基于Legendre小波的扫描仪色空间转换研究  

Research on Scanner Color Space Conversion Based on Legendre Wavelet

作　　者：黎新伍[1] 

机构地区：[1]江西财经大学电子商务系,南昌330013

出　　处：《信号处理》2008年第5期872-875,共4页Journal of Signal Processing

基　　金：江西省教育厅重点资助项目(2007259);中国航空科学基金(02153071)资助

摘　　要：扫描仪色彩管理是图像复制过程中保证图像色彩一致性的关键技术。在分析扫描仪呈色原理的基础上,提出了一种扫描仪色彩管理新模型。首先模型采用标准色靶作为实验样本,并以彩色梯尺区色块代表整个色空间以解决实验色块选取难题。其次,利用Legendre小波和BP神经网络构造了扫描仪色彩管理Legendre神经网络模型。该模型结合扫描仪色彩管理的实际应用,提出了若干加快模型收敛的方法。最后给出的实验结果表明,与BP神经网络比较,本文模型能够提高19%色彩转换精度和62%的处理速度。Color management for scanner is one of the key techniques in the color image reproduction. A New scanner color management model is presented based on analyzing scanner color rendering principle. First, it takes standard color target for experimental sample and substitutes color blocks in color shade district for complete color space to solve the difficulties of experimental color blocks selecting. Second, Legendre neural network model for scanner color management is selected when considering the advantages of Legendre wavelet and BP neural network and the model designs some measures to speed up the convergence of the model. Finally the experimental results show that the model can improve 19 percent in color conversion accuracy and 62 percent in processing speed compared with ordinary BP network model.

关 键 词：色空间转换 LEGENDRE小波 BP神经网络 标准色靶 

分 类 号：TP334.22[自动化与计算机技术—计算机系统结构] O174[自动化与计算机技术—计算机科学与技术]