3D facial animation generation technology based on 2D image

Abstract

In order to improve the modeling efficiency and ensure the authenticity of the model, this paper proposes a 3D facial animation generation method based on front and side images. First, the feature point information reflecting the shape characteristics of the face is extracted from the front and side photos of a given face, and then these feature point information is used to modify the general face model to obtain a specific face model. Then, the face texture information in the front and side photos is used to texture map the specific face model, and finally a realistic virtual 3D face animation is obtained. In the aspect of feature extraction, this paper uses an improved color Snake model to accurately locate contour lines. The experimental results show that the system still has good processing performance, and the average processing time is less than 0.1s. On the set V, the light source angle is greater than 78 °, the illumination is seriously uneven, and the recognition rate is very low, which indicates that the texture features are not well extracted. But even in this case, the recognition rate after light compensation has doubled, which proves that the two-dimensional image can realize the three-dimensional facial animation generation technology.