一种基于神经网络和Neugebaure方程的输入图像色彩校正模型

A Color Error Correction Model for Input Image Based on Neural Natworks and Neugebauer Equation

在分析扫描物呈色原理和误差产生原因的基础上,提出了一种扫描输入图像的色空间转换的修正模型.首先重新解释Neugebauer方程的参数含意,使原本只适用于印刷网点图像的Neugebauer方程也适用于非印刷网点图像.随后结合Neugebauer方程和RBF神经网络,分单色、双色和三色逐步导出输入图像的色彩修正方程.最后给出的实验结果表明,与目前具有较高转换精度的主流模型比较,本文模型能够明显提高转换精度.

A modified model of color space conversion is presented based on analyzing the color rendering principle of scanning object and the causes of color error. The paper reinterprets the parameters of Neugebauer equation which can be used only for printing dot image originally and makes it be for non-printing dot image. Neugebauer equation and RBF neural network are combined to deduce the modified model for color space conversion for scanning input image through single color correction, bicolor correction and tricolor correction stepwise. The experimental results show that the modified model can improve color space conversion accuracy compared with mainstream equations which have relative high conversion accuracy.