神经网络模型在分色算法中具有预测不确定和精度较低的缺点,在其基础上提出了补偿模型。该模型将神经网络预测CMYK 的各通道差值拟合成L*a*b* 多项式的函数,并对CMYK 值进行补偿。通过对该模型的实验测试表明,采用CMYK 值进行补偿后与原模型相比,提高了分色算法的精度,并且其分色精度要高于其他常用的模型。

A compensation model based on neural network was put forward to solve the prediction uncertainty low precision problems of neural network model in color separation algorithm. The model synthesizes each channel difference of neural network predicting CMYK to L*a*b* polynomial function, and compensates for the CMYK values. Experimental results showed that the new model has improved accuracy of color separation algorithm after CMYK value compensation; separation accuracy of the model is higher than other commonly used models.