基于广义回归神经网络的数码打样色彩空间转换方法的研究

由设备无关的色彩空间CIE L*a*b*与设备相关色彩空间CMYK转换是图像输出设备特征化和色彩管理模块的关键技术。基于数码打样样张的测量数据采用广义回归网络分别建立了CMYK与CIE L*a*b*色彩空间转换的正反向模型，并分别应用色差公式进行精度检验，研究结果表明基于广义回归网络建立起来的CMYK与CIE L*a*b*色彩空间转换模型是实现色空间转换的有效方法，该模型无论从训练的简便性、训练速度、还是精度上都比BP神经网络模型有优势。

Color space conversion algorithm of generalized regression neural network based on digital proof

Color conversion algorithm of CMYK and CIE L*a*b* is the key technique in output device characterization and color management module. In this paper the forward and reverse color space conversion models between CMYK and CIE L*a*b* based on Generalized Regression Neural Network (GRNN) were built using the color data of the digital proof. Then the accuracy of the models were tested. Finally, The result shows that it is an efficient method to build the color space conversion between CIE L*a*b* and CMYK using GRNN, which gains an advantage over BP network not only in training convenience, speed but also precision. network not only in training convenience, speed but also precision.