VPL geometry processing for open multilevel and multiscalar databases

Abstract. This study investigates open-source VPL algorithms for storage, geolocalization and comparison of diagnostic multi-sensor data on 3D semi-automatic geometric database extracted from low-cost image-based models and shared on open web-based platform to support information, conservation/restoration and monitoring of multiscalar elements (e.g. artworks, architectural or urban objects).
The information system consists of an interoperable parametric box/synoptic mesh/grid for semi-automatic definition and segmentation of 2D and 3D models of artworks, regardless of size, materials and digitalisation technique to validate interoperability between VPL algorithm and different information systems/approaches (e.g. BIM/HBIM, EM, multiplatform graphic engines): restorers and diagnosers can geolocate and annotate numerous results, observations, hypotheses or indications in geometric cells to visualize and compare heterogeneous visible and invisible structural and material data at different levels of detail. The database consists of a neutral grid extracted from the 3D reality-based mesh to wrap points cloud, surface or mesh: box or grid cells work as containers for storage of diagnostic information; the texture is however displayed in 3D modeler/viewer as visual reference for data geolocalization. The system allows semi-automatic clustering of areas, cells, colours attribution and labelling of text and image from link data list. The algorithm has been tested on high resolution images and architectural/structural object-oriented elements (VPL-BIM interoperability) and it is in progress on digital photogrammetric models of artworks. Work in progress consists of testing the algorithmic database on the largest number of works and elements, also considering the integration of automatic AI Data annotation/comparison as possible future works.