Theory And Technology Study On Flexographic Spot-color Ink Matching

At present,the majority of small printing presses use artificial color matching as their ink color matching method,while artificial color matching method has great dependency on colourist and the cycle of color matching is very long,what's more it's very easy to cause ink waste.Currently the research on computer color matching is mainly based on KM theory and Neugebauer equation.Color matching algorithm based on Neugebauer equation is relatively poor on color matching accuracy.Color matching algorithm based on the KM theory is widely used in some color matching software,although its color accuracy is relatively high,color matching process contains a large number of matrix operations and color matching process need to measure the spectral reflectance of each ink and substrate,usually the price of the instrument is very high,which undoubtedly lead to an increase in printing costs for some small printing press.So it is very important to propose a new ink color matching algorithm.In this paper,firstly make a detailed introduction to these commonly used color matching algorithms and do experiments to verify their effects.Then bionic algorithm is applied to color matching according to the existing algorithm's some shortcomings and insufficiency.This article selected the most widely used BP neural network and it is applied to color matching.Due to the initialization of the weights and the thresholds is very important for network performance,so use genetic algorithm and partical swarm optimization algorithm to optimize the weights and thresholds of BP neural network in the early stage of training.It is no longer use random weights and thresholds and the color matching result is improved.The results of this paper are as follows:(1)First the experiments are done to verify the color matching algorithms based on KM theory and the Neugebauer equation.The new double constant theory is used to verify the color matching effect.Experimental results show that the color matching algorithm based on new double constant can meet the requirements of color accuracy.But the calculation process involving a large number of matrix operations.It can meet the requirements of color accuracy after several iterations.The process is still complex and color matching process need to measure the spectral reflectance of each ink and substrate,usually the price of the instrument is very high,which undoubtedly lead to an increase in printing costs for some small printing press.There are many modified methods based on Neugebauer equation.In this paper,we do experiments to verify color matching effects of Neugebauer equation color matching algorithms based on first order liner regression and the second order curve regression.Experimental results show that the matching results of modified Neugebauer color matching method based on first-order linear regression and the second order curve regression can't meet the requirements of color accuracy,the Neugebauer equation color matching method based on exponentially modified can meet the requirements of color accuracy,but it need to partition according to hue,that is to say establish different Neugebauer equations according to different hues.It will make computer matching system be complicated.(2)In this paper,five improved BP neural networks are used for ink color matching,the results of these five methods are different,but they can meet the requirements of color accuracy.Among them,the BFGS quasi Newton algorithm is the best.In the initial training,the initial weights and thresholds of BP neural network are determined randomly,and the selection of initial weights and thresholds is very important to the performance of the network.(3)According to the above reasons,the genetic algorithm and partical swarm optimization are used to optimize the BP neural network.Using genetic algorithm and partical swarm optimization to optimize the neural network,in essence,is to optimize the weights and thresholds of the initial training period.The experimental results show that the matching result of optimized BP neural network is better than color matching algorithm using only BP neural network.But the genetic algorithm optimization has a fatal shortcoming,that is,the training time is relatively long,which is not allowed for some highly efficient enterprises.Particle swarm optimization algorithm overcomes this shortcoming,and the accuracy and stability of the algorithm can meet the needs of the actual production.