Computer based color imaging is rapidly moving from the realm of the specialist to the general public. To enable users the greatest benefit from advances in color image reproduction technology, the computer systems have to modify to incorporate this new technology. Two possible approaches are described - by the adoption of color management system software, and by the adoption of device independent color variables. Although both systems are likely to perform equally well when properly configured, the later is likely to be the long term choice as it will yield the most robust, lower cost and transparent system.

Previous and recent revisions of the ATD model for color perception and visual adaption are incorporated into the version that is fully described in this paper.

The ATD color space developed by Guth has shown that is can predict the response of human vision to a wide range of stimulus conditions. ATD is a good candidate for predicting the proper transforms under which the output of differing hard copy platforms can be made to have the same appearance. This paper will show how Guth's ATD model can be modified to produce appearance matches over the entire gamut of various ink sets. The paper will also discuss the many dimensions of gamut limits and gamut mapping.

The color matching ellipsoid theory has been extensively used to improve the relationship between the human perception and numerical representation of color differences evaluation. This theory can be described thanks to logarithmic functions which approximate the orientation, the position and the size of ellipses. By carrying out these observations to the study of the a^*b^* chromaticity plane we have obtained a partition of the plane. this partition separates each area for which the distribution of ellipses can be locally approximated by the same logarithmic function. Then we can interpolate the orientation and the dimension of ellipses for each position of each area. This logarithmic function points out that the perception of colors is nonuniform at once 'locally' for each area considered separately and 'globally' for the set of areas. Inversely, we can claim that if we take the 'inverse function' of this logarithmic function we can transform the distribution of ellipses in a convenient form in order to obtain an uniform distribution of circles. This process can be applied to each area of the chromatic plane and can be scaled to compute a distribution composed of equal-sized circles. Then, this transformation allows to set a new color space which is more uniform than the L^*a^*b^* color space; we will call it the L degree(s)a degree(s)b degree(s) color space.

It is known that colour constancy is more robust when the changes in illumination are restricted to the phases of daylight than it is for changes in most arti ficial 1 ight sources, suggesting that the ability to correctly identify surface colours may depend in part, on the observers familiarity with the light source. The present study examines whether a relatively short period of practice in judging surface colours under different illuminants leads to an improvement in colour constancy, and if so whether the improvement is restricted to the illuminants under which the judgeinents were made. Five subjects were asked to identify by name (chosen from 11 basic colour terms) a total of 70 colour patches under 4 different broad band illuininants. Subjects then named each colour patch a further 20 times distributed over a 2 week period, 10 times under each of two of the broad band illuminants. Finally, subjects again named the patches under all 4 illuminants. It was found that the degree of categorical colour constancy under broad band illumination was significantly better after the learning trials, and that the improvement was greater for the illuminants used in the learning trials.

Many color calibration and enhancement strategies exist for digital systems. Typically, these approaches are optimized to work well with one class of images, but may produce unsatisfactory results for other types of images. For example, a colorimetric strategy may work well when printing photographic scenes, but may give inferior results for business graphic images because of device color gamut limitations. On the other hand, a color enhancement strategy that works well for business graphics images may distort the color reproduction of skintones and other important photographic colors. This paper describes a method for specifying different color mapping strategies in various regions of color space, while providing a mechanism for smooth transitions between the different regions. The method involves a two step process: (1) constraints are applied so some subset of the points in the input color space explicitly specifying the color mapping function; (2) the color mapping for the remainder of the color values is then determined using an interpolation algorithm that preserves continuity and smoothness. The interpolation algorithm that was developed is based on a computer graphics morphing technique. This method was used to develop the UltraColor gamut mapping strategy, which combines a colorimetric mapping for colors with low saturation levels, with a color enhancement technique for colors with high saturation levels. The result is a single color transformation that produces superior quality for all classes of imagery. UltraColor has been incorporated in several models of Kodak printers including the Kodak ColorEase PS and the Kodak XLS 8600 PS thermal dye sublimation printers.

Accurate scanning of color image, which is absolutely essential for good color reproduction, depends upon the creation of a set of filters matched to the optical system of the scanner. Previous work developed a measure of goodness for a set of color filters. The measure provides an optimization criterion which can be maximized to obtain optimal scanning filters of arbitrary spectral shapes. This paper reports on the results of fabricating and testing filters designed by this method. The results show that the filters can be fabricated accurately enough to produce increased accuracy in color measurement over existing filters.

One of the major challenges in the prepress environment consists of controlling the electronic color reproduction process such that a perfect match of any original can be realized. Whether this goal can be reached depends on many factors such as the dynamic range of the input device (scanner, camera), the color gamut of the output device (dye sublimation printer, ink-jet printer, offset), the color management software etc. The characterization of the color behavior of the peripheral devices is therefore very important. Photographs and positive transparents reflect the original scene pretty well; for negative originals, however, there is no obvious link to either the original scene or a particular print of the negative under consideration. In this paper, we establish a method to scan negatives and to convert the scanned data to a calibrated RGB space, which is known colorimetrically. This method is based on the reconstruction of the original exposure conditions (i.e., original scene) which generated the negative. Since the characteristics of negative film are quite diverse, a special calibration is required for each combination of scanner and film type.

