Measuring changes in cultural heritage objects with Reflectance Transformation Imaging

• Published in: 2013 Digital Heritage International Congress (DigitalHeritage) Vol. 1; pp. 189 - 192
• Main Authors: Manfredi, Marcello, Williamson, Greg, Kronkright, Dale, Doehne, Eric, Jacobs, Megan, Marengo, Emilio, Bearman, Gregory
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2013
• Subjects: Cultural differences; Cultural Heritage; Imaging; measuring damage; Monitoring; normals; Polynomial Texture Mapping; Polynomials; PTM; Reflectance Transform Imaging; Rephotography; RTI; Sea surface; Sensitivity; Surface morphology
• ISBN: 1479931683; 9781479931682
• DOI: 10.1109/DigitalHeritage.2013.6743730

Sites and objects of cultural heritage - from art to ancient inscriptions to ruins - are under constant attack by time and the environment. While much is known about how material components change from laboratory-based artificial aging, very little is known about the process or rates of change of actual objects and sites in situ. Reflectance Transformation Imaging (RTI) is a quantitative method that captures surface normals. In our case, it provides detailed information on the geometry of the object surface. We show that RTI can be quantified for use as a method for measuring change in cultural heritage objects. The past decade has seen the rapid evolution and application of computational photography methods to document important works of human heritage, from art and architecture to archives and archaeology. The next logical step involves defining just how reproducible and precise these methods can be to use them to measure rates of change for important works of cultural heritage. The need is to move to calibrated, quantitative image datasets for reproducible imaging. We measure the precision of computed surface normals , which define the basic repeatability of RTI. Our results show that the average included solid angle for RTI sensitivity fitted to the Hemispherical Harmonics (HSH) polynomial function is 0.003 steradians (3 sigma), while the older Polynomial texture map (PTM) method is much less sensitive (0.5 steradians). The absolute sensitivity of the method is the minimum variation of the normal that can be statistically considered a change of the object. It is calculated considering the average value of the normal of each single pixel. The solid angle of the cone of variation represents the statistical limit (3 *σ). Analysis of multiple RTI data sets from objects that have changed between image capture sessions results in a map of change that can easily be evaluated by conservators.