Improved Given 3D Face Reconstruction Based On Orthogonal Photographs

Generating realistic 3D human face models has been a persistent challenge in computer graphics.The applications of 3D human face models include such diverse fields as character animation for films and advertising,computer games,virtual reality, teleconferencing,user-interface,surgical facial planning, and criminal discrimination.This paper focuses on the field of facial reconstruction and proposes an improved method that reconstructs textured 3D face model based on orthogonal photos with a few user interactions.The research work mainly includes the following aspects:1) We study the research meaning and the present conditions of domestic and inter- national research of the face recognition.we also study the technologies of 3D reconstru- ction and analyse the difficulties to the existing techniques of 3D reconstruction.2) According to the FDP in the standard of MPEG-4,we define some feature points in the human face based on the characteristics of human face.3) We also research the method of deriving the 3D coordinate of the feature point from the 2D orthogonal images.With this feature points,we get a given face model by deforming the generic face model.Using the nearest principle of Euclidian distance, we deal the non-feature points with feature points.4) We use multiresolution technique to build a Gaussian pyramid and a Laplacian pyramid for the frontal and profile facial images,and merge the boundary of the original texture image.The process of building a pyramid is to produce image sequence with different scales.5) We use cylinder texture mapping and OpenGL to generate a realistic 3D face model.6) Finally,we make experiments on the face database which we get with PC camera. And the results show the method can make a good reconstruction.