3D survey and documentation in building archaeology: The medieval church of San Niccolò in Montieri

Not far from Massa-Marittima (Tuscany, Italy), stands the medieval castle of Montieri, built in 1133 by the Bishop of Volterra. In the Middle Ages, thanks to its metalbearing ores, the castle was the focus for a complex history. During the last few years, the University of Siena was involved in studying in the archaeology and architecture of the castle and the surrounding old silver mines. The last survey campaign was focused on the area so-called "La Canonica", San Niccolö ecclesiastical complex. The excavation came up to light the existence of a church characterized by an unusual plan with six apses (the only example of its kind in Italy) [1]. During the survey an innovative image-based modelling (IBM) workflow for quick 3D acquisition in archaeology was tested. Building archaeology aims at reconstructing the history of existing building and analyzing materials, building techniques and, above all, the "vertical" stratigraphy. Therefore, during the survey of the Church, in order to get a data-set suitable for this kind of analysis, we had to provide a thorough metric analysis, for precise archaeological documentation and to better understand the development of the building. For this purpose, IBM as acquisition method was chosen. This recent technology allows a complete and detailed 3D to be obtained, in very short time and with very low budget, from a set of un-calibrated photos. The modeling process is carried out automatically by software that combines algorithms of computer vision. The automatic approach implies a lower precision in comparison with conventional scanning techniques, nevertheless, recent tests have demonstrated that it can be sufficient reliable and have already being used with success in archaeology [2]. The first step was the establishment of a local topographical network as a reference coordinates system. The topographic network constituted the basis for scaling and alignment operations and for controlling the geometrical precision of the 3D model. Then, to obtain a dataset of photos, a photographic campaign acquisition was carried out. Images were gathered using a camera with a 35 mm focal length and taken from different positions to better cover all the parts of the monument. Once acquired, the photos were imported into Photoscan Agisoftware (PA) to build the 3D models. As first step the software computed the camera position for each photo to derive the point cloud model. To optimize camera positions and to provide a reference grid to scale and orient the models, the markers acquired in the topographic campaign were use as control points (Fig.1). As second step 3D model reconstruction was computed, and then the model generated was optimized in Meshlab, a software for editing mesh. Finally the optimized model was imported in PA to build the texture. The IBM approach has proved to have multiple advantages such as short time of acquisition and post-processing of the data, very detailed textured models and economical equipment. The final model assured a good documentation by allowing a detailed characterization of the stone surface. The model was rendered to obtain rectified images of façades of the apses and to derive thematic maps of the construction phases, lithotypes and stratigraphy of the Church. Finally the use of the 3D model turned out to be very useful during the excavation campaign too, in order to obtain a visual perception of the entire building from non-traditional point of views and to better understand the evolution of the architecture (Fig. 2).