一种基于RBF神经网络的LCD显示器光谱特征化模型  

A Spectral Characterization Model of Liquid Crystal Display Based on RBF Neural Network

作　　者：于海琦[1] 刘真[1] 田全慧[2] 

机构地区：[1]上海理工大学,上海200093 [2]上海出版印刷高等专科学校,上海200093

出　　处：《包装工程》2015年第19期130-134,共5页Packaging Engineering

基　　金：国家自然科学基金-青年基金(61301231);上海市研究生创新基金(JWCXSL1402)

摘　　要：目的研究LCD显示器的光谱特征化。方法提出一种基于RBF神经网络的显示器光谱特征化模型;扩展神经网络模型输入变量的项数,以提高特征化模型的精度。结果实验结果表明:[rg rb gb]项的引入,提高了特征化模型的光谱和色度精度,以及网络的泛化能力;引入[r2 g2 b2],[r2 g2 b2],[rg2 rb2gr2 gb2 br2 bg2]均会导致模型精度下降及泛化能力降低;以[r g b rg rb gb]作为神经网络输入变量的特征化模型,在精度和泛化能力上均是最优化的,实现了平均色差为0.14的色度精度。结论选择扩展项[rg rb gb]作为输入变量的RBF神经网络模型对LCD显示器进行光谱特征化,是一种高精度显示器特征化的最优模型。The aim of this work was to study the spectral characterization of LCD. A spectral characterization model based on RBF neural network was proposed in this paper. The prediction accuracy of model was improved by extending the input variables of neural network. Experimental results showed that introduction of [rg rb gb] item could effectively improve the characterization chromaticity and spectral precision of the model as well as the generalization ability of the network, while introduction of [r^2 g^2 b^2] , [r^2 g^2 b^2] , [rg^2 rb^2 gr^2 gb^2 br^2 bg^2] item could decrease both the characterization precision of the model and the generalization ability of the network. The characterization model with input variable of [r g b rg rb gb] terms of RBF neural network achieved the optimal precision and generalization ability, reaching the colorimetric accuracy of 0.14. Thus, RBF neural network model with input variables of [rg rb gb] was the most optimized model for spectral characterization of LCD with high resolution.

关 键 词：RBF神经网络 光谱特征化 LCD显示器 扩展项 

分 类 号：TS801.3[轻工技术与工程]