RBF神经网络在显示器色空间转换中的应用

目的研究RBF神经网络对显示器色彩空间转换预测准确性的方法。方法通过编程借助Measure Tool软件自动测量获取建模和测试数据,通过反复测试选择建模合适的函数和参数,最后用RBF神经网络模型进行仿真实验,以获取较好的RGB转换到Lab色空间的转换模型。结果 RBF神经网络模型测试得到的色块平均色差达到0.75,最大色差达到19.7。结论该方法建模简单方便,网络训练速度快,转换精度高,对显示器色彩空间转换具有较好的非线性拟合能力和更高的预测准确性。

Application of RBF Neural Network in Color Space Conversion of Display

Objective To study the method for RBF neural network accuracy prediction of color space conversion of display. Methods The modeling and testing data were obtained by automatic measurement through programming with MeasureTool software, and suitable functions and parameters for modeling were selected by repeated testing, finally simulation experiment was carried out using RBF neural network model, in order to obtain a better model for conversion of RGB to Lab color space. Results The average color difference obtained in RBF neural network model tests was 0.75, and the maximum color difference reached 19.7. Conclusion The results showed that this modeling method was simple and convenient, with fast network training speed, high conversion accuracy, and had good non-linear fitting capability and higher prediction accuracy in color space conversion of display.