3D Face Reconstruction Based On Improved Morphable Model

With the rapid development of photography and imaging technology,3D reconstruction technology has been widely used in many fields such as remote sensing,medical care,industry,education,film and television,and entertainment.Specific to the direction of computer vision and graphics,3D face reconstruction has gradually become a core content of 3D modeling and plays an important role in 3D face recognition and computer animation.There are four main methods for 3D face reconstruction,including parametric models,muscle models,visual models,and deformation models.The current deformation model is the most widely used.Relatively speaking,the model has a good sense of reality and a high degree of automation.In the process of establishing the deformation model,accurate and rapid registration between 3D face samples is the most difficult and most critical link.For the face defornation model of 3D data registration and non-automated model matching problem,the work of this paper is mainly reflected in the following two aspects:1.Combine rigid and non-rigid iterative closest point algorithms to achieve coarse to fine 3D face registration.First,the feature points of the 3D target face and the template face are extracted,and according to the corresponding relationship between the feature points,a rigid iterative closest point algorithm is used to obtain the transformation parameters between two 3D face,thereby adjusting the 3D target face to the 3D template face Coordinate system,and scale changes to complete the 3D face between the rough registration.On this basis,the elastic deformation based on Laplace function 1s added to make the target face closer to the template face,and then the non-rigid iterative closest point algorithm is used to improve the registration speed without loss of registration accuracy.2.The active shape model is used to extract facial feature points as initial values to auto,ate model matching.In the initialization phase,the active shape model was introduced to extract the features of the face's contours,eyes,nose,eyebrows,and mouth,and template matching was achieved by adjusting a series of combination parameters such as model,lighting,and camera position.The current model searches for the combined parameters starting from the extracted facial feature points,reducing the search space,improving the matching speed,and completing the model matching and 3D face reconstruction automation.