Model Merging Through Mean Value Coordinates

Abstract

Merging a model to another model, including the mesh and its texture, is a useful technology for generating new models. This paper introduces a novel method to merge models through Mean Value Coordinates (MVC). The method firstly computes the MVC of the vertices of a source mesh with respect to its merging boundary. As the target mesh is specified, the differences between the two meshes’ boundaries are computed. The source mesh is transformed fluently according to the differences through MVC, and then construct a new mesh with the target. With the same framework, smoothing the seaming line can be achieved by interpolating tangent vector variations. And a new texture is similarly constituted by interpolating the color differences on the boundaries to make a consistent model. This method avoids the mesh parameterizations or the solving a large linear system, but merges through computing closeform expressions. Experiments show that model merging with this method is less artifacts and faster than some other methods.