In color reproduction research, a linear model designed to minimize the error between original surface reflectance spectra and reproduced spectra is useful in the process of producing an accurate color match between the original image and reproduction under a variety of illuminants, but it is inappropriate in efficiency. We propose an efficient linear model based on surface reflectance spectra and a unified wavelength function of CIE 1931 standard observer representing human perceptual property. The surface spectra weighted with the unified wavelength function were introduced to minimize the human perceptual error between original reflectance spectra and reproduced spectra and to reduce the number of the spectral basis functions. The performance of reflectance spectra-to-CIELAB transformation on our proposed linear model is tested and compared with a conventional model based on reflectance spectra under a variety of illuminants. The results of our linear model is superior to that of the conventional model. With Munsell 400 color patches, D65 illuminant and 4-dimensional linear model, the mean color difference of our model is 1.28 CIELAB unit. And an algorithm for color scanner characterization using our model is made and tested, and the results are shown.

This paper introduced a novel system concept for high color/spatial resolution movie film electronic capture. The camera system scans each frame of movie film to the full spatial/spectral resolution of the film. The image detector output is digitized to a unique color space that involves the dye density of each of the film's three color layers (discussed in a different paper at this conference). This paper briefly discusses the system concept. A side topic of this paper is to present the camera color calibration methodology that ensures accurate color capture and storage. From the spectrophotometer output of the scanned image, the XYZ values of each color sample can be calculated. The traditional method applies a global transformation matrix to translate the scanned RGB values into the XYZ space. In its simplest form the transformation matrix is a 3 X 3 matrix translating RGB into XYZ. Other approaches have considered various linear combinations of RGB (e.g. R^*G, R²) with some success. Our approach is to consider local regions of RGB color space and determine the transformation matrix based only on those local RGB and measured XYZ values.

Several 3D interpolation techniques for the color space transformation are compared via software simulations. Comparisons are made among four geometric interpolations, trilinear, prism, pyramid, and tetrahedral, with respect to the look-up table (LUT) size and packing. Three different LUT sizes and two ways of packing, uniform and nonuniform, are applied to the forward and inverse transformations of the XeroxRGB and CIELAB. Each simulation is tested by a set of 3072 points that are sampled around the entire RGB color space. Results indicate: (1) Interpolation errors of various 3D geometric interpolations are about the same and the errors with respect to the true values decrease as the LUT size increases. (2) The interpolation error peaks at the center of the cell and diminishes at nodes (lattice points). (3) The highest error occurs at the darkest region. For equally spaced LUTs, the error drops quickly as the level increases. (4) Nonuniform LUTs have a much lower fundamental error peak but the errors are rippled to the higher levels; this gives a more even error distribution and a better average value. From this study, it is conceivable that the colorimetric reproduction can be achieved to a very high degree of precision. With proper packing, the 3D interpolation provides the capability to closely approximate the true values in all regions of the color space. From the considerations of the implementation cost and computation speed, the tetrahedral interpolation is particularly attractive.

There exist several attempts to develop a system for device-independent color reproduction in desktop publishing. However, because conventional color management system is calibrated only by the colorimetric values under the standard illuminant, perceptual difference between the original and the reproduced colors will occur when these colors are viewed under non-standard illuminant. In this study, we present a method for device-and-illuminant independent color reproduction by adopting the spectral reflectance as an intermediate color representation. We trained three layered neural networks to realize the transformation from CMY values of the proof printer to the principal components (PCs) of the spectral reflectance, and the transformation from these PCs to C'M'Y' values of dye-sublimation printer. after the learning process, we evaluated the accuracy of the transformation from CMY to C'M'Y' through PCs by the trained neural networks.

To improve the color fidelity of 4 color reproduction and to increase the flexibility of Gray Component Replacement (GCR) for the text and continuous images, a novel GCR algorithm based on CIE L^*a^*b^* signals is proposed. The algorithm consist of (1) maximum (achromatic) black determination part, (2) black adjustment part based on chroma, and (3) 3 color determination part. On this configuration, black signal is determined ahead of MCY signals, and the freedom of 3 input i.e L^*a^*b^* 4 output i.e. CMYBk conversion is concentrated in (2). By using xerographic color printer, by neural network technique for resolving this, the algorithm is examined. As a result, it is shown that the algorithm can conserve the color fidelity in any GCR rate, and which is applicable on both of text and continuous images.

The conventional method of performing spatial filtering of RGB images is to subject each plane to the same processing, usually convolution with a filter kernel. Filtering is commonly used in the processing of photographic or photo-realistic images to sharpen or blur images, and to produce aesthetically-pleasing effects. For image sharpening, the technique of subjecting each plane to the same processing produces objectionable color errors in some circumstances, and that techniques which convert the image to a color space that separates luminance from chrominance and performing the filtering only on the luminance component can produce better results. The problem with this approach has been the computational cost of making the transformation, first to the luminance- chrominance space, and back to RGB. This paper presents an algorithm which operates on an RGB image and provides results which are free from chromaticity changes. It achieves these results with fewer computations than filtering the luminance component in a luminance-chrominance color space. In fact, the computations required are usually simpler than processing each RGB plane.

