Score fusion of face and voice using Dempster-Shafer theory for person authentication

Abstract

The present paper describes a method for person authentication which is based on the fusion of face and voice at the score level using Dempster-Shafer. Our experiments on the publicly available scores of the XM2VTS Benchmark database show a consistent improvement in performance compared to each individual modality. We have compared the proposed fusion with the sum rule and the log-likelihood ratio based fusion, the results have shown that our method enhances the performance compared to the methods cited above.