Closing gaps by clustering unseen directions

Although in recent years the 3D-scanning field has reached a good level of maturity, it is still far from being perceived by common users as a 3D-photography approach, as simple as standard photography is. The main reason for that is that obtaining good 3D models without human intervention is still very hard. In particular, two problems remain open: automatic registration of single shots and planning of the acquisition session. In this paper we address the second issue and propose a solution to improve the coverage of automatically acquired objects. Rather than searching for the next-best-view in order to minimise the number of acquisitions, we propose a simple and easy-to-implement algorithm limiting our scope to closing gaps (i.e. filling unsampled regions) in roughly acquired models. The idea is very simple: detect holes in the current model and cluster their estimated normals in order to determine new views. Some results are shown to support our approach