Study On The Color Output Character Based On Optimization Theory

The application of digital technology in media has effected the transmission of mediacontent deeply. As a important type of media， the true extent of digital replication is closelyrelated to the effect of color reproduction. As a basic factor of image， color plays animportant role in imgae collection， storage， processing and copying. However， due todifferences in color equipment principle， structure， manufacturing processes and drivers，etc.， resulting in the same color digital image rendering on different devices in differentcolors. The color management system (CMS) that is based on this need， through the colorcorrection， the color characterization， and color conversion， to complete the accuratereproduction of color information. With the development and optimization theory and thecomputing power and storage capacity of the computer， computer and optimization theory isapplied to build device characterization model and color conversion model in colormanagement gradually. On the basis of emerging artificial neural networks recently， geneticalgorithm and particle swarm optimization algorithm theory， in the article， the device colorpoint set gamut is built based on the pixels of a digital image， application of optimizationtheory in device characterization model and device gamut mapping model is discussed andresearched deeply， several kinds of device forward and reverse characterization model andgamut mapping model are built and evaluated. The main work and innovations of the paper isthe following:(1)On the basis of digital image pixels， the display device and the printing pointsetgamut is built by integrating the forward device characterization and the pointset gamut andthe geometry gamut is compared by the experimentn. Experiment shows that, under the basisof great progress of computer calculation and storage ability, the point set gamut method candescribe the deivce gamut effectively.(2)A farword device characterization model is built based on BP neural networkequipment. Based on classification of the study sample on hue angle， a device reversecharacterizaition model is proposede based on BP neural network， which solve the problemthat BP neural network is difficult to applied in building device reverse characterization modelpreliminary.(3)The RBF neural network is introduced to build farword characterization model evice. Take the common parameters SPREAD of RBF network as the starting point， a method ofhow the optimal SPERAD value is determined by a computer program within the experiencescope automatically. Three kinds of devices forward characterization model based on the RBFfunction are built and compared with that based on BP neural network. Experiment resultshows that, compared with device forward characterization model based on BP neuralnetwork, the effection and precisise of device forward characterization model based on RBFnetwork improved greatly.(4)Based on the point set gamut and ultrafine GBD gamut partition， two gamutmapping method based on ultrafine GBD is proposed， which are nominated as UFGBD1andUFGBD2respectively. Based on BP neural network， genetic algorithm and particle swarmoptimization， the gamut mapping method based on BP neural network(BPNNGM)， thegamut mapping method based on quasi-genetic algorithm(QGAGM) and the gamut mappingmethod based on quasi-particle swarm (QPSOGM) are proposed and compared withCARISMA. Experiment result shows that, compared with mature gamut matching algrithm,five gamut matching algrithm proposed in this paper reach the same performance nearly.(5)For the reason that color management and printing processes are closely related， aprinting ink quantity estimation model based on image block processing is proposed， and itsexperiment is designed on typical products offset and gravuer in the actual productioncondition. Experiment result shows that, the ink estimation model reaches rather high precise,it is could be used to estimate ink in factal production with its guiding meaning.Finally， on the basis of summing the main content and achievements, and analyzing andprobing the researching defects, some suggestion and idea in the furure is prposed.