Creating Curvature Adapted Subdivision Control Meshes from Scan Data

Often, designers have real-life models which need to be converted to a mathematical representation for further processing. For the designer to be able to manipulate the data sensibly and in a controlled manner the number of data points have to be reduced. However, if the new reduced representation of the shape is sparse everywhere, high frequency detail in the model will be lost. In this work we modify an existing quad meshing algorithm to convert a dense triangle mesh capturing the shape of the real-life model to a quad-dominant mesh of varying density. Our distribution of vertices allows to represent high frequency features in the surface, without increasing the density of the mesh elsewhere unnecessarily. Our quad mesh approximates the scan data up to a predefined error margin. This quad mesh is then transformed into a subdivision control mesh, which corresponds to a limit subdivision surface which closely resembles the scan data.