Fusion of Multibiometrics Based on a New Robust Linear Programming

Multibiometrics provides a reliable method for identity authentication and has the potential to be widely applied. The success of a multibiometrics method depends critically on its ability to fuse complementary information supplied by different modalities, where the most challenging problem is to evaluate the importance of different modalities. In addition, identity authentication at a distance has become a development trend of multibiometrics. In this paper, we propose a new robust linear programming method to fuse multibiometrics by combining the modalities optimally. The proposed method can provide a reasonable trade off between conservatism and robustness. Experimental results on CASIA-Iris-Distance, a public and challenging multibiometric database, demonstrate the effectiveness and robustness of this method.