Facial Expression Recognition Using Distance Signature Feature

Abstract

Distance feature has great significance in recognizing facial expressions. Identifying accurate landmarks is a vital as well as challenging issue in the field of affective computing. Appearance model is used to detect the salient landmarks on human faces. These salient landmarks form a grid on the human face. Distances are determined from the one landmark point to another landmark point in grid and normalized. A novel concept of corresponding stability index is introduced which eventually is found to play important role to recognize the facial expressions. Statistical analysis such as range, moment, skewness, kurtosis, and entropy are calculated from normalized distance signature to supplement the feature set. This enhanced feature set is supplied into a Multilayer Perceptron (MLP) to arrive at different expression categories encompassing anger, sadness, fear, disgust, surprise, and happiness. We experimented our proposed system on Cohn-Kanade (CK+), JAFFE, MMI, and MUG databases to training and testing our experiment and establish its superiority performance over the other existing competitors.