Automated Facial Expression Recognition framework for analyzing Dynamic Expression in Occluded Images

Abstract

The automatic emotion detection method plays important role in many fields such as human computer interaction, health informatics (Depression, autism), and education (eg. Tutoring system) and driving system. The salient facial patches can be used to extract the discriminate features of the fascia and is used to identify the universal expression such as neutral, anger, disgust, fear, joy, sadness and surprise. The proposed system provides a novel framework to identify the Universal expressions along with the other subtle expressions by using the facial point tracking method. Automated facial expression recognition (AFER) algorithm is proposed to identify the dynamics of facial expression in the temporal domain. In addition, the facial expressions of the occluded images can be determined by the use of iterative face recovery and recognition by input approximation, which achieves the normalized facial appearance and thereby the expression in the face.