Recognition of facial expression using action unit classification technique

Facial expression recognition is the demanding task in computer vision. It helps the human beings to deliver their emotions to others. Till this time recognition rate are not up to the level of expectation. To improve the recognition rate of facial expression, the dynamic bayesian network method has been chosen to represent facial evolvement in relation to different facial level activity. Experimental results are shown to illustrate the feasibility and effectiveness of dynamic bayesian network method. In this paper the Gabor wavelet and SUSAN operator (Smallest Univalue segment assimilating nucleus) has been adopted which will extracts various features from the faces that result in improved accuracy. In order to recognize our facial expression Adaboost classifier is adopted.