Facial Expression Recognition Based on Basic Expressions and Intensities Using K-Means Clustering

Facial expression recognition provides rich information for human-robot interaction with respect to emotions. This paper presents a facial expression recognition method that recognizes facial expressions as well as intensity and mixture ratio of basic facial expressions. In this system, K-means clustering method is applied on Cohn-Kanade image database. Algorithms are adopted to align the input facial images to get texture features. A novel method is proposed to recognize mixture ratio of very basic facial expressions which comprises of 6 facial expressions viz. happy, angry, fear, neutral, sad and surprise and also the intensity of the expression. Back propagation algorithm is used to obtain the recognition scores, which are then used again to classify the facial expression results. Experimental results verified that the proposed method can effectively recognize the image and expression intensity and provides an accuracy of 98 %.