Detection of Facial Expression using Fisher, Multi-SVM and Pattern Network and Comparison

Expression of face is very remarkable posture underneath the derma of the face. Expression of faces is one of the ways of human communication, which deliver so many things without talking verbally. The main purpose of this project is to develop a system for detecting facial expression of a given image among the seven basic human emotion expressions such as Angry, Sad, Happy, Contempt, Surprise, Disgust and Fear. This is performed using three different methods. The first method used is based on Eigen faces and Fisher face, using this method the obtained accuracy is 95.81%. The second method used here is HOG feature extraction and using these features to train the multi-SVM, and obtain the expression of test image. Using multi-SVM the obtained accuracy is 99.58%. The third method used is pattern neural network for emotion recognition of face image, for this also HOG features are used for training the network, and the accuracy obtained using pattern neural network is 90.79%