The architectural 3D survey vs archaeological 3D survey

In the last few years, the field of building architecture and archaeology survey's trend has been to implement a methodology which essentially converges to several procedures used indiscriminately in both environments. Such a methodology consists of: a first phase of data acquisition, carried out through well-established 3D survey's procedures, which merges range based and image based model, linked to a database; a second phase of data processing and feature extraction of the relieved object and a third phase, following closely the previous two, of 3D modelling restitution organized by elements with related data. Our research focus is to define specific procedures usable in the two environments and to make them more objective through the development of original algorithms which automatize several steps. Our proposed pipeline, both in the field of archaeological and architectural survey, consists of a first and essentially univocal phase of data acquisition and organization whose product is an implemented point cloud, organized by separate 3D elements linked with a corresponding database. It's in the second phase of data processing where the two procedures substantially diverge: in the case of archaeological structure survey we propose to generate a triangulated mesh aimed to obtain continuous and textured 3D models of the single fragments. Feature extraction is applied on the mesh model, through original focused algorithms, after its orientation in the original reference system. Moreover, in the specific case of virtual reconstruction, we use a reference model based on the knowledge and analysis of the database, in which every single fragment will be repositioned following its hypothetical original place. In the case of architectural survey the mesh definition procedure is not needed for feature extraction, which is applied directly on the point cloud. In this environment we construct a reference model based on available data and features which leads to the final model fitted to local variations due to deformations, defects, variations etc. It is clear that, despite the common use of a reference model made out of knowledge and following a similar geometrical/analytical construction based on collected database information, the final output in the two environments considered here are substantially different: one is a mesh based model (combining mesh and geometry based model, in the virtual reconstruction), the other is a geometry based model, locally adapted.