Showing papers presented at "Color Imaging Conference in 2003"

A Large Image Database for Color Constancy Research

Abstract: We present a study on various statistics relevant to research on color constancy. Many of these analyses could not have been done before simply because a large database for color constancy was not available. Our image database consists of approximately 11,000 images in which the RGB color of the ambient illuminant in each scene is measured. To build such a large database we used a novel set-up consisting of a digital video camera with a neutral gray sphere attached to the camera so that the sphere always appears in the field of view. Using a gray sphere instead of the standard gray card facilitates measurement of the variation in illumination as a function of incident angle. The study focuses on the analysis of the distribution of various illuminants in the natural scenes and the correlation between the rg-chromaticity of colors recorded by the camera and the rg-chromaticity of the ambient illuminant. We also investigate the possibility of improving the performance of the naive Gray World algorithm by considering a sequence of consecutive frames instead of a single image. The set of images is publicly available and can also be used as a database for testing color constancy algorithms.

Underwater color constancy: enhancement of automatic live fish recognition

Abstract: We present in this paper some advances in color restoration of underwater images, especially with regard to the strong and non uniform color cast which is typical of underwater images. The proposed color correction method is based on ACE model, an unsupervised color equalization algorithm. ACE is a perceptual approach inspired by some adaptation mechanisms of the human visual system, in particular lightness constancy and color constancy. A perceptual approach presents a lot of advantages: it is unsupervised, robust and has local filtering properties, that lead to more effective results. The restored images give better results when displayed or processed (fish segmentation and feature extraction). The presented preliminary results are satisfying and promising.

Rendering HDR images

Abstract: Color imaging systems are continuously improving, and have now improved to the point of capturing high dynamic range scenes. Unfortunately most commercially available color display devices, such as CRTs and LCDs, are limited in their dynamic range. It is necessary to tone-map, or render, the high dynamic range images in order to display them onto a lower dynamic range device. This paper describes the use of an image appearance model, iCAM, to render high dynamic range images for display. Image appearance models have greater flexibility over dedicated tone-scaling algorithms as they are designed to predict how images perceptually appear, and not designed for the singular purpose of rendering. In this paper we discuss the use of an image appearance framework, and describe specific implementation details for using that framework to render high dynamic range images.

Color-to-grayscale conversion to maintain discriminability

Abstract: Monochrome devices that receive color imagery must perform a conversion from color to grayscale. The most common approach is to calculate the luminance signal from the three color signals. The problem with this approach is that the distinction between two colors of similar luminance (but different hue) is lost. This can be a significant problem when rendering colors within graphical objects such as pie charts and bar charts, which are often chosen for maximum discriminability. This paper proposes a method of converting color business graphics to grayscale in a manner that preserves discriminability. Colors are first sorted according to their original lightness values. They are then spaced equally in gray, or spaced according to their 3-D color difference from colors adjacent to them along the lightness dimension. This is most useful when maximum differentiability is desired in images containing a small number of colors, such as pie charts and bar graphs. Subjective experiments indicate that the proposed algorithms outperform standard color-to-grayscale conversions.

A Multi-Resolution, Full-Colour Spatial Gamut Mapping Algorithm.

Abstract: A multi–resolution, full–colour spatial gamut mapping algorithm (GMA) is proposed in this paper Its aim is to maintain as much of an original image’s overall, and in particular spatial, information as possible within the limits of a reproduction medium’s gamut First, the original image is decomposed into different spatial frequency bands Second, lightness compression and initial gamut mapping are applied to the lowest frequency band image Third, the next higher frequency band is added to the gamut mapped image and the result is processed by subsequent gamut mapping transformations The third step is repeated until the highest frequency band is reached The effect of this algorithm is that intra–image differences in the original image are well maintained in the gamut mapped reproduction A psychophysical experiment is then described whose results show that this algorithm is in the pair of most accurate GMAs and can outperform all other algorithms tested here for images which are less accurately reproduced by all GMAs

