HOG, Wavelet Moments and Cartoon Features based Facial Expression Recognition using Random Forest Classifier

Abstract

Facial expressions play an important role in nonverbal communication as they provide information about people's feelings and are used in a wide variety of fields, especially in education, health, law and entertainment. Recognizing emotions is one of the most complex areas of science. In recent years, more and more applications have tried to automate it. These innovative applications concern several areas such as helping children with autism, video games, human-machine interaction. This paper presents a novel technique of facial expression recognition using HOG, Wavelet Moments and Cartoon descriptors based feature extraction methods. Random Forest Classifier is utilized to calculate the similarity score using two public datasets of facial expressions; KDEF and JAFFE. The performance evaluation is done on the basis of a confusion matrix plot with accuracy, precision and sensitivity graphs.