Research And Application Of Fast 3D Face Reconstruction Based On Front Facial Image

Three-dimensional face reconstruction has an important significance in face recognition,virtual reality,health care,education and so on.And it is an attractive issue in computer vision and graphics.Researchers have proposed various techniques for different applications in the last decades.However,most general users care more about the resemblance in vision and simple interaction.In this paper,we present a fast 3D face reconstruction approach.The approach only needs a suitable frontal image captured by a web camera,from which the 2D facial features are extracted.The predefined feature points of the 3D face model are mapped to 2D space by orthographic projection and matched with the feature points obtained from the input image.According to the contour feature points,the most similar face model is selected as a generic face model.Then the generic face model and image are partitioned into patches in the same way.And each patch of the model is deformed respectively,according to the corresponding features in the image.Finally,the facial texture is calculated and mapped to the face model by each patch.Firstly,the background and significance of the research and the theory of 3D face reconstruction methods are introduced.Considering to the shortcomings of the existing methods,a general 3D face model is designed.We describe the construction of 3D face model,patches and the model files in detail.Every step of fast 3D face reconstruction is described in detail,including face feature extraction,face feature matching,embedded deformation and texture mapping.Finally,the application of 3D face fast reconstruction technology is analyzed and technology is applied to fast 3D face reconstruction system,personalized virtual fitting and 3D printing.The results show that the proposed method could generate realistic 3D face.The method is robust and easy to implement.It could be applied to many virtual reality applications directly.