Analytical Strategies for Large-Area X-Ray Spectroscopic Imaging Measurements of Cultural Heritage Objects: A Case Study in Illuminated Manuscripts

X-ray fluorescence (XRF) mapping is a noninvasive imaging method that reveals the elemental composition of a material. XRF mapping is valuable for studying cultural heritage materials because detailed composition information about trace elements may provide new information about origins of historic pigments. Frequently, researchers collecting XRF data wish to analyze the relationship between spatial distribution and concentration of multiple elements in the scan. Previously, this analysis has been limited to directly comparing two or three elements at a time. This project focuses on expanding XRF mapping data analysis to include all elements present in the sample at a time, so that the entire sample can be described by a small number of characteristic ratios. New Python code was written to test several methods of pixel classification and we find that generally SVM plus k-means classification is a versatile method.