Facial Behaviour Realization using Statistical Features

Abstract

In this article, the authors attempt to characterize the facial emotions utilizing a few statistically measurable elements. The objective is to demarcate the emotions based on their arousal level. Several low and high arousal emotions such as anger, surprise, sadness, happiness, fear, and disgust are investigated to segment them based on the level of arousal. Initially, the facial images are loaded and the sector of interest is extracted to assess the factual component of the face. The versatile Gabor filter is applied to each of the facial images to extract the discriminate feature vectors. Finally, several statistical parameters are computed from the Gabor feature vectors of each facial emotional expression to characterization and identification based on the level of arousal. To exhibit the stated acknowledgment strategy, JAFFE facial information base and the MATLAB 18 (b) platform are incorporated. Simulation results reveal, it is possible to demarcate the high arousal emotional states from the low arousal states graphically for the sake of identification.