The3D Human Head Generation Technology Research Based On A Photograph

In virtual reality,3D games, animation, medical research and other fields, users demand more and more to the reality of characters. At the same time, as the continuous improvement of the hardware equipment,3D face reconstruction research has become the hotspot in recent years. Realistic3D face reconstruction based on the picture is not only the core technology in stereo vision, but also directly affect the user realistic experience. In this paper, we have the purpose of applying the3D reconstruction result to the NCUT virtual campus project, which make user roaming in the scene with real characters and enhance the immersion feeling and sense of reality.This paper first introduces the research background and significance of this topic, analyses the present research situation both at home and abroad, and proposes the research idea of deforming CANDIDE-3face mode based on a single positive picture. Deform neutral face model using radial basis function and displacement propagation algorithm, use cylindrical projection for specific face texture mapping. The main work and study are as follows:1) Realize the face detection and feature extraction method, calibrate the face region, and take overall adjustment to the neutral face model, which make the adjustment of model automatic.2) Based on CANDIDE-3neutral face model, this paper proposes the3D feature points calibration and extraction method. With moderate data quantity, it’s convenient for the model deformation and calculation.3) Realize the general face deformation algorithm based on radial basis function. First of all, manual interactively move feature points, record each feature point displacement, and then compute non-feature point displacement with the radial basis function approximation, which deform neutral face model to specific face.4) Propose the displacement propagation algorithm for moving non-feature points. First of all, calculate the non-feature point adjacent to each feature point, when manual adjust feature points, the adjacent non-feature points will.do corresponding movement in accordance with their respective weight, and at the same time ensure the model mesh elastic and not distortion.5) Adopt cylindrical projection method to the deformed specific face model, take texture mapping to the face model, and then get a realistic3D face model.This paper gives the3D face reconstruction system based on a picture, it’s tested that the proposed feature point extraction, the radial basis function and displacement propagation algorithm and the cylindrical projection method can rapidly reconstruct3D face model. What’s more, we achieved good fidelity in the case of less strips and vertices.