A multi-ink color-separation algorithm maximizing color constancy

Abstract: Current color-printing technologies may use three or more inks, e.g., CMY, CMYK, CMYKcm, CMYKGO, CMYKRGB. When the number of inks exceeds three, there is the usual color-management one-to-many mapping problem. Because the spectral properties of many modern inks are optimized for maximum color gamut and in some cases, black ink may not be used for pictorial images, many prints have poor color constancy. Changes in lighting dramatically change color balance, particularly for neutrals. An algorithm was developed for multi-ink printing in which the one-to-many mapping problem was overcome by selecting ink combinations with the best color constancy between illuminants F11 and D50. The algorithm was tested using a pigmented-ink inkjet proofing printer. CMYKGO prints color-separated using this algorithm were compared with a generic ICC profile for CMYKcm prints. The CMYK inks were common to both prints. The new algorithm improved color constancy significantly.

Estimating the Directions to Light Sources Using Images of Eye for Reconstructing 3D Human Face.

Abstract: This paper proposes a technique to estimate the direction of light sources based on the image of eye where the light source is imaged as reflection. The estimated directions of light sources are used to reconstruct 3D shape of face based on the photometric stereo technique. By using the reconstructed 3D shape, we can reproduce faces under various illuminants and from various viewing points. Without knowing the standing position of the subject, we can estimate the directions of light sources on human face from there images of eye for three light sources. In the process of this estimation, it is assumed that human eye is the sphere, and the geometrical constrains for reflection are used in the imaging system. The position and the size of human eye are also estimated as a result of the process. Since the standing position of the subject taken is not restricted in capturing the images, our imaging system is very practical to be used. The effectiveness of the proposed techniques are demonstrated by experiments. Introduction Reconstructing shape of human face has been required to reproduce faces under various illuminants and from various viewing points. In order to obtain the shape of human face, without using the expensive 3D scanner, the photometric stereo[3], which is a method for shape estimation that uses some intensity images obtained under different lighting conditions, has been used. However, with photometric stereo method, it is impossible to estimate the surface normal and the surface reflectance without a priori knowledge of the light source direction and the light source intensity. Therefore, in order to get the direction of light sources on human face exactly, it is necessary to determine the relation between the positions of light sources and subject. In this paper, we propose a method for estimating the direction of light source on human face by using the image of eye where the light source is imaged as reflection. Since the shape of human eye is almost sphere, and can reflect light, we can assume that human eye is used as mirrored ball to capture the environmental illuminants. According to mirrored ball technique[1,2], using the position of highlight peaks in the image of mirrored ball, where the light sources are reflected by on mirrored ball, the direction of light sources can be calculated correctly. However, this method requires the accurate position and radius of the mirrored ball. Since the radius of eye belongs with people, we cannot measure it easily. Moreover, the position of both subject and his human eye are not decided. Therefore it is difficult to estimate the direction of light sources without knowing the position and the radius of the human eye. In this paper, the position and the radius of the human eye is also estimated from the image of eye. In this estimation, we assumed that the position of three light sources in the camera coordination is known. However, we do not know where the subject stands or sit in front of the imaging system. By using the estimated direction of light sources, we can estimate the surface normal and the surface reflectance of human face by photometric stereo method. We can reconstruct 3D human face by integrating the estimated surface normal in the view coordinate. The effectiveness of this method is demonstrated by experiments. Geometry model of mirrored eye Figure 1 show the geometry model of proposed imaging system. This imaging process is based on the process of mirrored ball techniques [1,2]. Since the position of human face is not restricted, the human eye will be placed at any location in a natural environment. The camera is assumed to be a pinhole camera. Therefore, we can assume the process of imaging as if a screen is placed in front the camera. The distance of camera and screen is focal length. Figure 1 shows the projection of a human face in a 3D space onto a screen by the camera in perspective. In this figure, N is the surface normal at highlight peak, L is the light source directional vector on human eye, and V is the directional vector of camera at highlight peak. D is the distance between the camera and human eye, ( ) 0 0, y x is the center of screen. The directional vector of camera at the i highlight peak ( ) i i y x , on the image can be expressed using the focal lengthα and ( ) 0 0 , y x as follow, Figure 1. Geometry of the proposed imaging system [ ] i i y y x x α − − − = , , 0 0 V . (1) The coordinates of highlight peak ( ) i i y x , can be obtained from image since the center coordinates of image is known. The focal length α of camera is also obtained by camera calibration. The directional of camera is easily calculated as is shown in Figure 1. The surface normal vector N is determined using the sphere property. Let r be the radius of eye and ( ) c c y x , be the center coordinates of eye on the image. The surface normal vector at the i highlight peak ( ) i i y x , can be calculated by

