A Fast 3-D Face Reconstruction Method
Changming Meng, Fan Zhou, Ruomei Wang · 2014
Plausible human face reconstruction is an active topic of virtual human research and has amounts of achievements. However, most of them require abundant image source and complex manual operations. In this paper, we propose a method for reconstructing 3D face models using only one frontal face image. Six neutral head models of male and female in three human races are used as the prototypes and one of them is selected to be deformed into the particular face every time. Feature points are extracted from frontal image automatically. Then we link the feature points extracted from the 2D image to the points on the 3D model, and deform the model according to the relative position of the feature points using the radial basis function (RBF) algorithm. In order to make the deformed model more realistic, we use face texture restoration and texture mapping technology. The experiments show that this method can quickly and automatically generate a personalized and realistic face model.