Abstract
A method is tested for analyzing color video images to determine film thickness for transparent coatings on silicon. The method is robust to reasonable variations in light level, video parameters, and computer image capture spectral characteristics. The main standardizing feature of the method is to include characterization of a bare substrate along with the unknown samples (either within the view frame or as a separate sequence under identical illumination conditions during video capture). Then, the color appearance for the unknown sample can be referenced to the color appearance of the bare substrate, thereby enabling a quantitative match between the measured color and the predicted color characteristics. The technique is tested with transparent spin-on-glass (SOG) coatings that were calibrated separately. The limits of applicability and the level of accuracy that can be obtained are discussed in detail.