Color conversion for a multi-primary display using linear interpolation on equi-luminance plane method (LIQUID)

Abstract: A color-conversion method, which converts an XYZ tristimulus value to a set of display signals for a multi-primary display, has been developed. It is necessary for a display-device driver in a color-management system to select an optimal set of display signals that presents a given colorimetric value on a display screen. A multi-primary display, which has more than four primaries, has more than two sets of display signals that generate the same colorimetric value theoretically. A display-device driver, therefore, has to be designed according to a certain rule in order to choose a unique set of display signals for one colorimetric value only. To meet the requirement, a linear interpolation method, which uses three points on an equi-luminance plane in XYZ tristimulus value space, has been developed. The developed method assigns three points on an equi-luminance plane in XYZ tristimulus value space to ensure smoothness of a continuous gradation of display signals. The colorimetric accuracy of the developed method was evaluated with a six-primary DLP projector. Colorimetric error in CIELAB space between the targets and reproductions is about 0.8 on average and 2.9 at maximum.

Interactive Goniochromatic Color Design

Abstract: An interactive program has been developed to assist in the design of new goniochromatic colors. The program gives the user a unique set of controls over a second order polynomial that defines these color families at a sequence of aspecular reflection angles. One approach, based on traditional metallic colors, allows the user to adjust the average hue, saturation, and brightness of all of the colors interpolated by the polynomial. Another method, appropriate for the newer effect colors, permits the designer to establish face and flop colors to be reached at either end of the interpolation. In a final technique, variations produced by adjusting model parameters can be evaluated and selected.

Colorimetric characterization of LCD and DLP projection displays

Abstract: Successful color-management of projection systems depends on knowledge of their characteristics. In this study, two typical portable projectors were characterized. The projectors are based on different technologies, liquid-crystal display (LCD) and digital light-processing (DLP). Measurements were made with a spectroradiometer. The properties measured were spectral characteristics and the intensity of the primary and white colors, basic colorimetric characteristics, inter-channel dependency, tone characteristics, color-tracking characteristics, spatial non-uniformity, dependency on background, and temporal stability. Based on the characterization results, the possibility of color-management of the tested projectors is discussed.

AM-FM screen design using donut filters

Abstract: In this paper we introduce a class of linear filters called 'donut filters' for the design of halftone screens that enable robust printing with stochastic0 clustered dots. The donut filter approach is a simple, yet efficient method to produce pleasing stochastic clustered-dot halftone patterns (a.k.a AM-FM halftones) suitable for systems with poor isolated dot reproduction and/or significant dot-gain. The radial profile of a donut filter resembles the radial cross section of a donut shape, with low impulse response at the center that rises to a peak and drops off rapidly as the pixel distance from the center is increased. A simple extension for the joint design of any number of colorant screens is given. This extension makes use of several optimal linear filters that may be treated as a single donut multi-filter having matrix-valued coefficients. A key contribution of this paper is the design of the parametric donut filters to be used at each graylevel. We show that given a desired spatial pair-correlation profile (a.k.a. spatial halftone statistics), optimum donut filters may be generated, such that the donut filter based screen design produces patterns possessing the desired profile in the maximum-likelihood sense. In fact, 'optimal green-noise' halftone screens having the spatial statistics described by Lau, Arce and Gallagher may be produced as a special case of our design. We will also demonstrate donut filter designs that do not use an 'optimum green-noise' target profile in the design and yet produce excellent stochastic clustered-dot halftone screens.

