同色异谱现象是光谱反射率重建与颜色再现中的一个重要问题。 采用三基色CCD相机采集CIE标准光源D65下的颜色样品信号, 建立非线性复合模型, 使用主成分分析结合神经网络的方法（PCA-NET）改进基于同色异谱黑理论的R矩阵算法, 对标准Munsell色卡光谱重建进行研究。 在保证给定照明条件下的色度精度同时, 对光谱重建的结果进行了实验评价和讨论。 实验结果表明, 在给定的照明条件下, PCA-NET算法能够准确的拟合相机输出信号与主成分系数之间的非线性关系, 将其代替线性算法应用于R矩阵算法中时, 测试集的平均均方根值是未改进R矩阵算法的0.76, 平均标准差是R矩阵算法的0.85, 可有效提高光谱反射率的重建精度。 改进后的R矩阵算法具有精度较高、 操作简单易实现的特点, 可用于对重建色度精度及光谱精度均要求较高的领域。

Metamerism phenomenon is an important problem in spectral reflectance reconstruction and color reproduction. In this paper, a 3-primary color CCD camera is used to acquire spectral information in CIE standard illuminant D65 and a nonlinear composite model is established, including principal component analysis and neural network method (PCA-NET) to modify the Matrix R Method based on the Metameric Black theory. The standard Munsell color card is used in spectral reflectance reconstruction experiment and the results are evaluated and discussed. The experimental results verified that the PCA-NET algorithm can accurately fit the nonlinear relationship between the output signal of the camera and the principal component coefficients; and it can be used in the R matrix algorithm instead of the linear algorithm; the new method can serve as a promising technique for building a spectral image database whihc is better than the original Matrix R Method. In the fixed illumination environment, the mean RMS of the test set is 0.76 improved, and the mean STD of the test set is 0.85 improved, which can effectively improve the accuracy of spectral reflectance reconstruction. The modified matrix R method has the advantages of higher accuracy and easy implementation, and it can be used in the field of color reproduction and spectral reflectance reconstruction.