Emotion modeling

Abstract

Human emotion is a very important part of human to human communication. To be able to incorporate emotions into a Human Computer Interaction (HCI) system would greatly improve the quality of the HCI.Much work has been done in the Emotion recognition field. Image filters, feature extraction, feature tracking, movement detection and much more has been studied for use in Facial Expression Recognition. Besides facial expressions, emotion recognition in speech and EEG signals has also been studied extensively. Different techniques for Facial Expression Recognition are explored in this project, including MLP-based and LDA-based classifiers. A Cascading approach to classification is also explored where classifying “easier” emotions first can help to simplify the problem of classifying the “harder” emotions later. The results from these classifiers are comparable with previous works and are viable alternatives to emotion classification.