Gray tracking correction for TFT-LCDs

Abstract: The consistency of the white point with the input gray level is referred in the world of display technology as gray tracking. Gray tracking is an issue for Twisted Nematic (TN) TFT-LCD screens which typically show a bluish color shift of the white point when the gray level decreases. This paper discusses some causes of this color shift in TN displays and proposes a method for correcting it by using the existing video card look-up tables (VLUT) in the graphic controller driving the panel. The correction process performs the gamma correction and the gray tracking correction using a single set of three 1D look-up tables. The gray tracking correction uses the luminance and the chrominance information of the R, G, B channels. For a target gamma and white point, each VLUT entry that corresponds to a certain luminance and chrominance (target color), the method computes the output RGB values of the VLUT such that, the resulted gray has the minimum color difference to the target color. The method proves to be effective in removing the colorcast on the TN TFT-LCD screens. The method is different from a previous published paper from the same author by minimizing the loss of the luminance due to the tuning of the chrominance.

Influence of Spectral Sensitivity Functions on color demosaicing

Abstract: Color images acquired through single chip digital cameras using a color filter array (CFA) contain a mixture of luminance and opponent chromatic information that share their representation in the spatial Fourier spectrum. This mixture could result in aliasing if the bandwidths of these signals are too wide and their spectra overlap. In such a case, reconstructing three-color per pixel images without error is impossible. One way to improve the reconstruction is to have sensitivity functions that are highly correlated, reducing the bandwidth of the opponent chromatic components. However, this diminishes the ability to reproduce colors accurately as noise is amplified when converting an image to the final color encoding. In this paper, we are looking for an optimum between accurate image reconstruction through demosaicing and accurate color rendering. We design a camera simulation, first using a hyperspectral model of random color images and a demosaicing algorithm based on frequency selection. We find that there is an optimum and confirm our results using a natural hyperspectral image.

Controling color of liquid-crystal displays

Abstract: Two newly derived characterization models for a liquid-crystal (LC) display have been tested for five LC-based displays. Data measured from a series of test colors indicated that all LC-based displays showed similar characteristics, including an S-shaped tone curve and poor channel chromaticity constancy. Because they include a hyperbolic function in their definition, the models do not have analytical inverses, and so iterative mathematical techniques are applied. It was shown that a new characterization model based on a hyperbolic function fits the tone curve very accurately with only four coefficients per channel for any type of LCD. In addition, it was also shown that the first derivative of the function provides a means of accurate correction of the chromaticity variation.

The Spatial Properties of Contrast

Abstract: Contrast is the well-known observation in which a gray in a white surround appears darker than in a black one. An example is a 32 by 32 pixel gray area subtending 1.5 degrees visual angle on a black background 256 pixels on a side. The gray appears darker when surrounded by a band of white 12 pixels wide. The white band is made up of 2112 individual white pixels. If these white pixels are redistributed uniformly in the black background the gray appears much lighter. This paper measures the gray appearance as influenced by 2112 white pixel in a 27 different spatial configurations. The set of different spatial patterns of white pixels that generate the same matching lightness for gray are defined as equivalent backgrounds. The paper then analyzes the spatial properties of equivalent backgrounds. Gray appears darkest when the white surrounds the gray and is contiguous with it. In the case of the distributed white pixels, the gray appears lighter. This paper presents an analysis of the spatial properties of intermediate surrounds that give gray the equal visual appearances.

Color and Color Constancy in a Translation Model for Object Recognition

Abstract: Color is of interest to those working in computer vision largely because it is assumed to be helpful for recognition. This assumption has driven much work in color based image indexing, and computational color constancy. However, in many ways, indexing is a poor model for recognition. In this paper we use a recently developed statistical model of recognition which learns to link image region features with words, based on a large unstructured data set. The system is general in that it learns what is recognizable given the data. It also supports a principled testing paradigm which we exploit here to evaluate the use of color. In particular, we look at color space choice, degradation due to illumination change, and dealing with this degradation. We evaluate two general approaches to dealing with this color constancy problem. Specifically we address whether it is better to build color variation due to illumination into a recognition system, or, instead, apply color constancy preprocessing to images before they are processed by the recognition system.

A modular procedure for automatic red-eye correction in digital photos

Abstract: The paper describes an algorithm for the automatic removal of "redeye" from digital photos. First an adaptive color cast removal algorithm is applied to correct the color photo. This phase not only facilitates the subsequent steps of processing, but also improves the overall appearance of the output image. A skin detector, based mainly on analysis of the chromatic distribution of the image, creates a probability map of skin-like regions. A multi-resolution neural network approach is then exploited to create an analogous probability map of candidate faces. These two distributions are then combined to identify the most probable facial regions in the image. Redeye is searched for within these regions, seeking areas with high “redness” and applying geometric constraints to limit the number of false hits. The redeye removal algorithm is then applied automatically to the red eyes identified. Candidate areas are opportunely smoothed to avoid unnatural transitions between the corrected and original parts of the eyes. Experimental results of application of this procedure on a set of over 300 images are presented.

