Human Face Reconstruction Technology And Application Based On Local Control Model

The techniques for face modeling and animation have made great progress since 1970s. The development of them can be divided into 3 stages: Parameter module in 1970s (Parke), muscle model in 1980s (facial action coding system, simple muscle module and pseudo muscle model) and the modeling and drawing techniques based on image in recent years. For years the result has been applied in many fields like movie, communication, medical treatment and games. However, due to the complex construction of human face, it is still very meaningful to simulate the human face and its motion style.There exist many methods to construct the human face. In chapter 2 some fundamental techniques of human face construction and animation are introduced.In chapter 3 we introduced some basic methods to generate a personalized facial model. Excellent facial model and facial animation can be made using these techniques. However, as explained in this paper, these methods need the assistance of some expensive equipment or complex and special operation. To solve this problem, we presented a facial region control model in this paper and applied it to face reconstruction from an orthogonal imageIn chapter 3 and chapter 4 we mainly introduce local control model and its application in human face reconstruction from orthogonal images. The facial local control model adopts the 2-level grid facial model. Fixed grids are used in the inner level to represent the facial skeleton. The outer level is divided into several independent regions, each controlled by several controlling points. The changes of the points inside the regions are controlled by the corresponding distortion parameters, which can be computed from the feature points or adjusted manually. The face reconstruction method based on orthogonal images can be divided into several stages like marking the feature points on the frontal image and the lateral one, computing the depth values of the feature points, computing the distortion parameters automatically, adjusting the distortion parameters manually, smoothing the meshes, creating the textures and Compared to other human face reconstruction methods using orthogonal images, our method can be more precise by using facial local control model. As a result, the reconstruction result is more vivid.Finally, in chapter 5 we provided some comparison results and applied our method to practical project. In "" we adopted the methods introduced in chapter 3 and chapter 4. From the results we can see that the effect is satisfying.