Research On 3D Reconstruction And Facial Animation Based On Facial Features

The three-dimensional reconstruction technique of the face and the animation technique of the face are the important researches in the field of computer graphics.The traditional face 3D reconstruction technology relies mainly on expensive 3D scanning equipments and a large amount of time-consuming and post-manual processing,and requires the scanned subject to maintain a fixed posture for a considerable period of time.In recent years,the development of large-scale computing power of computer-related hardware,especially the development of GPUs and distributed computing,has made it possible to come up with a real-time and efficient solution.At present,the methods adopted at home and abroad can be roughly divided into the following types: traditional 3D face reconstruction method,3D reconstruction based on2 D image,3D reconstruction based on depth camera,and 3D face reconstruction based on end-to-end.In recent years,academic and industrial have made great breakthroughs in the fields of detection and recognition.The research on human body,especially human face,has become a research topic of great concern,and are also commonly used in the real world.The human face is the most recognizable part of the individual,but the features of the human face in the two-dimensional space are generally constrained by the uncertain factors such as the illumination in the real world,the expressions of the face,the angles of the face,and the occlusion of the face.The research on face in three-dimensional space has become the next breakthrough direction.Based on this trend,this paper studies the face in three-dimensional space.Complex 3D spatial structure features and rich subtle features make face 3D reconstruction and related facial animation technology the major problems in computer graphics.In this paper,the mapping relationship between the key points of the two-dimensional image of the face and the key points of the three-dimensional model of the face is studied,and the variation of the three-dimensional space is calculated according to the two-dimensional key points,thereby gradually reconstructing and optimizing the three-dimensional model of the face.Another exploration of this paper is facial animation technology,which is mainly divided into: a method of customizing bone point driving and a method of calculating individual coefficients and expressioncoefficients by blendshape.Based on the above methods and algorithms,this paper studies and develops corresponding directly operable application to demonstrate the research result.The system renconstructs the face model and adds custom textures,even to the eyeballs,different skin tones,different hairstyles,and different lighting.At the same time,based on the reconstructed face model,combined with the blendshape technique,we can set a more subtle gesture expression for the face model,build a database of the same subject,and enrich the depth of the data under the premise of the original data breadth.The database can provides convenient and rich practical data for the face-related film and television,the development of game production,virtual reality,medical beauty and so on.Combining related computer graphics technology,in order to enrich the research of face in 2D space and 3D space,the research content of this paper mainly includes the following parts: 3D face reconstruction based on 2D image,improvement of blendshape technique,face animation exploration.