School of Engineering and Digital Arts > Image and Information Engineering
Depositing User: Yiqing Liang
Date Deposited: 11 Aug 2009 16:18
Last Modified: 14 Jan 2010 14:37
Resource URI: http://kar.kent.ac.uk/id/eprint/9830 (The current URI for this page, for reference purposes)
  • Depositors only (login required):