CAA2016

Fusion of 2D and 3D Imaging Data for the Non-Destructive Assessment of the Baptistery di San Giovanni in Florence, Italy

Michael Hess, Michael Yeager, Vid Petrovic, Falko Kuester

Abstract
Three-dimensional data capture has long enabled visually compelling renderings of target environments, but in most cases the generated data are not utilized to their fullest potential for scientific analysis and interpretation. In addition to realistic rendering, 3D data streams from techniques like laser scanning and photogrammetry can be used for analysis of raw 3D point data and they can also serve as a digital scaffold which can be used to spatially anchor other data streams into one holistic model that serves as the digital surrogate. A case study at the Baptistery di San Giovanni in Florence, Italy is discussed here, wherein multiple 2D and 3D non-destructive techniques were used to digitally document the monument in order to study its construction. Terrestrial laser scanning (TLS) was used to measure accurate, high-resolution geometry throughout the structure. Portions of the raw data were analyzed in order to quantify and visualize the effects of centuries of structural changes and aging. The dense set of 3D coordinates also serves as the geometric scaffold used to anchor thermal imaging, ground penetrating radar and photogrammetry data. The analyzed point data as well as the data from the other imaging modalities can now be visualized together within the entire laser scanning model of the Baptistery. The utilized visualization environment allows for interactive exploration and manipulation of the holistic digital surrogate. Interactive visualization of the data enables more effective communication of the imaging results to stakeholders and facilitates collaboration with different domain experts for further analysis and interpretation of the multimodal data. The presented methodology for multimodal data fusion can be repeated to incorporate other data types and the flexible visualization environment supports the evolution and growth of input data.