目的 实现LCD显示器RGB颜色空间到颜色光谱高效的特征化。方法 利用主成分分析法对光谱数据进行降维处理以及借助RBF神经网络研究输入变量数据范围、视觉加权函数和颜色数量对特征化模型的精度影响。结果 主成分个数为6时可以很好地保留光谱原来的信息；输入变量范围为0到2.55，CIE1931视觉函数作为加权函数，颜色数量为364时特征化精度高，客观验证99个颜色转换的平均色差为0.36，最大色差为1.59，总样本的平均色差为0.17。结论 输入变量数据范围对模型影响最大，视觉加权函数和颜色数量次之，因此在特征化时要考虑输入变量范围、视觉加权函数和颜色数量，这样可以提高模型的精度。文中提出的模型是一种精度较高的特征化模型，具有一定实际应用价值。

The paper aims to realize efficient characterization of the LCD display RGB color space to color spectrum. The principal component analysis method was used to reduce the dimensionality of the spectral data and the RBF neural network was used to study the influence of the input variable data range, visual weighting function and color quantity on the accuracy of the characterization model. The original information of the spectrum could be properly preserved when the number of principal components was 6. The characterization accuracy was high when the number of principal components was 6, the input variable range was 0 to 2.55, the CIE1931 visual function was used as the weighting function, and the number of colors was 364. The average color difference of 99 color conversions was objectively verified to be 0.36, the maximum color difference was 1.59, and the average color difference of all color patches was 0.17. The input variable data range has the greatest impact on the model, and the weighting function and the number of colors are the second. Therefore, the input variable range, visual weighting function and number of colors should be considered in the characterization to improve the accuracy of the model. The model proposed is a high-precision characterization model with certain practical application value.