Classification of Facial Expression Using SVM for Emotion Care Service System

This paper presents a real-time approach to classify facial expression from a sequence of input images to provide emotion care service in developing a wellbeing life care system. The facial expression recognition from video images is useful to handle with sequential changes of facial expression. However, it needs more cost in training images and constructing database rather than using a still image. In this paper, we present automatic technique which infers emotions by recognizing facial expression from input video in real time. To classify the facial expression the feature displacements traced by the optical flow are used for input parameters to a support vector machine (SVM). The classification result of facial expression from input video will be used for providing personal emotion-care service depending on the emotional state.