为了实现彩色信息的标准化显示,需要对CRT的色度空间进行标定,也就是CRT的R、G、B空间与CIE的标准色度空间的相互转换问题.人工神经网络近年来被广泛应用于多种颜色空间的变换过程中.在分析前人经验的基础上,提出了一种采用四隐层BP网的CRT的R、G、B与CIE的X、Y、Z色度空间的变换方法.实验结果表明,该方法的收敛性和训练时间均优于前人采用2个或3个隐层的方案,而且通过对512个训练样本的实验,其平均转换精度接近1.5个CIELUV色差单位.

CRTs(Cathode Ray Tubes)are major display devices in computers. In practical cases, in order to display colors on a CRT on standard, we must do color calibration work for it, which is the problem of color notation conversion between the RGB space of the CRT and the XYZ space of CIE system. Neural networks is one of methods in color notation conversion. In this paper, we proposes a BP Neural networks with four-hidden-layer to perform the color notation conversion from RGB space to XYZ space in a computer controlled CRT system. The experimental results show our method is better than the method which using only 2~3 hidden layers, and the color conversion precision of our experiment is about 1.5 CIELUV units.