Automated Recognition Of Emotions From Facial Expressions Using Back Propagation Neural Networks

In everyday human interaction, facial expressions playa key role. Most of the automated emotion recognition systems recognize the basic set of six emotions anger, disgust, sad, happy, fear and surprise. However these basic emotions occur rarely rather communication happens through combination of one or more emotions. To promote human-machine interaction we propose an automated emotion recognition system. The proposed system is an intelligent system which uses artificial neural networks and digital image processing for emotion recognition. JAFFE database is used with 10 subjects bearing primary six emotions and a neutral expression. Neural network in the automated system is trained and tested against the database.