A fusion methodology based on Dempster-Shafer evidence theory for two biometric applications

Different features carry more or less rich and varied pieces of information to characterize a pattern. The fusion of these different sources of information can provide an opportunity to develop more efficient biometric system compared when using a feature vector. Thus a new automatic fusion methodology using different sources of information (different feature sets) is presented here. Dempster-Shafer evidence theory is employed for this purpose. For performance evaluation significantly large data sets of the biometric sources signature and hand shape are used. The results on combining different feature vectors compared to a single vector with our approach prove the importance of a fusion process