IDENTIFICATION BIOMÉTRIQUE PAR FUSION MULTIMODALE DE L’EMPREINTE DIGITALE

Abstract

The uni-modal biometric systems suffer from several problems which are due to the origin of the use of a single biometric trait susceptible to noise, poor capture, poverty in terms of confidential biometric points and in particular the deterioration of the quality of the biometric entry. Introduction of biometric systems multi-modal is a solution to these problems. We have proposed a system of authentication / identification automatic of individuals by multiple instances and repeated fingerprint, with a fusion at the score level. We used SURF algorithm for the extraction of characteristics, taking advantage of its robustness thanks to the invariability, scale and the rotation, more of its speed. The multi-modal recognition process makes it possible to strengthen the biometric systems under terms of security, reliability and relevance. Indeed, 100% were the overall recognition rate for the proposed method while the former mentioned system we’re rated as 65%