A self-learning based color correction system which employs the CMAC (Cerebellar Model Articulation Controller) neural network with B-Spline receptive field functions is proposed. The CMAC neural network learns color input and output characteristics of target printer and models the printer. An inverse printer model is then built based on the CMAC printer model. This scheme is adaptive because the CMAC neural network can model all kinds of printer. Experimental results show that the average color difference between the desired color and the printout can be reduced to 3.7 ΔEab when a Kodak DS8650PS dye sublimation color printer is used.
Jin-Jou Chen, King-Lung Huang, "Adaptive Color Correction by High-Order CMAC Neural Network" in Proc. IS&T 5th Color and Imaging Conf., 1997, pp 182 - 186, https://doi.org/10.2352/CIC.1997.5.1.art00036