Digital reconstruction of historic roof structures: developing a workflow for a highly automated analysis

Abstract

Planning on adaptive reuse, maintenance and restoration of historic timber structuresrequiresextensive architectural and structural analysis of the actual condition. Current methods for a modellingof roof constructions consist of several manual steps  including  the  time-consuming  dimensional modelling. The  continuous  development  of terrestrial laser  scanners increases  the  accuracy,  comfort  and  speed  of  the  surveying  work inroof constructions. Resultingpoint  clouds enabledetailed visualisation of theconstructionsrepresented by single points or polygonal meshes, but in fact donot containinformation  about  the  structural  system  and  the  beam  elements.  The  developed  workflow  containsseveral  processing steps on the point cloud dataset. The most important among them arethenormal vector computation, the segmentation of points to extract planarfaces, a classification of planarsegmentsto detect the beam side facesand finally theparametric modelling of the beams on the basis of classified segments. Thisenablesa highly automated transitionfrom raw point cloud data to a geometric model containing beams of the structural system. The geometric model,as well as additional information  about  the  structural  properties  of involved wooden  beams  and  their  joints,is necessaryinput  for  a  furtherstructural modellingof timber constructions. The results of the workflow confirm that the proposed methods work well for beams  with a rectangularcross-section  and  minor  deformations. Scan  shadows  and occlusionof  beamsby additional installationsor interlockingbeamsdecreases  the  modelling  performance,  but  in generala high  level  ofaccuracy  and completeness isachieved ata high degree of automationHighlights: This article presents a novel approach to automated reconstruction of beam structures by modelling geometry from segmented point clouds.Wooden beams are modelled as cuboids, thus a rectangular cross-section with minor deformation is required.An accuracy of less than 1 cm can be reached for modelled beams, compared to the reference LiDAR point cloud.