3D Facial Action Units and Expression Recognition using a Crisp Logic

This work proposes a method for recognizing the main 13 Facial Action Units and the 6 basic emotions. The methodologies rely on Differential Geometry to extract relevant discriminant features from the query faces, and on some linear quantities used as measures: Euclidean, geodesic, and angles between 17 automatically extracted soft-tissue landmarks. A thresholding system which evaluates local properties of connected regions, selected through tailored geometrical descriptors, supports the identification of the AUs. Then, a technique based on crisp logic allows the identification of the global expression. The three-dimensional context has been preferred due to its invariance to different lightening/make-up/camouflage conditions.