A fuzzy-neural system for removal of the color cast for digitally captured images under unknown illuminants

This paper proposes a novel color balance technique by establishing a fuzzy inference system. A database is constructed to include certain properties of various color samples captured under standard illuminants. Next, a fuzzy system is established, whose configuration and parameters are determined using the aforementioned database. Finally, test image captured under unknown illuminant may be passed through the fuzzy system to produce a set of ratios for correcting and removing the color cast. An experiment and comparative study justify that the suggested approach is preferable to existing methods with regard to execution time and the performance.