Facial Expression Transfer Based On 3D Morphable Model

The research of Expression Transfer is challenging and has tremendous application value. This paper focuses on Facial Expression Transfer technologies which are based on the 3DMM (3D Morphable Model). In this paper, a Expression Transfer framework based on 3DMM is designed and implemented. The main contribution of this paper is summarized as the following two aspects:1) In this paper, we apply the combination of AAM method and 3DMM shape model to construct 3D face model. This method is more efficient than face model matching based on the traditional 3DMM. At the same time, training 2D AAM does not require the expression database with 3D high-precision face, the cost is greatly reduced.2) In the traditional Facial Expression Transfer method based on 3DMM, facial wrinkles and other details of face can't been migrated well. In this paper, this problem has been solved. We apply the existing mature technology, a face detail transfer approach based on a Laplacian Differential Model for Facial Expression Transfer based on 3DMM, which is an effective complement for the 3D face shape due to insufficient accuracy of the details.In summary, the method of Facial Expression Transfer based on 3DMM is optimized and improved in this paper, We designed and implemented a useful 3D facial expression transfer framework. Experimental results prove the proposed method is effective, which can transfer the source face expression to the target face efficiently with natural appearance.