| 摘要: |
| 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。 |
| 关键词: 色彩空间 色块 RBF神经网络 |
| DOI： |
| 分类号:TS801.3 |
| 基金项目:河南省教育厅自然科学研究项目 （13B510932）；河南工程学院数字印刷和纺织品印花色彩控制研究中心资助项目（YJJJ2013003） |
|
| Application of RBF Neural Network in Color Space Conversion of Display |
|
HONG Liang1, LI Yu2, CHU Gao-li1, WANG Na1
|
|
1.Henan Institute of Engineering， Zhengzhou 450007， China;2.Henan Provincial Press and Publication School， Zhengzhou 450044， China
|
| Abstract: |
| 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. |
| Key words: color space color patch RBF neural network |
