Color calibration of an RGB digital camera for the microscopic observation of highly specular materials

Color calibration of imaging devices has been previously studied in a varied number of situations where the materials observed have diffuse or only slightly specular surfaces. Most of the calibration methods available in the literature consist in using standard diffuse color charts in order to determine the mathematical operations necessary to transform the colors measured by the imaging device into the reference colors obtained from the target chart. Unfortunately, there are many problems, such as sensor saturation, that arise when using these methods to calibrate devices intended for the observation of highly specular samples, especially in the 0 degrees:0 degrees illumination/observation geometry used in microscopic imaging systems. In this paper, we explore several color calibration methods adapted for the observation of highly specular materials, and propose one method in particular in which we use colored filters and a calibrated mirror in order to obtain a set of specular colored samples. By using 72 samples for learning, we tested the different methods on 50 other samples and obtained, with the best one, an average CIE-DeltaE94 color difference of 1.93 units, which is a fairly good performance for color measurements at the microscopic scale.