Research On Key Issues For Spectral-based Inkjet Color Reproduction

Spectral-based color reproduction can reduce the metamerism problem in the tranditional colorimetric color reproduction workflow and fufil the devices-independent, scenes-independent and obsevers independent color reprodution. so it can achieve the real and object reproduction of the original color. This thesis focusd on fulfilling the spectral color reproduction using the ink-jet printing technology. After investigating several key issues of the ink-jet printing workflow, such as sub-printer devision, ink limitation algorithm development, spectral prediction, spectral gamut mapping, spectral separation and modeling metric optimization, a new inkjet based spectral color reproduction workflow was proposed. Several typical color samples were used as example to testify the accuracy of the proposed method and the result appears to be quite good.The thesis analyzed the inherent problems of tranditional colorimetric color reproduction according to the fundament theory of color science, and discussed the theoretical advantage and technical architecture of spectral color reproduction comprehensively. After that, the current research status, together with the main topics and key technical problems of ink-jet spectral color reproduction was introduced.The accuracy and stability of the printer and color measurement devices was quantified by proceeding relevant experiments, and a new sub-printer deviding method was proposed based on which one can achieve the gamut maximizaion and simply the printer modeling workflow.A multi-dimensional interpolation based ink limitation algorthm was proposed. The threhold of ink amount was detemined using a calculation and micorexamination combined method. By adjusting the parameter in the algorithm, the proposed method can avoid the ink spilling problem and realized the gamut maximization. After that, a fast ink limitation method was proposed in order to improve the modeling efficiency.This paper analyzed the state of art halftone spectral models and its typical optimazaion method comprehensively according to experment results. Based on that, a new model, which used the BP neural network to simulate the best n value and outperformed the classical CYNSN (Celluar Yule-Nielsen Spectral Neugebauer) model was proposed.The performance of five typical optmizaion algorithms for spectral separation was investigated. Based on that, a gamut-judging based spectral color separtion method was deveopped which achieved high accuracy in-gamut color reproduction. From the prospective of gamut judging, the global printer modeling method was developped, as well as a spectral gamut mapping algorithm. To testify its performance, typical pigments colors of dunhuang cultural heritage, standard colorchecker testchart together with print samples were used as origianl for color reproduction. Considering the human visual charactristic, a spectral and colorimetric combined metric controlled spectral separation method was proposed, which improved the colorimetric accuracy obviously at the cost of slight reduction of spectral accuracy. At last, the advantage of spectral color reproduction in the aspect of metamerism reduction was discussed according to the previous research.