3D reconstruction of archaeological contexts using image segmentation and Structure from Motion

Into the large set of techniques to elaborate 3D reconstructions, the Structure from Motion algorithms stand out because of its low cost and versatility, being these algorithms a series of computational procedures to the automated generation of dense point clouds from a set of photographs. The goal of this paper is to propose a workflow for the 3D reconstruction of archaeological objects and contexts employing a low cost UAV (Unmanned aerial vehicle) for the data capture phase, searching to evaluate and problematize the need of use expensive and specific hardware. After this data capture stage the 3D reconstruction process begins. This workflow is composed by three great phases: The generation of binary masks to select regions of interest, the production of dense point clouds and its transformation into a triangular mesh. Finally, sketch out the challenge that means aspire to determine which is the best of the generated reconstructions under quantitative frames.