Physiological Feature-preserved Facial Modelling
Ming‐Feng Wu · 2006
In this paper we present an approach for physiological feature preserved facial modelling.We employ a multi-layered strategy,where user directly handled control points(called primary control points) are taken as an input layer and the other control points(called secondary control points) are taken as an output layer.An artificial neural network is built upon these layers.A set of face models is used to train the network using Error BackPropagation approach.The trained constraints are then transferred to the output layer to guide the interpolation of editing information over the facial model.The result can apply to any face models with the same topology.Experimental results show that the proposed multi-layered approach not only provides efficient and exact editing operation,but also preserves the fidelity of the physiological feature of facial models.