Showing papers on "Color constancy published in 2000"

Color signals in natural scenes: characteristics of reflectance spectra and effects of natural illuminants

Abstract: Multispectral images of natural scenes were collected from both forests and coral reefs to represent typical, complex scenes that might be viewed by modern animals. Both reflectance spectra and modeled visual color signals in these scenes were decorrelated spectrally by principal-component analysis. Nearly 98% of the variance of reflectance spectra and color signals can be described by the first three principal components for both forest and coral reef scenes, which implies that three well-designed visual channels can recover almost all of the spectral information of natural scenes. A variety of natural illuminants affects color signals of forest scenes only slightly, but the variation in ambient irradiance spectra that is due to the absorption of light by water has dramatic influences on the spectral characteristics of coral reef scenes.

Improving gamut mapping color constancy

Abstract: The color constancy problem, that is, estimating the color of the scene illuminant from a set of image data recorded under an unknown light, is an important problem in computer vision and digital photography. The gamut mapping approach to color constancy is, to date, one of the most successful solutions to this problem. In this algorithm the set of mappings taking the image colors recorded under an unknown illuminant to the gamut of all colors observed under a standard illuminant is characterized. Then, at a second stage, a single mapping is selected from this feasible set. In the first version of this algorithm Forsyth (1990) mapped sensor values recorded under one illuminant to those recorded under a second, using a three-dimensional (3-D) diagonal matrix. However because the intensity of the scene illuminant cannot be recovered Finlayson (see IEEE Trans. Pattern Anal. Machine Intell. vol.18, no.10, p.1034-38, 1996) modified Forsyth's algorithm to work in a two-dimensional (2-D) chromaticity space and set out to recover only 2-D chromaticity mappings. While the chromaticity mapping overcomes the intensity problem it is not clear that something has not been lost in the process. The first result of this paper is to show that only intensity information is lost. Formally, we prove that the feasible set calculated by Forsyth's original algorithm, projected into 2-D, is the same as the feasible set calculated by the 2-D algorithm. Thus, there is no advantage in using the 3-D algorithm and we can use the simpler, 2-D version of the algorithm to characterize the set of feasible illuminants. Another problem with the chromaticity mapping is that it is perspective in nature and so chromaticities and chromaticity maps are perspectively distorted. Previous work demonstrated that the effects of perspective distortion were serious for the 2-D algorithm. Indeed, in order to select a sensible single mapping from the feasible set this set must first be mapped back up to 3-D. We extend this work to the case where a constraint on the possible color of the illuminant is factored into the gamut mapping algorithm. We show here that the illumination constraint can be enforced during selection without explicitly intersecting the two constraint sets. In the final part of this paper we reappraise the selection task. Gamut mapping returns the set of feasible illuminant maps. Our new algorithm is tested using real and synthetic images. The results of these tests show that the algorithm presented delivers excellent color constancy.

The Colors Seen behind Transparent Filters

Abstract: How do the colors and lightnesses of surfaces seen to lie behind a transparent filter depend on the chromatic properties of the filter? A convergence model developed in prior work (D'Zmura et al, 1997 Perception 26 471-492; Chen and D'Zmura, 1998 Perception 27 595-608) suggests that the visual system interprets a filter's transformation of color in terms of a convergence in color space. Such a convergence is described by a color shift and a change in contrast. We tested the model using an asymmetric matching task. Observers adjusted, in computer graphic simulation, the color of a surface seen behind a transparent filter in order to match the color of a surface seen in plain view. The convergence model fits the color-matching results nearly as well as a more general affine-transformation model, even though the latter has many more parameters. Other models, including von Kries scaling, did not perform as well. These results suggest that the color constancy revealed in this task is described best by a model that takes into account both color shifts and changes in contrast.

Characterization of natural illuminants in forests and the use of digital video data to reconstruct illuminant spectra

Abstract: We describe illumination spectra in forests and show that they can be accurately recovered from recorded digital video images. Natural illuminant spectra of 238 samples measured in temperate forests were characterized by principal-component analysis. The spectra can be accurately approximated by the mean and the first two principal components. Compared with illumination under open skies, the loci of forest illuminants are displaced toward the green region in the chromaticity plots, and unlike open sky illumination they cannot be characterized by correlated color temperature. We show that it is possible to recover illuminant spectra accurately from digital video images by a linear least-squares-fit estimation technique. The use of digital video data in spectral analysis provides a promising new approach to the studies of the spatial and temporal variation of illumination in natural scenes and the understanding of color vision in natural environments.

