3D facial analysis for identification and man-machine communication

An accurate detection and analysis of the human face can provide valuable information and indications about the identity of a person, their intentions, reactions or emotions. These are the basis for many new applications, such as personal identification in security areas, medical diagnostics and therapy (autism, facial paralysis), capturing the car interior or new intuitive man-machine interfaces or dialogue systems. All of these applications require a highly accurate analysis and representation of the 3D shape and movement of the human face. In recent years, significant advances have been made in the field of 3D imaging. However, current methods for detailed and accurate recording require very complex structures and calculation methods or a person-specific a priori model, which is why these methods cannot be used for the above-mentioned applications so far.

In this research project, new methods for highly accurate, passive, dynamic 3D facial imaging are developed from simple stereodata without prior knowledge of the captured person. Due to the simple design and the non-specific approach, the processes should have potential for a broad application in a wide range of applications.