基于CMAC辨识理论的设备彩色特征化方法的研究  

Studying a Color Space Characterization Method Using CMAC Identification Theory

作　　者：智川[1,2] 周世生[1] 石毅 

机构地区：[1]西安理工大学,西安710048 [2]陕西科技大学,西安710021 [3]陕西华拓科技有限责任公司,西安710043

出　　处：《机械科学与技术》2012年第5期740-743,共4页Mechanical Science and Technology for Aerospace Engineering

基　　金：陕西省"13115"科技创新工程基金项目(2009ZDGC-06)资助

摘　　要：彩色设备的特征化是指设备相关颜色空间与标准色空间之间的转换,它是色彩管理技术的关键技术之一。以RGB颜色空间与CIE L*a*b*颜色空间转换为例,采用小脑模型神经网络(cer-ebellar model articulation controller,CMAC)进行设备颜色空间的特征化,以BP神经网络的方法和颜色空间转换模糊神经辨识方法对该模型进行比较研究,并利用Matlab软件进行了仿真。仿真结果显示,当训练样本总数为729个时,经过不到100次的迭代,网络既可以满足输出精度要求,并且检验样本的平均色差为1.35,表明该算法具有更快的模型转换速度和更高的转换精度。The color characterization of a device refers to the transformation between its related color space and its standard color space.It is one of the key technologies of color management.Taking the transformation between RGB color space and CIE L＊a＊b＊ color space as an example,we adopt the cerebellar model articulation controller（CMAC） identification theory to characterize the color space of a device,and then compare our color space characterization method with the BP neural network method and the fuzzy neural network identification method.We simulate the color space characterization with Matlab.The simulation results show that our new method can meet the output precision requirement by less than 100 times of iteration when the number of training samples is 729,and that the average color error of a testing sample is 1.35,indicating that the transformation speed and accuracy of our new method is superior.

关 键 词：小脑模型神经网络 设备彩色特征化 色差 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TP391[自动化与计算机技术—计算机科学与技术]