Contribution of S opponent cells to color appearance

Abstract: We measured the regions in isoluminant color space over which observers perceive red, yellow, green, and blue and examined the extent to which the colors vary in perceived amount within these regions. We compared color scaling of various isoluminant stimuli by using large spots, which activate all cone types, to that with tiny spots in the central foveola, where S cones, and thus S opponent (So) cell activity, are largely or entirely absent. The addition of So input to that from the L and M opponent cells changes the chromatic appearance of all colors, affecting each primary color in different chromatic regions in the directions and by the amount predicted by our color model. Shifts from white to the various chromatic stimuli we used produced sinusoidal variations in cone activation as a function of color angle for each cone type and in the responses of lateral geniculate cells. However, psychophysical color-scaling functions for 2° spots were nonsinusoidal, being much more peaked. The color-scaling functions are well fit by sine waves raised to exponents between 1 and 3. The same is true for the color responses of a large subpopulation of striate cortex cells. The narrow color tuning, the discrepancies between the spectral loci of the peaks of the color-scaling curves and those of lateral geniculate cells, and the changes in color appearance produced by eliminating So input provide evidence for a cortical processing stage at which the color axes are rotated by a combination of the outputs of So cells with those of L and M opponent cells in the manner that we postulated earlier. There seems to be an expansive response nonlinearity at this stage.

Invariant cone-excitation ratios may predict transparency.

Abstract: Cone-excitation ratios for pairs of surfaces are almost invariant under changes in illumination and offer a possible basis for color constancy [Proc. R. Soc. London Ser. B257, 115 (1994)]. We extend this idea to the perception of transparency on the basis of the close analogy between the changes in color signals that occur for surfaces when the illumination changes and the changes in color signals when the surfaces are covered by a filter. This study presents measurements and simulations to investigate the conditions under which cone-excitation ratios are statistically invariant for physically transparent systems. The invariance breaks down when the spectral transmission of the filters is low at some or all wavelengths. We suggest that cone-excitation ratios might be useful to define the stimulus conditions necessary for the perception of transparency.

Color appearance of surfaces viewed through fog.

Abstract: How do the colors of surfaces seen through fog depend on the chromatic properties of the fog? Prior work (eg Chen and D'Zmura, 1998 Perception27 595–608) shows that the colors of surfaces seen through a transparent filter can be described by a convergence model. The convergence model takes into account color shift and change in contrast. Whether the convergence model can also be applied to fog was tested experimentally with an asymmetric matching task. In computer graphic simulation, observers adjusted the color of a surface seen through fog in order to match the color of a surface seen in the absence of fog. The convergence model fits the data well. The results suggest that the color constancy revealed in this task with fog is described best by a model that takes into account both shift in color and change in contrast.

Relational color constancy in achromatic and isoluminant images

Abstract: Relational color constancy, which refers to the constancy of perceived relations between surface colors under changes in illuminant, may be based on the computation of spatial ratios of cone excitations. As this activity need occur only within rather than between cone pathways, relational color constancy might be assumed to be based on relative luminance processing. This hypothesis was tested in a psychophysical experiment in which observers viewed simulated images of Mondrian patterns undergoing colorimetric changes that could be attributed either to an illuminant change or to a nonilluminant change; the images were isoluminant, achromatic, or unmodified. Observers reliably discriminated the two types of changes in all three conditions, implying that relational color constancy is not based on luminance cues alone. A computer simulation showed that in these isoluminant and achromatic images spatial ratios of cone excitations and of combinations of cone excitations were almost invariant under illuminant changes and that discrimination performance could be predicted from deviations in these ratios.

Color constancy in goldfish: the limits.

Abstract: Color constancy was investigated in behavioral training experiments on colors ranging from blue to yellow, located in the color space close to Planck's locus representing the main changes in natural skylight. Two individual goldfish were trained to peck at a test field of medium hue out of a series of 13–15 yellowish and bluish test fields presented simultaneously on a black background. During training the tank in which the fish were swimming freely was illuminated with white light. Correct choices were rewarded with food. During the tests differently saturated yellow or blue illumination was used. The degree of color constancy was inferred from the choice behavior under these illuminations. Perfect color constancy was found up to a certain degree of saturation of the colored light. Beyond this level test fields other than the training test field were chosen, indicating imperfect color constancy. Color constancy was quantified by applying color metrics on the basis of the goldfish cone sensitivity functions.

