Caractérisation de formes architecturales. Une approche expérimentale intégrant complexité et intelligibilité des représentations numériques
English title
Characterisation of architectural shapes. An experimental approach integrating the complexity and intelligibility of digital representations
Supervisors
Judith le Maire de Romsée, Professor (ULB) and Livio De Luca, Research Director (CNRS)
Members of the jury
Victor Brunfaut, Professor, Faculté d'Architecture (ULB) – Président / Jean-Yves Blaise, Researcher (CNRS), FR / Marc Daniel, University Professor, H.D.R (CNRS), FR / Pascal Liévaux, Head of the Department for Research and Scientific Policy (Ministry of Culture and Communication), FR
Abstract
These last three decades, the integration of digital technology in the field of architectural practice, and particularly in the study of building heritage, generated major upheavals. Through a historical perspective of architectural survey, this research discusses the cognitive dimension intrinsic to its practice and to the technological challenges it faces.
On that basis, it appears that the digitizations coming from the application of the latest acquisition methods (photogrammetry & lasergrammetry) express high completeness (visual and metric) with the artifacts surveyed, but also produce significant data volumes. This "information overload" does not reinforces the architectural representation in its role of knowledge vehicle. Facing this lack of intelligibility, the purpose of this research is to provide an epistemological answer considering the need of intelligibility without loosing the value of digitizations. This research offers an original low level (non-interpretative) approach, where the meaning of elements ("semantic" and "geometric" structure) comes from the morphological similarities observed on a corpus of digitized shapes. The statistical analysis and the use of different morphological descriptors help to formalize characteristic signatures of the corpus; they correspond to the formal expression of a determined number of geometric features. This approach provides a methodological solution for "big data" processing.
Based on accumulation of data, it offers a complementary analytical response to "traditional" high-level approaches (interpretative) where the shape is characterized with pre-structured knowledge specific to the architectural field. If one of the challenges is to apply geometric analytical methods on "big data", the other one is to confront, through the comparative study, high and low level observations. The purpose is to refine the understanding of the stylistic propagations in time and in space. To experiment this approach, 31 columns of the cloister of the Abbey of Saint-Michel-de-Cuxa have undergone a thorough morphological analysis.
Keywords: Architectural representation, survey, digital, architectural heritage, shape analysis, semantic description.
