Classification of Human Facial Expression:: an Application of Image Processing and Machine Learning

This book describes a computer vision system fordetection and classification of 7 basic facialexpressions. Facial expressions are communicated bysubtle changes in one or more discrete features suchas tightening the lips, raising the eyebrows, openingand closing of eyes or certain combination of them,which can be identified through monitoring thechanges in muscles movements located around the aboveregions. In our research, an analytic facerepresentation consists of 15 feature points has beenused that identifies the principle muscle actions andprovides visual observation of the discrete featuresresponsible for each of the 7 basic emotions. Featurepoints from the region of mouth have been detected bysegmenting the lip contour applying a variationalformulation of the level set method. A multi-detectorapproach of facial feature point detection has beenutilized for identifying the points of interest fromthe region of eye, eyebrow and nose. Feature vectorscomposed of 15 features are then obtained from thesefeature points and used to train a SVM classifier sothat the system can classify facial expressions froman unknown face image with a certain level of accuracy.