Proposal of an integrated 3D architectural survey method for application in historic agri-food building analysis and representation

Abstract

In Italy historic agri-food buildings can be considered a relevant material expression and testimony of century-old agriculture and food processing practices handed down by generations. Recently they have gained ever-growing importance as a part of the wider architectural heritage. As such, they deserve dedicated general surveys to build a thorough knowledge of their distinctive characteristics and investigate their current condition, this way setting the basis for the implementation of planning and management actions for their sustainable valorisation. To this end, building information modelling (BIM) can be considered an efficient strategy to preserve construction information by creating 3D models based on surveys of the built heritage. To acquire in fast and accurate way geometric, reflectance and colour data of rural buildings, as 3D point-cloud, the terrestrial laser scanner (TLS) represents a powerful tool. The traditional TLS-based survey methods, in the context of historic agricultural buildings, have several limitations, mainly due to the presence of inaccessible parts and bulky machinery once used for processing and storage. In the present research, to overcome these issues and thus have a complete survey, we describe a proposal of an integrated methodology for obtaining 3D point-cloud data of existing rural agri-food buildings based on the integrated use of TLS, handheld scanner (HS), and unmanned aerial vehicles (UAV) instruments. The proposed methodology was tested in surveying three historic agri-food buildings and the accuracy of the obtained 3D point-cloud was calculated by means of the root mean square error (RMSE) on the X, Y, and Z alignment of the two different 3D point-clouds in correspondence of the used B/W target. Moreover, a measure of the distance between two merged 3D point-clouds, in their overlap area has been performed using the multiscale model to model cloud comparison (M3C2). RMSE analysis always shows values lesser than 1 cm and M3C2 shows values between 0 and about 6 cm.