The reproduction of color as practiced in the graphic arts, professional and consumer photography has undergone a remarkable transformation in the past decade and a half. What was once very much a craft, skill, and art has become a computational process supported by the disciplines of color science, image science, image processing, signal processing, and computer science. The search for a solution to the communication of color information in distributed computer systems that received much attention dating from early efforts at standardization; e.g. Office Document Architecture (ODA) has led to a fundamental shift in the measurement basis for color information in digital color reproduction systems. The colorimetric specification of color replaced, in large measure, the device-dependent representations that were based upon sensitometry. This enabled much of the science of human vision to be brought to bear in a computational manner in the process of color reproduction. This has placed new demands on the science of color measurement and the related color spaces which have evolved under the auspices of the CIE. The computational potential of this representation has already produced significant changes in both capture and output devices.Digital cameras may now contain complex internal pipelines offering a range of processes such as chromatic adaptation computation, color space conversions, and algorithms based on the adaptive processes of the human visual system. Digital printers may have equally complex image processing pipelines. To make this complexity accessible to the user an architectural solution is being sought in the form of color management (International Color Consortium, ICC). The result has been that many of the processes executed by skilled operators have become menu choices or automatic operations that are computationally executed. Much progress has been made in turning color reproduction into a systematic process, but the question remains: How far can this be taken? Much stress has been placed upon colorimetry to provide the foundation for this approach to color reproduction. This keynote will examine what challenges lie ahead for color science and contrast them with the expectations that have emerged in the consumer and professional marketplaces. The subjective nature of color reproduction must be accommodated at the same time as taking full advantage of the objective processes that can simplify and streamline the generation of high-quality color output. What is the role of color science research in realizing this goal? This paper will also offer suggestions as to the logical placement and access for the artistic skills of the professional practitioner in photography and the graphic arts.

The parametric factors k_L, k_C and k_H that scale the CIELAB components ΔL*, ΔC* and ΔH* in the CIE94 colour difference formula are unity under reference conditions. When the conditions are changed, the scaling factors may be adapted to account for the influence of specific experimental conditions on perceived colour differences. We determined thresholds for the visibility of static background noise and for the visibility of a test symbol. The noise was present in only one of the L*, C* or H* dimensions, and the test symbol was an increment to the background, also in one of the dimensions L*, C* or H*. In order to maintain a perceptual uniform difference metric between test symbol and noisy background we arrived at k_L = 0.15, k_C = 0.52, and k_H = 2.21, such that a just noticeable difference corresponds to ΔE*₉₄=1. When the dimension (L*, C* or H*) of the incremental test symbol is the same as that of the noise in the background, the threshold for the test symbol increases linearly with the noise. When the dimensions are different, the thresholds for the test symbol remain constant (background noise in L*) or slowly increase (background noise in C* or H*).

Reproducing colour transparencies on hard copy is a common cross-media reproduction task, in which the original and reproduction have different viewing conditions. Colour appearance models CIECAM97s and CIECAM02 have not been successful at predicting the effect of the different viewing modes, partly as a result of ambiguities over the effects of surround and background.New values were derived for surround parameters c and Nc, which gave an improved prediction of the appearance of the transparency. Also evaluated was a function which weights the surround and background luminances by the distance from the stimulus in determining the background luminance factor Yb. Although this technique yielded a small improvement in the prediction of the transparency appearance, it was not significantly better than the grey world assumption which sets Yb to 20%.

The aim of this paper is to explore the space of diagonal colour constancy solutions. In gamut mapping approaches, finding the illuminant of a scene implies to find the set of feasible maps and afterwards to apply certain decision criterion to select a proper solution. This last step has been usually based on a heuristic computation over the feasible set. However an analysis on how are the solutions of this feasible set is not known by the authors. This is the essential contribution of this paper, since we explore on a reduced version of the feasible set some specific properties of the solutions. Criteria such as, maximum volume, feasible set average, maximum area on chromaticity plane or grey world solutions have been explored, and this works conclude that this usual criteria do not always assure finding optimal solutions, and therefore, further work remains to be done in this sense. Finally, we outline that some criteria related to the position of the optimal mapped image on the chromaticity plane should be taken into account.

