Abstract:
This research presents methods of identifying (Facial Expression
Recognition). The objective of this recearch is to present a combact texture
oriented method , along with the dimensions reductions,so it would be used in
the training of three neural networks: ( Single Layer Neural Networks (SLN),
Back Propagation Algorithm (BPA) and Cerebellar Model Articulation
Controller (CMAC) ) for identifying facial expressions. The proposed methods
are called ( intelligent) methods because they can assimilate the variations in
facial emotions and hence proved to be better for untrained facial expressions.
Conventional methods have limitations, so facial expressions should follow
some constraints. Gabor wavelet is used in different angles to extract possible
textures of the facial expression, in order to achieve the expression detection
accuracy. Higher dimensions of the extracted texture features are further
reduced into a two-dimensional vector by using Fisher’s linear discriminant
function in order to increase the accuracy of the proposed methods. Tarining and
testing have been done on JAFFE database on certain facial expressions ( angry,
disgust, happy, sad, surprise and fear) .The performance comparisons of the
proposed algorithms are presented. The resultes obtained are acceptable
according to international standards.