Spectral prediction and dot surface estimation models for halftone prints

Abstract: We propose a new spectral prediction model as well as new approaches for modeling ink spreading which occurs when printing ink layer superpositions. The spectral prediction model enhances the classical Clapper-Yule model by taking into account the fact that proportionally more incident light through a given colorant surface is reflected back onto the same colorant surface than onto other colorant surfaces. This is expressed by a weighted mean between a component specifying the part of the incident light which exits through the same colorant as the colorant from which it enters (Saunderson corrected Neugebauer component) and a component specifying the part of the incident light whose emerging light components exit from all colorants, with a probability to exit from a given colorant equal to that colorant surface coverage (Clapper-Yule component). We also propose two models for taking into account ink spreading, a phenomenon which occurs when printing an ink halftone in superposition with one or several solid inks. Besides the physical dot gain present within a single ink halftone print, we consider in the first model the ink spreading which occurs when an ink halftone is printed on top of one or two solid inks. In the second more advanced model, we generalize this concept to ink halftones printed on top or below solid inks. We formulate for both ink spreading models systems of equations which allow to compute effective ink coverages as a combination of the individual ink coverages which occur in the different superposition cases. The new spectral prediction model combined with advanced ink spreading yields excellent spectral predictions for clustered-dot color halftone prints, both in the case of offset (75 to 150 lpi) and in the case of thermal transfer printers (50 to 75 lpi)

Illuminant Multiplexed Imaging: Special Effects Using GCR.

Abstract: A novel spectral imaging technique is presented that allows multiple grayscale images to be combined in a single hardcopy print. The images are individually revealed when the print is illuminated by suitably chosen illuminants. In its basic form, the technique utilizes the complementary relation between cyan, magenta, and yellow colorants and narrow-band red, green, and blue illumination to combine and extract images. Colorant interactions are compensated through a simple calibration process. In addition, we present techniques demonstrating how gray component replacement (GCR) may be defined for illuminant multiplexed imaging and how it may be successfully exploited to allow embedding of additional patterns in the print that are revealed under uncontrolled illumination. The paper is accompanied by a light-booth based demonstration of illuminant multiplexed imaging and the effects described herein.

A solution to Mobile IP registration for AAA

Abstract: The Internet Engineering Task Force (IETF) Mobile IP working group has proposed the Mobile IP to support that an MH has the ability of roaming around a foreign domain. However, when it roams around a foreign domain, the authentication, authorization and accounting (AAA) for an MH must be done by the foreign domain. For this reason, the IETF suggests that the existence of some server capable of performing authentication, authorization, and accounting (AAA) services could help. So in this article we propose that employing simultaneously the public key cryptosystem and symmetric cryptosystem to accomplish the mission of Mobile IP registration for AAA.

Spectral imaging of Matisse's pot of geraniums: a case study

Abstract: The accuracy of color image-acquisition systems is most often evaluated using test targets of uniform color patches imaged under optimal conditions. In artwork imaging, system performance is judged visually, comparing the art with images rendered for display or print. Because the surface properties of the art may not be uniform, the spectral properties of the pigments may be different than the test targets, the sizes may be different, renderings are often metameric to the art, taking and viewing lighting geometries may be different, and the museum observers are more experienced than scientists in judging color accuracy visually, color accuracy as determined on a visual basis may be quite different than target performance. Therefore, an experiment was performed where a spectral-imaging system, designed for scientific purposes under laboratory conditions, was taken to a museum and tested in its photographic and conservation departments. The work of art evaluated was Henri Matisse’s Pot of Geraniums. Spectral and colorimetric comparisons were made between in situ small aperture spectrophotometry and imaging. The average performance was 3.7∆E00 and 3.1% spectral RMS; this was similar to an independent verification target of typical artist pigments applied to a canvas board. Viewed in close up, this level of accuracy yielded reasonable color matching for images rendered for display and print. Viewed overall, the matching quality worsened, a result of using diffuse lighting during image acquisition. Renderings appeared “flat” and reduced in perceived contrast. This indicates that when creating an image archive for both scientific and visual purposes, it will be necessary to use both directional and diffuse lighting geometries.

