Facial Expression Synthesis for a Desired Degree of Emotion Using Fuzzy Abduction

Abstract

Modulating facial expression of a subject to exhibit emotional content is an interesting subject of current research in Human-Computer Interactions. The ultimate aim of this research is to exhibit the emotion-changes of the computer on the monitor as a reaction to subjective input. If recognizing emotion from a given facial expression (of a subject) is referred to as forward (deductive) reasoning, the present problem may be considered as abduction. The logic of fuzzy sets has widely been used in the literature to reason under uncertainty. The present problem of abduction includes different sources of uncertainty, including inexact appearance in facial expression to describe a given degree of a specific emotion, noisy ambience and lack of specificity in features. The logic of fuzzy sets, which has proved itself successful to handle uncertainty in abduction, thus can be directly employed to handle the present problem. Experiments undertaken reveal that the proposed approach is capable of producing emotion-carrying facial expressions of desired degrees. The visual examination by subjective experts confirms that the produced emotional expressions lie within given degrees of emotion-carrying expressions.