Study On Color Reproduction And Press Work Detection Based On Multispectral

With the development of multi-spectral imaging, multi-spectral color reproduction hasbeen one of the most important issues in color science. Traditional color reproduction ismainly based on the chrominance matching of target colors, therefore it is easily influencedby metamerism and is difficult to achieve the goal of ‘what you see is what you get’.Multi-spectral color reproduction is based on the spectral matching of target colors. Theirmatching results are not influenced by environment such as illuminate changes, observerchanges and so on. Therefore, multi-spectral color reproduction could overcome the defects oftraditional color reproduction effectively, and it is important to the development of highfidelity printing, presswork anti-counterfeit, cross-media color reproduction.In inspect of printing reproduction based on multispectral, this paper first constructed thetheoretical framework of multi-spectral color reproduction. In addition, it expounded thefunctions and key techniques of every module in this theoretical framework. Due to spectraldata has the characteristics of high dimension, high data redundancy and high memory cost,two common methods of spectral data dimensionality reduction method (i.e., PrincipalComponents Analysis and Independent Components Analysis) were analyzed for theiradvantage and disadvantages, then a new dimensionality reduction method based on acombination of Principal Components Analysis and Independent Components Analysis wasproposed. The first step of this method was to reconstruct multi-spectral data by PrincipalComponents Analysis. Colorimetric information of target colors was introduced into spectralreconstruction in order to improve colorimetric accuracy. The second step of this method wasto correct residual spectral error which was resulted from the introduction of colorimetricinformation by the use of Independent Components Analysis. The numerical simulationresults showed that the introduction of colorimetric information could improve thechromaticity accuracy of process data. Simultaneously, the correction of residual spectralerror improved spectral accuracy of process data.In inspect of presswork detection based on multispectral, this paper first introduced thespectral characteristics of LED light source. Because of metameric colors have differentcolors under different illuminations, they were decomposed into fundamental stimulus andresidual stimulus on the basis of spectral decomposition theory. Color vision differencefunction was calculated according to the value of residual stimulus, and the LED light sourcewhich has the same peak wavelength with color vision difference function was selected as theoptimal lighting source which could enhance the contrasts between metameric color areas ofpresswork. The numerical simulation results showed that proposed method could enhancecontrasts between metameric colors effectively. At last, aiming at the detection of metamerismink, a detection method based on measuring instrument and a detection method based onhuman vision were designed respectively. In addition, a workflow of determining optimallighting source to detect metameric color areas of presswork was proposed.