3D Face Reconstruction Based On Random Fern

3D face reconstruction has been extensively used in a wide range of applications such as face recognition,film and television production,game design,biomedical technology and so on.With the expeditious development of computer technology,the problem of 3d face reconstruction has gradually become a hotspot in the fields of computer graphics,computer vision and pattern recognition.With the easy access to data,reconstruct 3D face from a single image has attracted much attention from researchers around the world.In this paper,we focus on the reconstruction of 3d face from a color image and a depth image.Considering explicitly regressing the face parameters with random ferns,we propose an algorithm that can directly reconstruct a 3D face mesh from a color image and a depth image.The experiment showed that our method can achieve good results.The main work of this paper is showed as following:1.Implement the bilinear face model based on Face Warehouse,a 3D face database.With our bilinear face model,a face mesh can be parameterized by identity parameter and expression parameter.2.Propose a 3D facial feature descriptor based on depth image,by extracting binary features from depth image which are the pixel difference between two projected points on face mesh.3.Propose a method for 3D face reconstruction from color image and depth image based on the 3D facial feature descriptor described above.The method apply semi-naive bayesian classification model to the classification of facial feature and to the regression of face parameters,and align face mesh by fitting sparse landmarks.4.Experiments exemplify that this method can achieve good reconstruction results.