| Traditional color reproduction technologies cannot achieve the consistent colorreproduction under different viewing conditions. To solve this problem, the image colorappearance reproduction technologies are researched in this dissertation. New schemesare proposed in the color appearance reproduction system framework, color appearancemodels, the color appearance reproduction of colorimetric images, multispectral images,and high dynamic range(HDR) images.First, the architecture of image color reproduction system is discussed. The twoexisting structure of color management, one is based on colorimetric and another isbased on spectral, which are analyzed, and taking the limitations of demanding theidentical viewing conditions between the reproduced image and its original one intoaccount, the color appearance reproduction under different viewing conditions ispresented. According to the two existing color management, two implementation colorappearance reproduction systems are proposed.Next, color appearance models are researched. By considering a relatively simplecolor management system, to demonstrate that traditional tristimulus match would havedifferent color appearance in disparate conditions. Next some terminologies of colorappearance are examined. This includes appearance attributes, viewing conditionattributes, and color appearance phenomena. Then methods of predicting colorappearance of color stimulus are researched. This includes chromatic adaptation modelsand color appearance models. Finally some color appearance models are comparedthrough a experiment, and CIECAM02is chosen for this dissertation.Then aiming at the issue of the color appearance reproduction of colorimetricimages, two solutions are proposed and implemented. To solve the problem that huenonconstancy of CIELAB color space affects the result of gamut mapping, ahue-constancy corrected method is proposed and a new color space tLAB is constructed,which was used to replace CIELAB during color management. Experiments show thatthe results of gamut mapping in the new tLAB space gain an advantage over thetraditional ones. Aiming at the problem that the traditional color reproduction cannotconsider disparate conditions, CIECAM02is used to supply the necessary adjustmentswhen the viewing conditions are different. By optimizing CIECAM02through fourcolor-difference datasets which includes RIT-DuPont, Witt1999, Leeds, and BFD-P, anew uniform color appearance space is constructed, which is used to replace CIELAB during color management. Experiments show that color management based on theoptimized CIECAM02can solve consistent color reproduction under disaparate viewingconditions, the results of gamut mapping gain an advantage over the traditional ones,and the new color difference evaluation can provide a quantitative measure that moreclosely corresponds to the color difference perceived by human visual system. Lastly, acolor appearance management system is constructed according to the studies above. Itutilizes the CMM+Profile mode and is compatible with the traditional colorimetric colormanagement system. The CMM supports the color appearance processing, and theProfile records the relationship between device dependent color spaces and deviceindependent color spaces.Aiming at the issue of the color appearance reproduction of multi-spectral images,a new algorithm based on color appearance mapping for multi-spectral imagereproduction is presented. Firstly, through introducing color appearance transformationand inverse transformation, color appearance matching is achieved between thereprodueced image and its original one. Secondly, to improve spectral precision of thereproduced image, based on the source spectra, the estimated spectra was corrected.Lastly, to reference the success experience of ICC color management, the CMM+Profilemode is also utilized for multispectral image color appearance reproduction.Finally, to solve the problem that different viewing conditions results in differentcolor appearance between a reproduced image and its original one during HDR imagereproduction, a new algorithm based on color appearance mapping for HDR imagereproduction is presented. It separates procedures of color appearance mapping and tonemapping in the structure, the former maintaines a perceptual match between the realworld and the displayed image, and the latter compresses the dynamic range. Thismethod can support mix-and-match color appearance models with tone reproductionoperators to suit any specific task. In color appearance mapping, considering theoriginal viewing condition is usually unknown, the estimation is supplied to improve theapplicability of the algorithm. In tone mapping, a tone reproduction operator based onadaptive regionalization is presented, through allocating the range of display luminancefor different regions, which raised perceptual contrast of the image. Experiments showthat the proposed algorithm gains advantages over the traditional ones in colorappearance maintaining, dynamic range compressing, and the performance of details. |
PDF Full Text Request