Human face is the most important content to distinguish and remember individuals from different characters. The rich information contained in it allows people to perform information transmission and identity authentication through face images. Compared with 2D plane faces, the face model in 3D space contains more complex and rich biological information. Therefore, 3D face reconstruction technology has long been one of the hotspots in the field of computer vision. In this paper, we propose a single-image 3D face reconstruction algorithm. In order to improve the accuracy and speed of image generation, we use an improved network framework based on GAN and a joint loss function, and finally we obtain a 3d face model containing texture features through detailed filling.
展开▼