In this work we create a computational model, which assesses chromaticity differences based on an ellipse data set. The used ellipse data sets are the MacAdam ellipses in the CIE 1931 (x,y)-chromaticity diagram and the ellipses which were used to derive the CIEDE2000 color-difference formula in the CIELAB color space. In general the ellipse data set can be any set of planar chromaticity ellipses. The chromaticity differences are calculated from the surfaces which are defined by the ellipse data set and the two chromaticity points whose the chromaticity difference is calculated. The distances along the surfaces are calculated by a method based on the Weighted Distance Transform On Curved Space (WDTOCS). The computational model corrects the planar values in chromaticity difference calculation.

Since an input device is not a colorimeter and its opto-electronic behavior is not ideal, its color gamut is smaller than that of the CIE-1931 XYZ standard observer. A chromatic discrimination model and packing algorithm to the color discrimination ellipses have been used to compute the number of distinguishable colors within the frontiers of MacAdam's optimal color loci. We have found that, due to the short dynamic range of the digital camera response, this distinguishes considerably fewer dark colors than light ones, but relatively much more colors with middle lightness (Y between 40 and 80, or L* between 69.5 and 91.7).

In this paper, a new technique to simulate apparel products is introduced. The technique includes two issues: texture mapping and color fidelity. A texture grid is generated interactively to map the pixels in the target area onto the texture space. Through this grid, user can subtly twist and stretch the texture to achieve realistic effects. As the proposed texture mapping technique is image based, the computation complication and the mapping distortion caused by three-dimensional (3D) models can be avoided. We employ the dichromatic model to achieve the color fidelity. The color signals in the target area are divided into two linear components according to the model. Then only one component is substituted by the new color specified by users while the other one is held. This substitution technique is accomplished automatically and can keep the illuminant distribution on the target area unchanged.

One example of color constancy is color transparency: when a surface is seen both in plain view and through a transparent overlay, the visual system still identifies it as a single surface. Previous studies suggest that color changes across a region of an image that can be described as translations and/or convergences in a linear trichromatic color space lead to the perception of transparency, but other transformations, such as shear and rotation, do not. Recently, other studies have added motion to their stimuli, claiming that this enhances the transparency effect.We tested whether complex configurations and motion are neutral with respect to the effects of systematic color changes. We defined several experimental conditions: a static versus moving stimulus condition, a simple (bipartite stimuli) versus a more complex configuration (checkerboard stimuli), equiluminant, filter and illumination overlay conditions. Different absolute color changes (vector lengths) were also chosen and varied systematically within the gamut of the monitor.The main results show that motion influences observers' responses for translations independently of stimulus complexity, luminance conditions, and vector lengths. A strong effect is observed for divergences that induce transparency perception in moving checkerboard conditions. However, while shears in a moving bipartite configuration tend to be transparent, this effect is completely cancelled for checkerboard like stimuli, even in motion. Finally, neither motion nor complex configuration effects have been found for convergences.

This study deals with the regulation of colour simultaneous contrast. We have proceeded to the measurement of the regulation effect of a scene in a situation where a strong induction generated by a large peripheral field is counterbalanced by a complex colour surround in the neighbouring. Results show that every neighbouring scene considerably reduces the chromatic contrast induced by the large peripheral colour field. Although every neighbouring scene has the same average chromatic content, the resulting colour appearance of the target seems to differ between scenes, and this may be ascribed to the spatio-chromatic organisation of the scene.

A uniform color space has, according to various literature, two definitions: (1) a global uniform color space is a space in which perceptional color difference agrees with the Euclidean distance; (2) a local uniform color space is a space in which discrimination elliptics/ellipsoids are unit circles/spheres everywhere. Unfortunately, it seemed that the relationship between them was not well understood and uniform color spaces have been constructed following these two different definitions independently. In this paper, we discuss the issue from a point of view of global Riemannian geometry and show that these two uniform color spaces are actually equivalent.Giving perceptive metric in a color space, an efficient algorithm is shown to construct a “pollar coordinate system” for the color space, which is the image of the pollar coordinate system in its uniform space.