Face identification by fusion scores 2D neutral and 3D expressions

The objective of this work is the implementation of algorithms to merge the scores obtained from two images of face with different modalities: one 2D and the other 3D expressions. The faces systems recognition present failures in the identification of faces in presence of expressions. In this work we contribute by an approach based on the fusion of scores by combination. In our face recognition system we use Gabor filter and a transformation EFM discriminant with similarity measure. By merging the information 3D expression and information 2D, we get a rate of recognition TR = 92.25 per cent value to take into account especially in this area, which is the identification of the persons with expressions and especially on a database the CASIA 3D or the likeness between the persons is very pronounced.