ABSTRACT
Typical 3D capture setups, such as structure from motion, Chapter 7, multiview stereo, Chapter 8, and RGBD sensors, Chapter 9, provide a dense sampling of the measured 3D object or scene. In most cases, the measured data can take the form of a dense point cloud sampling the geometry of the scene and carrying surface attributes such as diffuse color and more advanced appearance parameters. However, the majority of visual computing applications require a continuous surface representation of their models which either interpolates or approximates the point data. The standard format of this representation is an indexed triangle mesh which consists of a list of vertices and a list of triangles indexed over the vertices.
