A Empirical Framework for Facial Expression Analysis

For facial expression analysis different-different approaches have been implemented. First thing preprocessing step is done and after that face detection algorithm applied for face detection. After face detection facial feature tracking or feature is extracted on the behalf of obtained feature value. Differentdifferent classifier likes ANN, HMM, SVM is used to train the feature value obtained and then after testing is done to classify the classes to whom this feature value belongs. If high dimension feature value is obtained than PCA does work to reduce the dimension. Our approach is that apply Gabor filter on face to extract the feature. After feature extraction SVM is used to training and testing the feature value and finally similarity measure is evaluated to classify the classes to whom it belongs means it be from happy, sad, disgust, surprise, angry class. As many algorithm has been used for showing the Facial Expression of a Human. They have used different-different feature extraction method. Consideration of some similarity function has taken. Sometimes PCA, LDA was considered for dimensions reduction and for better accuracy.