Comparative analysis of decision-level fusion algorithms for 3D face recognition
18th International Conference on Pattern Recognition, ICPR 2006, Hong Kong, Çin, 20 - 24 Ağustos 2006, cilt.3, ss.1018-1021, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Cilt numarası: 3
- Doi Numarası: 10.1109/icpr.2006.387
- Basıldığı Şehir: Hong Kong
- Basıldığı Ülke: Çin
- Sayfa Sayıları: ss.1018-1021
- Boğaziçi Üniversitesi Adresli: Evet
Özet
3D shape-based face recognition algorithms can be improved by using decision-level fusion algorithms. In this work, we present a comparative analysis of various fusion algorithms, and also propose novel ones. The contributions of the paper can be summarized as: i) a through analysis of several decision-level fusion algorithms, ii) a dynamically estimated reliability-assisted fusion schemes, and iii) a novel implementation of LDA-based cascaded serial fusion algorithm. Experiments conducted on the 3D_RMA dataset confirm that serial fusion offers the best solution, and dynamic calculation of reliability estimates improves the accuracy of the standard, fusion schemes.