Characterization and visualization of reflective properties of surfaces

Images play a vital role in several fields of natural science research, including biology, physics, astrophysics, and computer science. In the natural sciences, images are commonly used in measurements or documentation; such applications include images made with telescopes, optical microscopes, or electron microscopes. In the humanities, images also play an important role in research. In art history, for example, many different types of images, from photos of small objects to three-dimensional reconstructions of buildings, help art historians to develop theories, to discuss them with other scholars, and to document the current state of artworks, e.g. in the process of restoration. This is particularly useful if the object is not easily accessible, in which case a common solution is to work with photographs. Digital photography has simplified the process of visual representation, because digital images can be easily shared and made accessible.
However, when it comes to more complex kinds of artworks like mosaics, these static and two-dimensional images are not able to reproduce the actual visual impression of the object. Similar considerations apply to a variety of other artifacts, such as early prints, books, parchments, and textiles. The challenge in the digitization of of these objects lies in their complex surface properties and reflection behavior.
A promising way to solve those limitations is the use of Reflectance Transformation Imaging. RTI is a set of computational photographic methods that capture a subject’s surface shape and color, making it possible to interactively re-light the subject from any direction by means of a mathematical model. The major drawback of RTI is the limitation of the applied mathematical model. Other drawbacks are the RTI imaging workflow and the fact that display of RTI requires a particular stand-alone application.
In this thesis, the author developed a data-driven scientific approach to reproduce surfaces composed of lambertian and glossy materials using the RTI technique with as few parameters as possible. This new approach has been called eRTI (enhanced Reflection Transformation Imaging). Furthermore the hardware needed to acquire RTI and eRTI has been improved, by collaborating with a local Swiss firm to develop a novel solution for image acquisition. Lastly a web-based viewer has been developed, to render eRTI images in any standard web browser, even on most mobile devices.
The qualities of eRTI have been tested using a novel approach that includes a quantitative and a qualitative method. The results show agreement between the techniques.