Dissertation > Excellent graduate degree dissertation topics show

Study of Spectral-based Neugebauer Characterization Model for a Multi-ink Printer

Author: WangBinYu
Tutor: XuHaiSong；M.Ronnier Luo
School: Zhejiang University
Course: Measuring Technology and Instruments
Keywords: colour reproduction spectral reconstruction multi-ink printer characterisation cellular Neugebauer model Yule-Nielsen effect colour separation criteria printer calibration
CLC: TP334.8
Type: Master's thesis
Year: 2012
Downloads: 56
Quote: 0
Read: Download Dissertation

Abstract

With the development of spectral imaging technology, spectral-based printer characterisation has become a hot research focus in the related filed. Reconstructing spectral reflectance of a colour target for reproduction offers many advantages, such as high accuracy and eliminating the metamerism. Compared to traditional 4-ink printers, multi-ink printers have much larger colour gamut and more candidate recipes, which provide a great platform to accurately reproduce spectral images. Hence, with the adoption of the cellular Yule-Nielsen spectral Neugebauer (CYNSN) model, the study of multi-ink printer charaterisation models is of great significance and has practical values.First of all, the physical characteristics of the 12-ink colour printer HP Designjet Z3200ps were investigated in terms of ink drying time, stability, uniformity, tone reproduction curve, colour gamut and ink limit. Next, the training data were designed and generated to build five 3-ink and six 4-ink forward CYNSN models. Meanwhile, the model parameters were determined by optimizing the dot gain model and Yule-Nielsen n-factor. The average color differencesΔE00(D65,10°) of the 11 forward models were less than, or near to 1. Subsequently, the linear regression iteration and the cell-searching algorithm proposed by J. Guo were employed to build 4 types of backward CYNSN models, each of which had good reversibility, and the mean color differencesΔE00(D65,10°) were approximately 0.2 for most models. Taking one type of backward model for example, the predicted recipes were printed to evaluate the actual accuracy of backward CYNSN models, and the reproduction performance was good (mean metamerism indice<0.7) but with some systemic error. In addition, the spectral-based model was verified to be more accurate with comparison to the CIE1931 XYZ tristimulus based model in terms of prediction accuracy and metamerism index.Colour separation criterion is a method to select the optimal recipe from several candidates. Seven colour separation criteria, which were based on spectral match quality metrics, were compared and analysed. The experimental results showed the prediction and actual performance of the 7 criteria were approximately the same for printing samples, which indicated that the simplest colour separation criterion. i.e. minimumΔE00(D65,10°), was the best option. Regarding to the Munsell testing samples, the avgEoo&MI, which combined one color difference with two metamerism indices, outperformed other selection criteria. Except a small part of printing samples with low lightness and yellowish Munsell samples with relatively high chroma and lightness, the avgEoo&MI worked well for most printing and Munsell testing samples with the average color differences and metamersim indice about equal to 2.Two calibration algorithms, based on wavelength calibration and principal component analysis (PCA) respectively, were proposed in this study. In order to achieve high prediction accuracy in new printing environment using original CYNSN models, both methods required generating a small set of calibration samples with new ink cartridges to model the relationship between the colour characteristics of original and new inks. The PCA-based PPM (print&predict model) had the best calibration performance for original forward CYNSN models. The color differencesΔE00(D65,10°) were decreased 0.11～1.84 for different models with the mean value 0.62 compared to wavelength calibration, and even outperformed the original models using original inks in some cases, since not only the colour change between original and new inks but also the systemic error of CYNSN models were taken into account in the PPM. As for the backward CYNSN models, two algorithms reached desired results. The wavelength calibration algorithm worked relatively better, while the performance of PCA-based method was poor in some human vision sensitive spectrum for some testing samples.Finally, the main work and conlusions of this dissertation are summarized, and the perspective of the future study is also forecasted.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Electronic digital computer (not a continuous role in computer ) > External device > Printing devices
© 2012 www.DissertationTopic.Net Mobile