The paper presents a dynamically visualization procedure for 3D histogram of color images. The procedure runs for RGB, YMC, HSV, HSL device dependent color spaces and for Lab, Luv device independent color spaces and it is easily extendable to other color spaces if the analytical form of color transformations is available. Each histogram value is represented in the color space as a colored ball, in a position corresponding to the place of color in the color space. The paper presents the procedures for nonlinear ball normalization, ordering of drawing, space edges drawing, translation, scaling and rotation of the histogram. The 3D histogram visualization procedure can be used in different applications described in the second part of the paper. It enables to get a clear representation of the range of colors of one image, to derive and compare the efficiency of different clusterization procedures for color classification, to display comparatively the gamut of different color devices, to select the color space for an optimal mapping procedure of the outside gamut colors for minimizing the hue error, to detect bad-alignment in RGB planes for a sequential process.

The Color Hardcopy Quality Factors (CHQF) study series began in 1990 and has grown to include a 'database' of hundreds of print samples spanning nearly a hundred different printing devices across a broad range of marking technologies. The key element of this series is the ability it offers users to analyze actual print samples using their own criteria, rather than relying on technical specifications and general benchmarking. The understanding that print quality analysis contains an inherent subjectivity is crucial to the design of the PostScript files used for CHQF testing. The samples which were generated provide a baseline to compare printing methods. This baseline is established through uniform, controlled input, custom test programs, the use of the device-independent PostScript language,and the wide range of devices included in testing. While PostScript was used for most testing for maximum device-independence, printing solutions which could not process PostScript were tested using Adobe Acrobat as a PostScript interpreter.

Dainippon Screen's Color Management System (SCMS) is currently under development for use with device-dependent CMYK-based equipment, and RGB- based devices. SCMS will provide color space transformation of linkage to the CIE 1976 L^*a^*b^* color space. The SCMS concept is based on Screen's CMYK image data formats. The SCMS was announced two years ago at IGAS '93 where many attendees viewed this new Screen approach to color management technology. At DRUPA '95, the SCMS concept for handling RGB-based devices and the transformation of color data between CMYK and RGB-based devices, will be revealed and explained, and this disclosure will be in conjunction with the introduction of products using this CMS technology. Before the full introduction of SCMS at DRUPA '95, this paper reveals the technical characteristics of the color management system as follows: (1) As precise as possible color data acquisition and measurement are carried out. (2) As precise as possible and as flexible as possible device modeling is carried out. (3) As precise a transformation as possible, which is as quick to calculate as possible, is carried out. (4) A different algorithm is used in the case of color matching on the colorimetric measurement, especially on display of creative work by designers.

This paper describes the results of an effort implement device- independent color on three types of devices. The goal was to produce near the same eye-brain response when the observer viewed the image produced by each device under the correct lighting conditions. The procedure used to calibrate and obtain each device profile is described. A commercial color management software system (CMS) was used to transform the input image data to device dependent data.

A hardware color transformation accelerator, in CMOS ASIC form, and its usage are described. The CHAMELEON ASIC performs multi-dimensional tetrahedral interpolations at a speed of 10 Million (32 bit) pixels per second. The achieved accuracy of the color transformation is a major topic, which strongly depends upon the kind of interpolation used. A report is made on the application of this ASIC in different fields of the graphics industry and the achieved results are compared to the desired market needs.

The purpose of this research is to study the practical differences of characterizing color output devices using tristimulus values from an X- Rite 938 spectrodensitometer and an X-Rite DTP51 auto-scanning colorimeter. Because colorimeters offer a lower price point, and in the case of the DTP51, automated measurements, they may be preferable for use in a localized color management solutions. This paper examines the performance of the DTP51 colorimeter compared to a 16-point hand-held spectrophotometer. The enabling technology used was Color Blind color management software (Color Solutions, Encinitas, CA) for the Macintosh platform. It was used to create the output device profiles, and apply them to the various media tested.

24-bit color display capability is gradually becoming the standard over 256K or 64K color displays on graphics workstations. Some main problems associated with this trend are the dramatic increase in memory storage requirement and the bandwidth requirement. This paper describes the operation of a new 4:1:1 True Color Difference Mode (TCDM) color image encoding scheme which encodes information in differences. This method can display near true color quality images while the memory requirement is reduced by two-third (i.e. 4 bytes for a set of 4 pixels). The exceptionally simple hardware required to implement the real-time decoding of this scheme is also presented. Nonlinear quantization and the use of differential encoding scheme enables the encoded data to emulate the original data. Visually, it is not possible to detect any difference between test pictures encoded with 24 bits/pixel and that encoded with TCDM using only 4 bytes to represent a set of 4 pixels.