Etablissement
Ecole Nationale Supérieure d'informatique
Affiliation
Département de Post-Graduation
Auteur
BELHADJ, Foudil
Directeur de thèse
AITAOUDIA Samy (Professeur)
Filière
Informatique Industriel
Diplôme
Doctorat
Titre
Biometric system for identification and authentification
Mots clés
Fingerprint identification, minutiae matching,local structures, minutiae insertion deletion.
Résumé
Fingerprint is the most imprtant biometric modality used for human identification. Automated Fingerprint Identification system (AFIS) can be achieved in three principal steps : Segmentation, feature extraction and matching. The reliability of an AFIS depends largely on the technique used in the latter step. The litterature proposes many mathods for the matching step that can be broadly classified into two categories : Image Based and Minutia based. this last one is the most used and is more efficient. Minutiae based matching process can be affected by serious problem especially : the rotaion, translation,distorsion and presence of noise. To deal with this problems, many authors design their matching methods to be achieved in two seps : (1) Local matching step based on local structures extrcated around a minutia. these structures are invariant under rptation, translation and scaling. tha goal of this step is to get relibale minutiae used for global alignment.(2) global matching which is a consolidation step to extract a final matching score. these mathods have generally a heigh time complexity and perform bad against the presence of noised (added/inserted) minutiae. In this work we intend to propose an efficient minutiae based fingerprint matching method that is robust to noised minutaie problem and has an n*log(n) time complexity.
Statut
Vérifié