A mechanism for filtering distractors for graphical passwords

Poet, R. and Renaud, K. (2007) A mechanism for filtering distractors for graphical passwords. In: International Graphonomics Society (IGS2007), Melbourne, Australia, 11-14 Nov 2007,




Graphical passwords hold some promise in a world of increasing numbers of passwords, which computer users are straining to remember. However, since these kinds of mechanisms are a fairly new innovation many design and implementation problems exist which could wellprevent these mechanisms from realising their full potential. This paper addresses the problem of choosing distractors for graphical passwords. It is important that the distractors should not be too similar tothe target image, since that could cause confusion and seriously reduce the potential for enhanced memorability and usability. We cannot use the obvious approach of screening out confusing or similar images by means of classification because images are difficult to categorise unambiguously and also because such classification is time-intensive and therefore does not scale. The mechanism of interest for thispaper makes use of user-produced doodles, which makes distractor choice by means of categorisation even more difficult. In this paper we present an algorithm that automatically recognises pertinent features in target doodles and then identifies similar doodles in order to disqualify these doodles as distractors for those targets within a particular challenge set. This algorithm produces few false negatives, which makes it particularly suitable for use in this domain.

Item Type:Conference Proceedings
Glasgow Author(s) Enlighten ID:Renaud, Professor Karen and Poet, Dr Ron
Authors: Poet, R., and Renaud, K.
Subjects:Q Science > QA Mathematics > QA75 Electronic computers. Computer science
College/School:College of Science and Engineering > School of Computing Science
Copyright Holders:Copyright © 2007 The Authors
Publisher Policy:Reproduced with the permission of the authors

University Staff: Request a correction | Enlighten Editors: Update this record