An innovative approach to automated texture mapping for architectural heritage preservation

Kawther Kharfouchi, Belkacem Labii, Walid Hamma, Hani Amir Aouissi, Soumia Kharfouchi

Rezumat/Abstract. This work will first focus on the various technical aspects of the design of photorealistic 3D renderings of architectural objects. Special attention will be given to the role of new technologies, such as photogrammetry or texture mapping, in the preservation and enhancement of the architectural heritage. The problem of restoring the visual appearance of 3D architectural heritage objects through texture mapping will also be addressed. Hence, an algorithm that achieve a piecewise texture mapping by minimizing length distortion based on artificial neural networks (ANN) and the theory of elasticity was presented. To begin, a surface triangulation S (comprising triangular facets and a boundary) is created. The obtained mesh is then partitioned into disjoint pieces, with each interior vertex belonging to a single piece. Each subsurface, corresponding to one of the obtained pieces, is then mapped independently from the others by imposing local length-preservation constraints and barycentric coordinates of the interior mesh vertices. The proposed method aims to achieve uniform and seamless texturing, allowing for photorealistic representation of 3D models of heritage buildings and cities. This work provides an effective mathematical framework to obtain geometrically precise texture generating process on surfaces of arbitrary architectural topology.

Cuvinte cheie/Key words: architectural heritage, 3D representation, 3D rendering, texture mapping, ANNs

Text integral/Full text