2019/11/07 by Gabriel Riutort‐Mayol, Michael Riis Andersen, Riutort-Mayol, Gabriel +5
Arts and Humanities · Chemistry · Environmental Science · Physics and Astronomy · Psychology · #Applications (stat.AP) #Color Science and Applications #Color perception and design #Conservation Techniques and Studies #FOS: Computer and information sciences #Remote Sensing in Agriculture #Spectroscopy and Chemometric Analyses
paper · pdf · doi:10.48550/arxiv.1911.03454
openalex publication_date 2019/11/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Microfading Spectrometry (MFS) is a method for assessing light sensitivity\ncolor (spectral) variations of cultural heritage objects. The MFS technique\nprovides measurements of the surface under study, where each point of the\nsurface gives rise to a time-series that represents potential spectral (color)\nchanges due to sunlight exposition over time. Color fading is expected to be\nnon-decreasing as a function of time and stabilize eventually. These properties\ncan be expressed in terms of the partial derivatives of the functions. We\npropose a spatio-temporal model that takes this information into account by\njointly modeling the spatio-temporal process and its derivative process using\nGaussian processes (GPs). We fitted the proposed model to MFS data collected\nfrom the surface of prehistoric rock art paintings. A multivariate covariance\nfunction in a GP allows modeling trichromatic image color variables jointly\nwith spatial distances and time points variables as inputs to evaluate the\ncovariance structure of the data. We demonstrated that the colorimetric\nvariables are useful for predicting the color fading time-series for new\nunobserved spatial locations. Furthermore, constraining the model using\nderivative sign observations for monotonicity was shown to be beneficial in\nterms of both predictive performance and application-specific interpretability.\n