Regular Grids: An Irregular Approach to the 3D Modelling Pipeline

The 3D modelling pipeline covers the process by which a physical object is scanned to create a set of points that lay on its surface. These data are then cleaned to remove outliers or noise, and the points are reconstructed into a digital representation of the original object. The aim of this thesis is to present novel grid-based methods and provide several case studies of areas in the 3D modelling pipeline in which they may be effectively put to use. The first is a demonstration of how using a grid can allow a significant reduction in memory required to perform the reconstruction. The second is the detection of surface features (ridges, peaks, troughs, etc.) during the surface reconstruction process. The third contribution is the alignment of two meshes with zero prior knowledge. This is particularly suited to aligning two related, but not identical, models. The final contribution is the comparison of two similar meshes with support for both qualitative and quantitative outputs.