High-accuracy 3-D modeling of cultural heritage: the digitizing of Donatello's "Maddalena"

IEEE Trans Image Process. 2004 Mar;13(3):370-80. doi: 10.1109/tip.2003.822592.

Abstract

Three-dimensional digital modeling of Heritage works of art through optical scanners, has been demonstrated in recent years with results of exceptional interest. However, the routine application of three-dimensional (3-D) modeling to Heritage conservation still requires the systematic investigation of a number of technical problems. In this paper, the acquisition process of the 3-D digital model of the Maddalena by Donatello, a wooden statue representing one of the major masterpieces of the Italian Renaissance which was swept away by the Florence flood of 1966 and successively restored, is described. The paper reports all the steps of the acquisition procedure, from the project planning to the solution of the various problems due to range camera calibration and to material non optically cooperative. Since the scientific focus is centered on the 3-D model overall dimensional accuracy, a methodology for its quality control is described. Such control has demonstrated how, in some situations, the ICP-based alignment can lead to incorrect results. To circumvent this difficulty we propose an alignment technique based on the fusion of ICP with close-range digital photogrammetry and a non-invasive procedure in order to generate a final accurate model. In the end detailed results are presented, demonstrating the improvement of the final model, and how the proposed sensor fusion ensure a pre-specified level of accuracy.

Publication types

  • Comparative Study
  • Evaluation Study
  • Research Support, Non-U.S. Gov't
  • Validation Study

MeSH terms

  • Algorithms*
  • Anthropology, Cultural / methods
  • Archaeology / methods*
  • Archives
  • Culture
  • Data Compression / methods*
  • Image Enhancement / methods
  • Image Interpretation, Computer-Assisted / methods*
  • Imaging, Three-Dimensional / methods*
  • Models, Statistical
  • Pattern Recognition, Automated
  • Quality Control
  • Reproducibility of Results
  • Sculpture*
  • Sensitivity and Specificity
  • Signal Processing, Computer-Assisted*