Multibiometric score-level fusion through optimization and training

This paper presents a method to do a fast score-level fusion between multibiometric data. The method uses optimization and training to generate the total score of multibiometric system. This score will depend on as many constants as the multibiometric type of data. These constants will be estimated solving an optimization problem built in such a way that the resulting linear combination between these constants and the scores makes it possible to discriminate between genuine and impostor.