Color texture classification by normalized color space representation

Abstract: This paper proposes a novel approach to color texture characterization and classification. Rather than developing new textural features, we propose to derive a family of new, reduced dimensionality color spaces named P/sub 1/P/sub 2/, that allow a good classification performance by the use of classical energy-distribution features, defined in a scalar spectral domain. The dimensionality reduction approach can be traced back to color constancy normalization and the reduced ordering principle and exhibits a strong perceptual background. We develop an adaption procedure for the selection of the proper color space within the new P/sub 1/P/sub 2/ family. The overall classification performance is very promising and the proposed methodology surmounts the current color texture characterization by features extracted from the luminance spectrum only.

Color blindness and a color human visual system model

Abstract: A physiologically motivated human color visual system model which represents visual information with one brightness component (A) and two chromatic components (C/sub 1/ and C/sub 2/) is used to create stimuli for testing the color perception of deuteranomalous trichromats. Two experiments are performed. Using simple ramp patterns, the first experiment finds that three deuteranomalous trichromat test subjects can distinguish variations only in the C/sub 2/ component of the color vision model. This finding is further tested in the second experiment: a set of paired comparison preference tests. Two altered versions of each of three natural color images are prepared by setting either one of the color components to a constant over the full image. Pairs of an original and a distorted image are presented to the test subjects, and they are asked to indicate which image they prefer. The experimental results indicate that the C/sub 1/ channel is severely attenuated in the deuteranomalous trichromat test subjects, and that nearly all their color sensation is mediated by the C/sub 2/ channel of the color vision model.

Extending color vision methods to bands beyond the visible

Abstract: There is a rapid growth in sensor technology in regions of the electromagnetic spectrum beyond the visible. Also, the parallel growth of powerful, inexpensive computers and digital electronics has made many new imaging applications possible. Although there are many approaches to sensor fusion, this paper provides a discussion of relevant infrared phenomenology and attempts to apply known methods of human color vision to achieve image fusion. Two specific topics of importance are color contrast enhancement and color constancy.

Color Property of the Recognized Visual Space of Illumination Controlled by Interior Color as the Initial Visual Information

Abstract: It was shown that the color property of the recognized visual space of illumination, RVSI was controlled by changing the initial visual information by arranging objects in the room all shifting toward orange direction. We constructed two miniature rooms, D and I, both illuminated by the same daylight type fluorescent lamps but arranged with furniture of different color, those in room I shifting toward color as if they were illuminated by an incandescent lamp. Subjects felt as if room I were illuminated by an incandescent lamp. A test patch was placed midair in each room and its color was judged. When the test patches were placed in room I their colors were all perceived to be shifted toward greenish blue compared to those of test patches placed in room D, in spite of having the same illumination. The results imply that the apparent color of an object is determined not by its chromaticity, but in relation to the color property of the RVSI of the room where the object is observed.

Color constancy: a biological model and its application for still and video images

Abstract: A novel biological model for color constancy (CC) is presented. This is the first biological model that succeeds in achieving automatic color correction of real still and video images (registered patent). It is based on two chromatic adaptation mechanisms in retinal color-coded cells: local adaptation and remote chromatic adaptation. Our simulations calculate the perceived image in order to correct image colors (as is commonly required in cameras). The results indicate that the contribution of adaptation mechanisms to CC is significant, robust, and succeeded in performing color correction of still images and video sequences under single and multiple illumination conditions.

Visual communication with retinex coding

Abstract: Visual communication with retinex coding seeks to suppress the spatial variation of the irradiance (e.g., shadows) across natural scenes and preserve only the spatial detail and the reflectance (or the lightness) of the surface itself. The separation of reflectance from irradiance begins with nonlinear retinex coding that sharply and clearly enhances edges and preserves their contrast, and it ends with a Wiener filter that restores images from this edge and contrast information. An approximate small-signal model of image gathering with retinex coding is found to consist of the familiar difference-of-Gaussian bandpass filter and a locally adaptive automatic-gain control. A linear representation of this model is used to develop expressions within the small-signal constraint for the information rate and the theoretical minimum data rate of the retinex-coded signal and for the maximum-realizable fidelity of the images restored from this signal. Extensive computations and simulations demonstrate that predictions based on these figures of merit correlate closely with perceptual and measured performance. Hence these predictions can serve as a general guide for the design of visual communication channels that produce images with a visual quality that consistently approaches the best possible sharpness, clarity, and reflectance constancy, even for nonuniform irradiances. The suppression of shadows in the restored image is found to be constrained inherently more by the sharpness of their penumbra than by their depth.

