A method to improve the color rendering accuracy in cultural heritage: preliminary results

Abstract

Color specification is an important challenge in many application domains including Cultural Heritage. The collection of metadata concerning Cultural Heritage involves the valorization, fruition and becomes part of the conservation process. It becomes essential to find methods that simplify and optimize the acquisition of such data as color information. In this regard, in this work we present the preliminary results of a project that involves the acquisition by 3D scanner of samples of different colors placed in a controlled environment and with different illumination conditions. To make more accurate the color rendering, the color coordinates of each sample were measured by a spectrophotometer. All the obtained measurements become part of a dataset with which to train a machine learning model that learns how to perform the transformation from the RGB to the CIELab color space in different lighting condition.

Color specification for color rendering

“…In order to estimate the 3D scanner color rendering, a region of interest (ROI) of 5x5 pixels was selected inside the images and the mean value for each R, G and B channel was calculated. Finally, the RGB to CIELab color transformation was performed using the python image processing toolbox scikit-image, the formula conversion parameters were set using the D65 as the reference illuminant and 10 degrees as the aperture angle of the observer (as performed in distance and contact measurements) [10,11].…”