Colorimetric Calibration of Video Cameras Using Two Color Constancy Models

Colorimetric calibration of color imaging devices is important to realize device independent color reproductions. However the calibration for digital or video cameras is difficult because a color image is acquired for unknown objects under various unknown illuminants. An experiment was carried out based on two computational color constancy models, the finite-dimensional linear model and the spectral sharpening model, to estimate colorimetric values using image data of a color chart captured by video camera. We estimated the colorimetric values by using a single reference reflectance with known spectral reflectance in a color chart under an unknown illuminant. The accuracy was evaluated by color differences in CIELAB color space and was compared on three different color charts under three different illuminants including a fluorescent lamp. It was confirmed that the two computational models do not require prior knowledge of illuminants and surfaces. The finite-dimensional linear model gave more accurate results than the spectral sharpening model in the simulations and experiments.