The perception of color at dawn and dusk

Abstract: The illumination of scenes by sunlight at dawn and dusk creates a rich array of colors and long high contrast shadows that make the appearance of the scene more beautiful and allow for better photographic imagery. Photographers often refer to this as when 'the light is right' as opposed to mid-day illumination when the light is 'flat and uninteresting'. When the light of the setting sun illuminates an object or person the surfaces impart a characteristic warm glow - the warm colors of the scene can almost appear self-luminous. This paper addresses this sensation with respect to color constancy, color appearance, chromatic adaptation, and digital photography. Because photographers strive for this kind of illumination it is vital that in the design of a color model for a photographic system that we understand and preserve the photographer's control of this effect.

Color correcting uncalibrated digital images

Abstract: Color images often must be color balanced to remove unwanted color casts. Color balancing uncalibrated images (e.g. downloaded from the Internet or scanned from an unknown film) adds additional challenges to the already difficult problem of color correction because neither the pre-processing to which the image was subjected, nor the camera sensors or camera balance are known. In this article, we propose a framework for dealing with some aspects of this type of image. In particular, we discuss the issue of color correcting images where an unknown gamma' non-linearity may be present. We show that the diagonal model, used for color correcting linear images, also works in the case of gamma corrected images. We also discuss the influence that unknown camera balance and unknown sensors have on color constancy algorithms. To perform color correction on uncalibrated images, we extend previous work on using a neural network for illumination, or white-point, estimation from the case of calibrated images to that of uncalibrated images of unknown origin. The results show that the chromaticity of the ambient illumination in uncalibrated images can be estimated with an average CIE Lab error around 5AE. Comparisons are made to the grayworld and white-patch methods.

Estimation of spectral distribution of scene illumination from a single image

Abstract: This article proposes an illuminant estimation algorithm that estimates the spectral power distribution of an incident light source from a single image. The proposed illumination recovery procedure has two phases. First, the surface spectral reflectances are recovered. In this case, the surface spectral reflectances recovered are limited to the maximum achromatic region (MAR) which is the most achromatic and highly bright region of an image, after applying intermediate color constancy process using a modified gray-world algorithm. Next, the surface reflectances of the maximum achromatic region are estimated using the principal component analysis method along with a set of given Munsell samples. Second, the spectral distribution of reflected lights of MAR is selected from the spectral database. That is, a color difference is compared between the reflected lights of the MAR and the spectral database, which is the set of reflected lights built by the given Munsell samples and a set of illuminants. Then the closest colors from the spectral database are selected. Finally, the illuminant of an image can be calculated dividing the average spectral distributions of reflected lights of MAR by the average surface reflectances of the MAR. In order to evaluate the proposed algorithm, experiments with artificial and real captured color-biased scenes were performed and numerical comparison examined. The proposed method was effective in estimating the spectral distribution of the given illuminants under various illuminants and scenes without white points.

How to Shop on the Web Without Seeing Red.

Abstract: — Using the World Wide Web to order goods and services is a rapidly increasing activity. Experience with mail-order catalogs has shown that failure of goods to match the catalog color is a major category of complaint. Ordering colored goods on the Web poses even greater challenges. Useful devices to standardize cathode-ray tubes, and other self-luminous displays, are available, but these only address the problem of display set-up. Other relevant problems include the effects of changes in the level of luminance, lack of color constancy of goods with changes in illuminant color, departures from a set of color-matching functions of the spectral sensitivities of acquisition devices, observer metamerism, the effect of the surround on the appearance of the display, and limitations in the gamut and in the resolution of typical displays. The extent to which these factors are likely to be important is discussed, and ways in which some of their effects might be mitigated are considered.

