Combining Snakes and Neural Networks for Off-Line Signature Verification

José F. Vélez 1 Ángel Sánchez 1 Ana B. Moreno 1 José L. Esteban 1
1 GAVAB - Grupo de Algorítmica para la Visión Artificial y la Biometría
Abstract : This paper introduces an improved snake algorithm based on the work by Kass et al. Our approach is applied to the off-line signature verification problem where signatures are scanned and then converted into binary images. This way no dynamic information of the signers is available. We also have considered some real conditions for the verification problem when applied to bank check. For example our system uses only one training signature per subject. Involved system parameters are tuned to solve the task in an effective and efficient manner. A two-layer perceptron is build for signature classification and it uses only two signature features (distance and matching factor) provided by the adjusted snake. Finally, a study of the system for a signature database is provided.
Keywords : neural networks. Off-line signatures verification snakes neural networks
Type de document :
Communication dans un congrès
Guy Lorette. Tenth International Workshop on Frontiers in Handwriting Recognition, Oct 2006, La Baule (France), Suvisoft, 2006
Liste complète des métadonnées

Littérature citée [5 références]  Voir  Masquer  Télécharger
https://hal.inria.fr/inria-00103939
Contributeur : Anne Jaigu <>
Soumis le : jeudi 5 octobre 2006 - 15:16:55
Dernière modification le : jeudi 5 octobre 2006 - 15:47:48
Document(s) archivé(s) le : mardi 6 avril 2010 - 18:31:37

Fichier

cr1026231491271.pdf

Identifiants

  • HAL Id : inria-00103939, version 1

Collections

IWFHR10

Citation

José F. Vélez, Ángel Sánchez, Ana B. Moreno, José L. Esteban. Combining Snakes and Neural Networks for Off-Line Signature Verification. Guy Lorette. Tenth International Workshop on Frontiers in Handwriting Recognition, Oct 2006, La Baule (France), Suvisoft, 2006. 〈inria-00103939〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

83

Téléchargements de fichiers

142

Contact


archive-ouverte@inria.fr