摘要
提出了一种基于 BP网络的色彩匹配方法 ,它既解决了 RGB色彩空间到 CMYK色彩空间的非线性转换关系 ,又可用于其它的色彩空间转换 .首先 ,给出样本数据的选取和转换方案 ,并结合色彩匹配的特点 ,利用迭代自组织数据分析算法实现对样本数据的动态聚类 .其次 ,在将 BP网络用于色彩匹配的过程中 ,设计了合理的网络结构 ,并对训练数据进行了规格化处理 .最后 。
Color matching method is one of the key technologies in the computer graphics and image processing fields. A color matching method based on BP network, BP _CMM, is presented in this paper and it not only can solve the nonlinear conversion form RGB color space to CMYK color space but also can be applied to complete conversion between other two different color spaces. Firstly, the selecting criterion and the transforming method of sample data are given, and clustering the sample data is achieved with the iterative self organizing data analysis technique algorithm relating to the properties of color matching. Then, in the course of applying BP network to color matching method, designing the reasonable network structure and standardizing the data are discussed. Finally, the network training and testing results are given and the experimental results show that BP\-_CMM is satisfactory.
出处
《计算机学报》
EI
CSCD
北大核心
2000年第8期819-823,共5页
Chinese Journal of Computers
基金
国家自然科学基金!( 69975 0 0 5 )
哈尔滨工业大学校基金
关键词
色彩匹配
动态聚类
BP网络
计算机图形学
color matching method, clustering, BP algorithm, color consistency