Facial expression recognition using VFC and snakes

The most effective and natural means for human beings is Facial expression that have the dexterity to communicate emotion and regulate inter-personal behaviour. We proposed a novel facial-expression analysis system design that focused on automatically recognize facial expressions and reducing the doubt and confusion between facial-expression classes. The information used in facial expression concentrates mostly on important parts of faces, which gives information of facial regions like mouth, eye and eyebrow. These regions then segmented from the facial expression images. For this, a new Extraction method is introduce to segment efficiently facial feature contours or outline using Vector Field Convolution (VFC) technique. Depending on the detected contours or outlines, extracting facial feature points, this helps in facial-expression deformations. To detect facial features we applied Log Gabor filters and for classification, SVM is applied. Among the detected points, a set of distances classify in model to define prediction rules through data mining technique. These prediction rules are able to classify facial expressions.