Recognition of emotional expressions from images of face

A novel framework is explored for recognition of six emotional expressions (e.g. joy, surprise, sadness, etc.) from a given face image. The framework is based on the assumption that information related to an emotion or expression lies in the subspace orthogonal to the subspace representing expression-neutral faces. Principal Component Analysis (PCA) is used for deriving the principal subspace of the face images showing no expression. Then a method is derived to find the orthogonal complement (OC) of the subspace defined by the PCA mentioned above. It is shown using different tools such as dendrogram and Davies-Bouldin cluster index that the OC of the principal subspace better represents the expressions as compared to the principal subspace using PCA. Extensive experiments are done to validate the recognition capability of the proposed OC space. Two well-known publicly available facial expression databases are used for the experiments. The proposed framework exhibits higher (9.66% on an average) recognition capability as compared to the present state-of-the-art works.