Automatic Synthesis of 3-D Facial Expression

Abstract

We explain the technique to produce the face expressionchange using the basic picture such as the expressionlessface and the smiling expression produced automatically bythe interpolation ability of the Kohonen neural networks(KNN), also known as competition networks. The 3D shape andcolor information of the facial surface which is obtainedby the range finder are used for the KNN's learning. Next,as the method of searching the optimal path of the faceexpression change, the DP matching is used. It uses theweight which was generated by the learning of KNN for thedistance calculation of the DP matching. Finally, weexamine the difference between the face expression taken bythe photograph and the face expression produced by theKNN.