Surface completion is an active area of research in computer graphics. Many times the mesh that represents a physical object has holes that may not be due to the scanning process, but rather because the item is broken or incomplete. Most algorithms for filling holes in meshes are context-based, that is, they seek patches from other areas of the original mesh.

Harary et al. go a step further. They propose a method that is coherent in the sense that each patch used to complete the hole is contained in the original mesh. This assures that the features in the surface used to fill the hole are not very different from those found on the rest of the mesh.

Although their algorithm has some limitations for large holes without many features in the immediate area, the examples the authors present show their methodology’s clear advantage when compared to other similar methods.