Dimensionality Reduction of Facial Features to Recognize Emotion State

Abstract

Lots of applications are based on facial emotion recognition such as action recognition in computer games, medical, and human–computer interaction. In past, number of approaches is proposed to recognize human emotions based on facial expressions by extracting facial features. As the number of facial features increases, the complexity level for recognizing expressions becomes high. Relatively, limited work is done on dimensionality reduction, which restricts the accuracy and robustness of the emotion. In this paper, work is done on dimensionality reduction of facial features and use of minimum number of facial features to represent emotion states.