基于同色异谱黑理论的非线性复合模型光谱重建研究

发布时间：2018-03-01 16:01

本文关键词： 光谱重建颜色再现 PCA 神经网络改进的R矩阵算法　出处：《光谱学与光谱分析》2017年03期 　论文类型：期刊论文

【摘要】：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。
[Abstract]:Metamerism phenomenon is an important problem of spectral reflectance reconstruction and color reproduction in color. The sample signal three color CCD camera CIE standard light source D65, a nonlinear composite model, combined with the method of neural network using principal component analysis (PCA-NET) improved R matrix algorithm of metamerism theory based on black study on the standard of Munsell card, to ensure accuracy of spectral reconstruction. At the same time given color lighting conditions, the spectral reconstruction results are evaluated and discussed. The experimental results show that the lighting in the given condition, the PCA-NET algorithm can be a non-linear relationship between the coefficient of fitting the camera output signal accurately and its principal components. Instead of a linear algorithm is applied to the R matrix algorithm, the test set the average RMS value of 0.76 is not improved R matrix algorithm, the average standard deviation is 0.85 R matrix algorithm, can have The reconstruction accuracy of spectral reflectance is improved effectively. The improved R matrix algorithm has the characteristics of high accuracy, simple operation and easy implementation, and it can be applied to the areas with higher requirement of reconstruction color accuracy and spectral accuracy.

【作者单位】：北京理工大学光电学院光电成像技术与系统教育部重点实验室;北京理工大学深圳研究院;
【基金】：国家自然科学基金项目(61575020) 广东省科技计划项目(2015A020214004)资助
【分类号】：TP391.41;O433;TP183

【相似文献】