Illumination-based and Device-independent Imaging Model and Spectral Response Functions

摘要

This work provides a theoretical framework for the colorimetric scene reproduction under different illuminants. First, based on the principle of CIE color matching experiment as well as image production pipeline in digital still cameras, we analyze and conclude that the Color Matching Function (CMF) of the color space determined by a display is the best Spectral Response Function (SRF) for scene reproduction on the certain conditions. Then, we wonder what the best SRFs for other illuminants beyond D65 are. In order to do this, we propose a new imaging formula in which the SRF is changing with illumination variation. Different from previous imaging formulae suitable for RAW format for D65 illuminants, our imaging formulae is a theoretical one that can achieve color constancy for an arbitrary camera under any illuminants and is suitable for final image formats. We also propose two methods to solve the optimal SRF, including the pseudo-inverse solution and Brandford method. Finally, we compare the proposed method with previous common-used method, and the experimental results can validate the performance of the proposed method.