Automatic human face modeling in model-based facial image coding

The paper describes a method for automatically modelling the human face. The model provides a realistic 3D structure and texture description of a specific face for model-based facial image coding. The modelling process fits a 3D general wire frame facial model (WFM) to a 2D image using important features extracted from the image, and finally imposes facial colour by a texture mapping process. Our method is novel in two aspects. The facial-feature-extracting algorithm has been developed to automate the process of fitting the general WFM to the specific real facial image. This is done by locating the boundaries of the face, mouth, and eyes, using active contour models (snakes) guided by artificial neural networks. Secondly, by introducing face orientation detection during the fitting process, the facial model construction method is robust with respect to its ability to adjust the specific facial 3D model for arbitrary orientations.