| Digital image devices, such as color display, color scanner, color printer and etc., are being extensively utilized in industry, life and scientific research. Since the principles of color acquiring, display and reproduction and accordingly their color rendering abilities of these equipments are distinctly different, it is difficult to transfer the color information directly with high fidelity cross different types of media, which is seriously restricting the applications of digital image devices to the modern and traditional color industries, computer aid design (CAD), biomedicine, and many other fields. Color management system (CMS) accurately defines the influencing parameters of color appearance, such as the colorimetric characteristics of color image devices, color rendering intent, illuminating and viewing conditions, and etc., to realize actual color reproduction based on color mapping of cross-media color spaces. On the other hand, color is the synthesized result of stimulus, human vision system, illuminating and viewing conditions as illuminant, viewing environment and geometry. Therefore, the color management system is very complicated, many issues of which are still hotspots in the international research domains of color science and computer image technology.As a key technique of CMS, the color characterizations of related digital image devices were studied in this thesis. For different image media, the technical methods and their corresponding mathematical models of color characterization were compared and further optimized. Meanwhile, the color gamut description methods of digital image devices were investigated and then were used to analyze the characteristics of several equipments. Lastly, based on the accurate color characterization of related devices, an integrated application example of color transfer and reproduction was presented, in which the ANSI IT8.7/2 standard color target, as a representative image, was firstly acquired by a scanner, then was displayed on a CRT as its softcopy, and its hardcopy was reproduced by a printer.Color scanner is important digital image input equipment. There are two main types of methods for scanner color characterization, i.e. the colorimetric one and the spectral one. The polynomial regression model for scanner colorimetric characterization was emphatically discussed, considering the optimization of the polynomial items, preprocessing of input parameters, and selection of appropriate output color space. However, this polynomial regression model suffered from thecolor metameric problem. Herewith, to minimize the metamerism, a model was developed based on the principal component analysis (PCA) and back-propagation (BP) artificial neural network for scanner spectral characterization, which could reconstruct more accurate spectrum in comparison with the related international experimental results.For the extensively used cathode ray tube (CRT) display, several colorimetric characterization models were compared with respect to their color prediction accuracies. By preprocessing of black-point removal from measured data, the model accuracy was efficiently improved. When 17 characterized colors were used for each channel, the piecewise linear interpolation assuming constant chromaticity coordinates (PLCC) model was the most accurate, while the best performance belonged to the gain-offset-gamma (GOG) model when 5 colors were characterized for each channel. The color display principle of liquid crystal display (LCD) is different from CRT's, so a new S-Shape model was proposed in this thesis for LCD colorimetric characterization based on the detailed measurement and analysis of LCD's tone reproduction characteristics (TRC). Compared with the existed S-Curve model, the S-Shape model was more convenient and reasonable for achieving comparative color prediction accuracy.The colorimetric characterization of ink-jet printer was more complicated than those of scanner or display. The transformation between the color spaces was generally performed by three-dimension look-up table (3D-LUT) with interpolation. The cubic interpolation was easy to be applied to the conversion from device-dependent color space of printer to CIE colorimetric system, but its inverse conversion was rather difficult with this method. Herein an algorithm using 3D-LUT with interpolation and matching based on cubic subdivision was presented, and the experimental results proved that the inverse transformation of printer color spaces could be realized more efficiently and accurately with this novel algorithm.The exact color gamut description of digital image device is the precondition of cross-media color gamut mapping. Compared with the discrete gamut description method, the analytic representation could reduce the memory size of saved data and improve the calculating efficiency of color gamut mapping. In this thesis, enlightened by the curve surface fitting in three-dimension space and combined with device gamut characteristics, a two variable high order polynomial (TVHOP) method and its mathematical model were developed. The CRT experimental resultsshowed that the color gamut could be accurately determined by this TVHOP model, in which only a small quantity of measurements of colors on gamut surfaces was required.With the basis of thorough investigation for color characterization of digital image devices, the ANSI IT8.7/2 standard color target was employed as a representative color image to carry out a color transfer and reproduction experiment from a scanner, passing by a display, to a printer. In this experiment, the color characterization model and color reproduction fidelity were synthetically tested, analyzed and evaluated, which proved the accuracy and efficiency of these models and technology presented in this thesis by comparing with related domestic and overseas studies.Finally, the main conclusions and innovations of this dissertation were summarized, and the perspectives and the emphases of future study were also forecasted.
|
PDF Full Text Request