An interactive perception-based model for characterization of display devices

Abstract: This paper describes a simple to use, yet accurate way to obtain the Tone Reproduction Curve (TRC) of display devices without the use of a measurement device Human vision is used to compare a series of dithered color patches against interactively changeable homogeneously colored display areas Results comparing this method with spectrophotometer measurements are given for three monitors

CRT calibration techniques for better accuracy including low-luminance colors

Abstract: We present a new CRT characterization technique that improves the accuracy of the characterization. This is achieved by optimizing the linear transformation matrix of the two stage model in the uniform CIE/L*a*b* space. We also introduce an approach to improve the characterization performance for low luminance colors. These methods are used to calibrate two CRT monitors and better accuracies are obtained compared to existing methods, especially for low luminance colors. We present a systematic way to adjust the white point of the monitor using hardware settings. This allows us to adjust the monitor white accurately without losing any digital counts which is the case if a software approach is used. We propose a novel search algorithm to achieve very high accuracy calibration for experiments where a limited number of colors has to be displayed. We apply this search algorithm to the case of monochrome image display application and verify the performance our method.

A Radial Sampling of the OSA Uniform Color Scales.

Abstract: The OSA Uniform Color Scales were derived using a unique geometry for the physical samples. Regular rhombohedral packing allows each sample to be compared to twelve other equally distant samples. While this sampling scheme provides an efficient geometry for sample comparison and allows multiple cleavage planes, it obscures the underlying perceptual attributes. However, it is relatively straightforward to compute a radial sampling of data points in OSA. This radial sampling results in a distance from the achromatic axis and an angular quantity and can be used to compare other color spaces. This paper presents a method and considerations for computing the radial sampling. The utility of this data is demonstrated by comparing the perceptual uniformity of five different color spaces. Introduction Color spaces and color order systems provide a means to organize and describe color. A uniform color space is one in which equal differences in any part of the space correspond to equal perceptual differences. There is an abundance of prior and ongoing research in the area of uniform color spaces. The OSA Uniform Color Scales or UCS is one such color space and was the result of extensive psychophyisical investigation and other considerations. The UCS exists both as physical samples and in equation form. The samples are based on a regular rhombohedral sampling scheme. This sampling results in twelve equally spaced neighbors for any given color and yields the closest uniform three dimensional sampling. This sampling scheme has a number of advantages, including visualization of color differences along multiple cleavage planes and efficient sampling for psychophysical evaluation. However this sampling scheme can make it more difficult to visualize color space uniformity. For example the OSA UCS space has been compared to other color spaces and color order systems, such as CIELAB or the Munsell Book of Color, or as a qualitative demonstration of perceptual uniformity for a physiologically based system of photometry and colorimetry. However the radial structure of the Munsell Book of Color and the regular rhombohedral structure of the OSA UCS can make visual comparisons difficult. However, aside from computational considerations there is no specific limitation or restriction on calculating alternative sampling schemes for the OSA UCS. This paper reviews the OSA UCS, the forward transformations and then proposes a radial sampling of the OSA UCS. Finally as an example of how this data can be used, the CIELAB, CIELUV, CIECAM97s, IPT and CIECAM02 color spaces are compared to each other using the radial OSA UCS data.

Influence Of Test Image Choice On Experimental Results.

Abstract: The choice of test images is an important part of the experimental evaluation of colour reproduction algorithms. As previous studies show low levels of correlation between individual image results, this study explores the impact of test image choice on the overall results of gamut mapping algorithm evaluation experiments. A set of fifteen test images is used in an experiment designed according to CIE TC8–03 Guidelines to evaluate the performance of six GMAs. The results first show the significant difference between how accurately individual images can be reproduced. Most importantly the results also show that five–image subsets can be selected so as to have each of the six GMAs in the top group for that set. Hence it is necessary to use greater numbers of test images in future experiments to obtain more robust and repeatable overall results.