Despite the rise of conservation science with an ever-increasing sophistication of chemical characterization tools, for over 100 years, the polarizing light microscope (PLM) has remained one of the most important analytical techniques for identifying artist pigments. This project aims to maximize the amount of information extracted from PLM by introducing digitization and image processing protocols designed to make the data collection/interpretation process more accurate and user-friendly. By optimizing the collection of refractive index and birefringence data with machine learning methods, the desired outcome of the research would be to revitalize PLM as a state-of-the-art technique for pigment characterization and for the method to find renewed and broad applicability in conservation science investigations. In this three-year project made possible by the National Endowment of the Humanities (NEH), the postdoctoral fellow will develop a single-shot polarizing light microscope that utilizes a new machine-vision camera with polarizers attached directly to the sensors at the pixel level. To identify pigments, the fellow will develop deep learning architectures to perform multi-class classification of the pigments from the PLM image. The postdoctoral fellow will be mentored by a project team consisting of researchers specializing in signal processing (Northwestern University, Chicago), optics (University of Arizona, Tuscon), and cultural heritage (M+, Hong Kong) and will utilize the Forbes pigment archives as source material.
Plus d’informations :
[Website Northwestern University]