Study On Modeling Of Color Separation Methods And Errors Analysis

Whether the color information of copy can be transferred correctly is a key to measure printing product quality. In prepress stage, it is a important procedure that the copy is be separated and color-separation information is be transferred to plate. This procedure have a influence to copy reproduced quality. It has a positive significance about research methods and technique on color-separation, and seek optimized modeling methods. Based on ECI2002 standard target recommend by FOGRA, designed the color-separation module used by printing simulation system. Used the digital analysis methods, and Systematically researched and technical analyzed four color-separation methods from the modeling data dependence, the relationship between copy and data, accuracy and the stability. Main research work and obtained conclusions are as follow:1) Based on existing color-separation methods, established coversion model from L*a*b* color space to CMY(K) color space by methods of Neugebauer equation,3-D LUT and neural network.3 methods were used by printing simulation system as 1st,2nd,3rd options.2) Established coversion model from L*a*b* color space to CMY(K) color space by multiple linear regression and nonlinear regression. The accuracy of nonlinear regression model in C, M,Y was better than linear regression model, and the results were 1.51%,2.09%,2.11%. This conclusion indicated, that conversion model had better accuracy with the method of nonlinear regression model.3) Established coversion model from L*a*b* color space to CMY(K) color space by multiple multivariate nonlinear regression. Procedure in modeling indicates, that thought process is distinct with multiple multivariate nonlinear regression, because this method not only certifies accuracy, but also model is simple and intuit. Multiple multivariate nonlinear regression is be used in color-separation module as 4th option.4) Contrasted and analyzed accuracy and stability about four color-separation model based on ECI2002 offset-print data. The research indicated, that the method of 3D-LUT had best accuracy, and the method of multiple multivariate nonlinear regression had best stability. 5) Analyzed dependence in sample quantity and corresponding superiority in area-modeling about four color-separation models. The research indicated, that the method of Neugebauer equation needed least modeling-data, and it was comfortable for copy with dominant hue red. The method of multiple multivariate nonlinear regression needed less sample quantity than the method of Neugebauer equation, and the method of BP neural network needed more sample quantity than Neugebauer equation.3D-LUT needed the most sample quantity. The method of BP neural network and the method of multiple multivariate nonlinear regression were comfortable for copy with dominant hue green and blue. The method of 3D-LUT was comfortable for copy with dominant hue blue.6) Designed and developed the color-separation module in printing simulation system based on Matlab.