Application Of Support Vector Machine In Offset Printing Quality Control

A intelligent algorithm which called support vector machine(SVM) is introduced to the printing quality control by the paper, a regression mode is built to set up the relationship between the color and density, otherwise, every factor of density has been analyzed by grey relation to decide which is the biggest one, another regression model has been built to find the relationship between the CMY color space and the Lab color space.The data mainly contains two sections, the information of density and color was gained by PressSign form a large number of patches, the information for color space conversion was gained form measurement IT8.7/4.PCPO was adopted to create a multi-single support vector regression model of determining solid ink density, faced to the situation that the same color difference corresponds to a different density, the grey relation was approached to consider L, a, b and printing contrast which is the biggest factor to the density.Two different methods were adopt to establish the model of color space conversion. One was using the least square support vector regression to set up a multi-multi model, another was creating three multi-single model for each output items, then combine the regression output items. Otherwise, how the sample size influenced the regression accuracy was verified.The result showed that for determining the best solid ink density, a more comprehensive and better result can be given by the model based on support vector machine and grey relation. The training error is small, the predictive capability is strong, the overall error is beyond the standard. Support vector regression can be applied to establish space conversion model, multiple input multiple output support vector regression performance is poor, the conversion model’s training accuracy is not enough, needs to be improved.