Novel digital paradigms to support the Architectural Heritage knowledge process

In the field of the documentation process for supporting conservation activities of Architectural Heritage (AH), it is necessary to reach a deep and complete knowledge evaluating the history, physical configuration, condition assessment, and cultural conditions in terms of social, political, and economic aspects, in relation to the building and the environment in which it is located, in order to determine precise protective measures. However, today, the ambiguity of workflows in the knowledge process, the fragmentation of previous documentation, located in paper and digital archives, and the use of inadequate tools make the conservation process very critical. The risk consists of errors in the assessment of intervention strategies with repercussions on the economic and social system producing the loss of value to the heritage asset. In accordance with what is stated in the "London Charter" (2009) - which establishes general methodological principles for digital visualization applications in the field of research and communication of AH - undoubtedly the technological contribution can change the way of perceiving the representation of the historic manufacts. Indeed, the introduction of huge digital archives, that can store data deriving from the different types of information, may enrich the 3D models better defining the historical heritage itself. The development of Computer Vision and digital modelling techniques has enabled increasingly exploiting ICT (Information and Communication Technologies) technologies, and specifically, information systems in the Heritage field. Virtual 2 repositories allow more effective control over the documentation process permitting the storage of different and heterogeneous data produced on built heritage. Specifically, Building Information Modeling (BIM) is defined as a complete digital reproduction of an architecture that contains the attributes of digital parametric models. Its introduction into the world of representation has radically changed the meaning of three-dimensional models, proposing them as systems information for architecture. Considering the panorama of AH refers to Historic Building Information Modeling (HBIM) which is a system of representation of built architecture composed of libraries of objects semantically structured and parameterized. HBIM creates new possibilities for sharing different information within a single digital environment and is a way to increase portability and provide data to a wider community of users. The BIM methodology can be a bridge between archival documentation, architectural survey and digital model, proving to be an effective tool as a semantic data archive, consisting not only of geometrically correct elements but also of alphanumeric and digital attributes (texts, multimedia files, URL links, etc.). The HBIM models are understood as data hubs that in the documentary field, can represent the historical artefacts taking into account the past and the present as a result of constructive transformations, enlargements and changes of use in a wide time span. HBIM models can be constructed from three-dimensional data derived from digital surveying techniques, such as laser scanner and photogrammetry. The process, called reverse engineering, no longer has the goal of translating geometric survey into two-dimensional plans, elevations and sections, but rather is a starting point for three-dimensional modelling, striving to move from solid to parametric objects, described geometrically and semantically. Furthermore, from literature emerges that, nowadays, there is no structured and shared digital framework that takes into account the knowledge gathered on the assessment of the state of the buildings, the morpho-metric survey, the report of conservation activities as a whole integrated system. One of the difficulties consists in the use of standardized parametric objects which not fulfil the complexity and 3 uniqueness of architectural elements (for example vaults, irregular masonry walls, etc.). The following PhD thesis proposes a methodological workflow that, starting from the digital survey, addresses the issue of modelling and semantic enrichment of HBIM models focusing on the pre-diagnostic analysis of the state of the places, reaching the process of managing data useful for the activities of conservation of historical heritage. The proposed research connects the aims of two disciplinary areas, architectural survey and representation, and refurbishment by implementing consolidated methodologies with the objective of defining an optimal strategy modelling the fourth dimension (time) of AH in the HBIM environment, to avoid risks of assessment of the state of the places and a coherent future conservation activity. The goal is to achieve complete and flexible management of the large amount of multidisciplinary information processed during the activities of knowledge, survey, analysis and planning of conservation activities. Therefore, the main objectives of this thesis consist, firstly, in the validation of a consolidated process of representation in BIM that allows to overcome the limitations due to the standardization of parametric objects respecting the specificities of the instances of the historic asset, secondly, in the implementation of semantic segmentation methods through standard (ragion-based, model-based, etc.) and Machine Learning (k-means clustering, random forest, etc.) algorithms to automatically structure and classify 2D and 3D output (orthophoto, UV Map, point cloud, textured mesh, etc.) of digital survey in order to quantify architectural components and the different states of alteration of surface materials essential for prediagnostic activities, and to reduce the time and manual steps of the Scan to BIM process; finally in the integration and the improvement of the semantic level of BIM models by extending the field of representation to several levels of detail, and including the knowledge needed to understand the characteristics of the built heritage that cannot be directly included in its physical components (i.e. history, construction phases, survey data, etc.), through the develop of methods for integrating the 4 information contained in BIM models through Database Management Systems and 4D simulation. The validation of the proposed methodology will be carried out on case studies chosen within the Italian Architectural Heritage and starts from the survey and ends with the development of three-dimensional models for the management of historical architecture spendable in areas of different application: accessibility and enhancement, archival and documentary research, conservation and artificial intelligence.