Edge based mean LBP for valence facial expression detection

Recognition of spontaneous emotion would significantly influence human-computer interaction and emotion-related studies in many related fields. This paper endeavors to explore a holistic method for detecting emotional facial expressions by examining local features. In recent years examining local features has gained traction for nuanced expression detection. The local binary pattern is one such technique. Using the mu LBP, this modified mean LBP adds a discriminating factor to the examined feature via the addition of an edge detector. Hence, the Sobel based μ local binary pattern for the extraction of features in the human face. Using this method, the extracted feature is classified into its valence classes (positive and negative) using an SVM classifier.