基于胞元式 RBF 神经网络的高保真分色模型研究

采用胞元式RBF神经网络模型对七色印刷输出系统构建了分色模型。首先,借鉴颜色空间分区理论将7个主色在整个颜色空间中划分为了6个颜色区域,在每个分区中选取了CIE L*a*b*明度值L上等间隔均匀采样的网点面积率,用于设计建模所需的训练样本,然后对每个分区划分胞元,并且为每个小胞元建立了基于RBF神经网络的分色模型。对于任意给定的要复制的目标色,利用提出的胞元搜索算法确定其所在的胞元位置后,使用相应的神经网络模型进行分色预测。实验结果表明,该分色算法能够达到较高的分色精度,可以满足高质量彩色复制的要求。

Research of Hi-Fi Color Separation Model Based on Cellular RBF Neural Network

Color separation models for seven-color printing system were established using cellular RBF neural network. Firstly, according to color space partition theory, the color space of the printing system was divided into 6 partitions by 7 primary colors. Dot areas, which were sampled from L of CIE L*a*b* with equal space, were then selected as training samples of RBF neural network model in each partition. Each partition was subdivided into several cells and the color separation algorithm based on RBF neural network model for each cell was established. Any target color would be separated into the CMYKRGB dot areas according to the color separation models of the numbered cell which were determined by the cell search algorithm proposed. The experiment result showed that the color separation algorithm can achieve high accuracy of color separation for seven-color print production.