Towards an expeditious as-is surface reconstruction

Abstract

Conservation and rehabilitation of structures such as urban buildings and heritage requires a growing effort on the establishment of sufficiently comprehensive documentation concerning the time evolution of these structures’ characteristics and health condition. These documentation archives and databases should be much more than just a collection of structure’s design plans, and desirably they should contain information about the as-built and as-is states of a certain structure. Due to the technological evolution of data acquisition systems, 3D laser scanning points’ clouds became an important data source that enables the subsequent representation of 3D surface’s geometries, and thus the constitution of data repositories. However, it is known that the volume of data obtained, can be enormous and the computation time to process it, a heavy task. The present work proposes an expeditious, continuous processing methodology that aims in a first step the generation of a points’ structured grid and its associated triangle structured mesh by defining an injective function from a virtual surface onto the surface’s cloud points. The additional possibility of selecting the virtual grid dimension to be used, enable representations with different levels of detail. In a second step, the resulting structured grids are used to reconstruct the corresponding surfaces by considering compactly-supported radial basis functions. The main contribution of this work is thus related to the proposal of an expeditious integrated processing methodology of points’ clouds. A set of case studies are presented to illustrate the performance of the present integrated methodology.