Multimodal biometric system based on hand geometry, palmprint and signature

Abstract

The aim of this paper is to exploit the best possible combinations of hand geometry, palmprint and offline signatures for multimodal biometric systems by integrating the information at score level fusion. Initially, Zernike moments are extracted for each biometric trait of a person and study the identification accuracy. Subsequently, the effect of identification accuracy using score level fusion of multiple traits of a person is studied. Experiments are conducted on GPDS hand geometry database, PolyU two dimensional palmprint database and UOM offline signature database to assess the actual advantage of the fusion of multiple biometric traits performed at score level fusion, in comparison to the unimodal biometric system. The proposed methodology has shown promising results.