Duliu, AlexandruJakob, VogelSerafima, SisterLasser, TobiasNavab, NassirGabriele Guidi and Roberto Scopigno and Pere Brunet2016-01-062016-01-062015978-1-5090-0048-7https://doi.org/10.1109/DigitalHeritage.2015.7413871When investigating artworks, having the ability to examine the reflectance spectra at every image pixel can us give insight into the materials and pigments used in its creation. Multispectral imaging is a great tool to perform this task, however, when acquiring images with a multispectral camera, in addition to the reflectance spectra, the surface interaction with the illumination is also captured. This hampers quantitative measurements of the surface. We introduce a novel multispectral imaging system and illumination compensation approach, that mitigates the influence of illumination on image formation. We showcase the performance of our system by producing highresolution multispectral mosaic images (low or high dynamic range) of artworks dated from the beginning of the XIX-th century, to the present day.multispectralillumination compensationmosaicinghigh dynamic rangelowrank decompositionrobust principle component analysis10.1109/DigitalHeritage.2015.7413871