Czyz, J.
Sadeghi, M
Kittler, J
Vandendorpe, Luc
[UCL]
In this paper we study two aspects of decision fusion for enhancing face authentication. First, sequential fusion of scores obtained on successive video frames of a user's face is used to reduce the error rate. Secondly, the opinions of several face authentication algorithms are combined so that the combined decision is more accurate than the best algorithm alone. The experiments performed on a realistic database demonstrate that the fully automatic multi-frame - multi-experts system proposed in this work allows a significant improvement over the static single-expert system.
Bibliographic reference |
Czyz, J. ; Sadeghi, M ; Kittler, J ; Vandendorpe, Luc. Decision fusion for face authentication.1st International Conference on Biometric Authentication (Hong Kong(Peoples R China), Jul 15-17, 2004). In: Lecture Notes in Computer Science, Vol. 3072, p. 686-693 (2004)In: First International Conference on Biometric Authentication, Springer-verlag Berlin : Berlin2004, p.686-693 |
Permanent URL |
http://hdl.handle.net/2078.1/61246 |