| Nowadays, with continuous development of computer technology and advances in multimedia technology, 3D face modeling technology has become one of the hottest topics in the field of computer graphics in recent years. Human face, as a special physiological organ, has a richer expression and more action experience than other human organs. Because of the special nature and morphological diversity of the face, 3D face modeling technology research faces many difficulties, especially reduction the 3D face modeling with a single photo which missing the depth information.However, in some cases, detailed facial information acquisition is not only expensive, but also d ifficult. The acquisition of single photo is relatively easier, which makes needs of the use a single photo for fast 3D face modeling more urgent.Under such a background, this thesis chooses a single face photo as input sample to reconstruct 3D face model, thereby achieving the reduction from two-dimensional images into three-dimensional models. Firstly, this thesis reviews current 3D face modeling technology at home and abroad, and summarizes the current mainstream 3D face modeling method. Secondly it describes the radial basis function and the related algorithm based on deformation model theory, especially for thin samples of different dimensions of the basic function of algorithm analysis, thin plate spline and use different dimensions of face image dense correspondence; Thirdly using principal component analysis method to establish the generic face model, which extends the construction of a specific 3D face model algorithm. During the reconstruction of three-dimensional human face of this thesis, the select of feature point is particularly important. In this thesis, the selected feature points proposed an improved algorithm based on ASM, which is more accurate than the traditional ASM algorithm positioning feature point. Finally, in the earlier theoretical studies and analysis of the algorithm based on, based on the theory of deformation model, three-dimensional face model Candide as a fundamental, through two-dimensional single photo and a generic 3D face model gradually match, with OpenCV and OpenGL procedures to complete the verification, to achieve the specific faces of the 3D face model reconstruction of specific experiment, and the final reconstruction results show. The final results show that: in this thesis, based on a single picture of the 3D face modeling technology, in the front, side and pitching surfaces are able to achieve better modeling results.This study is very important to promote the three-dimensional face model, which has positive significance in television advertising, medical beauty, game animation, video, certification and many other practical applications; while research tri-mode is also conductive to the good application in the future simulation of human medicine, virtual reality and so on. |