A non-standard 3D face mesh texture method based on template face approximation

Abstract

The invention discloses an irregular three-dimensional face mesh texture method based on template face approximation, comprising the following steps: establishing a texture mapping relationship between template three-dimensional face meshes and template texture image points; The grid is preliminarily aligned with the irregular 3D face grid; the error energy function between the aligned template 3D face grid and the irregular 3D face grid is constructed, and the error energy function is optimized to make the template three-dimensional The face mesh approximates the irregular 3D face mesh, and finally obtains the canonical 3D face mesh; according to the above texture mapping relationship, the texture mapping between the generated canonical 3D face mesh and the template texture image is realized, so as to generate a canonical 3D face mesh. Textured 3D face model. The standardized three-dimensional face grid of the present invention realizes texture mapping with the template texture image through template parameterized texture coordinates, and has the advantages of standardizing the input three-dimensional face grid topology and vertex number and low cost.

Description

technical field [0001] The invention relates to the technical field of face data computer processing, in particular to an irregular three-dimensional face mesh texture method based on template face approximation. Background technique [0002] Textured 3D face models are widely used in personalized consumption, film and television advertising, game entertainment, visual real-time communication, cosmetic surgery rehearsal, next-generation human-computer interaction and many other fields. In addition, texture mapping is the process of establishing the correspondence between the three-dimensional object surface and the two-dimensional image space pixel coordinates. Because it consumes less hardware computing resources and is more convenient to render, texture mapping is an effective method to enhance the realism of 3D face meshes. However, in many applications, the 3D face mesh constructed by some methods generally has the problem of irregular topology or a large number of vert...

AI Technical Summary

Problems solved by technology

The method of Newcombe et al. (Newcombe R A, Izadi S, Hilliges O, et al. KinectFusion: Real-time dense surface mapping and tracking[C] / / Mixed and augmented reality (ISMAR), 2011 10th IEEE international symposium on. IEEE, 2011: 127-136.) The depth camera Kinect can be used to perform 3D scanning and reconstruction of the real human body, and finally obtain the 3D human body with texture information, but the model mesh topology reconstructed by this method is relatively complex and the number of vertices is also large, and it takes a long time It takes time for manual repairs to be applied to engineering projects; and methods similar to using 3D scanners have this problem

Images