Dynamics of Asymmetric Color Matching

Abstract: It is known that the human color constancy is not complete. We conducted asymmetric color matching experiments to clarify a simple question: what was it that the observer matched. Observers made apparent-color matches between Munsell color chips under D65 illumination and a color chip presented on a CRT. The observers’ matches showed incomplete color constancy. We applied relative cone-weight transformation to each observer’s results under control condition, so as to equate the unique-white point for each illuminant condition. The result of this simple transformation showed good match to the actual data. Our results suggest what the observer was doing during the asymmetric color matches was picking up apparent-color signal, achieved from the test color chip under D65, and reconstructing it with respect to the unique-white by a simple cone response scaling.

Colour Constancy from Hyper-Spectral Data.

Abstract: This paper aims for color constant identification of object colors through the analysis of spectral color data. New computational color models are proposed which are not only invariant to illumination variations (color constancy) but also robust to a change in viewpoint and object geometry (color invariance). Color constancy and invariance is achieved by spectral imaging using a white reference, and based on color ratio’s (without a white reference). From the theoretical and experimental results it is concluded that the proposed computational methods for color constancy and invariance are highly robust to a change in SPD of the light source as well as a change in the pose of the object.

Computational colour constancy

Abstract: The colour response of a camera or of the human eye is confounded by illumination. The colours recorded under yellow and blue illuminants are respectively more yellowish and bluish than they ought to be. Removing colour bias due to illumination is called colour constancy. Opinions differ as to whether colour constancy is a reasonably well solved problem or one that is still far beyond our reach. These conflicting conclusions result from the different ways that the colour constancy problem is defined. In this paper we review the color constancy problem placing special emphasis on the different definitions used. The links between the different approaches will also be made clear.

Color Constancy Using the Inter-Reflection from a Reference Nose

Abstract: This paper introduces a novel camera attachment for measuring the illumination color spatially in the scene The illumination color is then used to transform color appearance in the image into that under white light The main idea is that the scene inter-reflection through a reference camera-attached surface “Nose” can, under some conditions, represent the illumination color directly The illumination measurement principle relies on the satisfaction of the gray world assumption in a local scene area or the appearance of highlights, from dielectric surfaces Scene inter-reflections are strongly blurred due to optical dispersion on the nose surface and defocusing of the nose surface image Blurring smoothes the intense highlights and it thus becomes possible to measure the nose inter-reflection under conditions in which intensity variation in the main image would exceed the sensor dynamic range We designed a nose surface to reflect a blurred scene version into a small image section, which is interpreted as a spatial illumination image The nose image is then mapped to the main image for adjusting every pixel color Experimental results showed that the nose inter-reflection color is a good measure of illumination color when the model assumptions are satisfied The nose method performance, operating on real images, is presented and compared with the Retinex and the scene-inserted white patch methods

Prefiltering with Retinex in color image retrieval

Abstract: We have examined the performance of various color-based retrieval strategies when coupled with a pre-filtering Retinex algorithm to see whether, and to what degree, Retinex improved the effectiveness of the retrieval, regardless of the strategy adopted. The retrieval strategies implemented included color and spatial-chromatic histogram matching, color coherence vector matching, and the weighted sum of the absolute differences between the first three moments of each color channel. The experimental results are reported and discussed.

Using Color Constancy to Advantage in Color Gamut Calculations.

Abstract: The human color constancy uses spatial comparisons. The relationships among neighboring pixels are far more important than the absolute differences between the colorimetric values of an original and its gamut-limited reproduction. If all the pixels in an image have a reproduction error in the same direction (red, green, blue, lightness, hue, chroma), then our color constancy mechanism helps to make large errors appear small. However, if all the errors are randomly distributed, then small errors appear large. This paper will describe experiments using constant errors to produce variable apparent errors and describe a technique of calculating the best appearance image using spatial comparisons. This calculation will be applied to colorgamut problems.

A spectral-imaging system and algorithms for recovering spectral functions

Abstract: Knowledge of such spectral functions as the surface-spectral reflectances of objects and the illuminant spectral power distribution is needed for realizing color constancy, accurate color reproduction, and rendering realistic images. This paper describes a spectral-imaging system using a liquid-crystal tunable filter and algorithms for recovering the spectral functions of illuminant and surface reflectance from the image data. First, we introduce a filtering mechanism and the camera system. Next, the camera outputs and the spectral functions are represented by finite-dimensional linear models. The algorithms for estimating the spectral functions are then presented using the image data. Finally the usefulness of the proposed imaging system and algorithms is shown in an experiment.