Showing papers on "Color constancy published in 1997"

A multiscale retinex for bridging the gap between color images and the human observation of scenes

Abstract: Direct observation and recorded color images of the same scenes are often strikingly different because human visual perception computes the conscious representation with vivid color and detail in shadows, and with resistance to spectral shifts in the scene illuminant. A computation for color images that approaches fidelity to scene observation must combine dynamic range compression, color consistency-a computational analog for human vision color constancy-and color and lightness tonal rendition. In this paper, we extend a previously designed single-scale center/surround retinex to a multiscale version that achieves simultaneous dynamic range compression/color consistency/lightness rendition. This extension fails to produce good color rendition for a class of images that contain violations of the gray-world assumption implicit to the theoretical foundation of the retinex. Therefore, we define a method of color restoration that corrects for this deficiency at the cost of a modest dilution in color consistency. Extensive testing of the multiscale retinex with color restoration on several test scenes and over a hundred images did not reveal any pathological behaviour.

Properties and performance of a center/surround retinex

Abstract: The last version of Land's (1986) retinex model for human vision's lightness and color constancy has been implemented and tested in image processing experiments. Previous research has established the mathematical foundations of Land's retinex but has not subjected his lightness theory to extensive image processing experiments. We have sought to define a practical implementation of the retinex without particular concern for its validity as a model for human lightness and color perception. We describe the trade-off between rendition and dynamic range compression that is governed by the surround space constant. Further, unlike previous results, we find that the placement of the logarithmic function is important and produces best results when placed after the surround formation. Also unlike previous results, we find the best rendition for a "canonical" gain/offset applied after the retinex operation. Various functional forms for the retinex surround are evaluated, and a Gaussian form is found to perform better than the inverse square suggested by Land. Images that violate the gray world assumptions (implicit to this retinex) are investigated to provide insight into cases where this retinex fails to produce a good rendition.

Bayesian color constancy.

Abstract: The problem of color constancy may be solved if we can recover the physical properties of illuminants and surfaces from photosensor responses. We consider this problem within the framework of Bayesian decision theory. First, we model the relation among illuminants, surfaces, and photosensor responses. Second, we construct prior distributions that describe the probability that particular illuminants and surfaces exist in the world. Given a set of photosensor responses, we can then use Bayes’s rule to compute the posterior distribution for the illuminants and the surfaces in the scene. There are two widely used methods for obtaining a single best estimate from a posterior distribution. These are maximum a posteriori (MAP) and minimum mean-squared-error (MMSE) estimation. We argue that neither is appropriate for perception problems. We describe a new estimator, which we call the maximum local mass (MLM) estimate, that integrates local probability density. The new method uses an optimality criterion that is appropriate for perception tasks: It finds the most probable approximately correct answer. For the case of low observation noise, we provide an efficient approximation. We develop the MLM estimator for the color-constancy problem in which flat matte surfaces are uniformly illuminated. In simulations we show that the MLM method performs better than the MAP estimator and better than a number of standard color-constancy algorithms. We note conditions under which even the optimal estimator produces poor estimates: when the spectral properties of the surfaces in the scene are biased. © 1997 Optical Society of America [S0740-3232(97)01607-4]

Color constancy in the nearly natural image. I. Asymmetric matches

Abstract: Most empirical work on color constancy is based on simple laboratory models of natural viewing conditions. These typically consist of spots seen against uniform backgrounds or computer simulations of flat surfaces seen under spatially uniform illumination. We report measurements made under more natural viewing conditions. The experiments were conducted in a room where the illumination was under computer control. Observers used a projection colorimeter to set asymmetric color matches across a spatial illumination gradient. Observers' matches can be described by either of two simple models. One model posits gain control in one-specific pathways. This diagonal model may be linked to ideas about the action of early visual mechanisms. The other model posits that the observer estimates and corrects for changes in illumination but does so imperfectly. This equivalent illuminant model provides a link between human performance and computational models of color constancy.

Method of improving a digital image

Abstract: A method of improving a digital image is provided. The image is initially represented by digital data indexed to represent positions on a display. The digital data is indicative of an intensity value Ii(x,y) for each position (x,y) in each i-th spectral band. The intensity value for each position in each i-th spectral band is adjusted to generate an adjusted intensity value for each position in equation (I), each i-th spectral band in accordance with where S is the number of unique spectral bands included in said digital data, Wn is a weighting factor and '*' denotes the convolution operator. Each surround function Fn(x,y) is uniquely scaled to improve an aspect of the digital image, e.g., dynamic range compression, color constancy, and lightness rendition. The adjusted intensity value for each position in each i-th spectral band is filtered with a common function and then presented to a display device. For color images, a novel color restoration step is added to give the image true-to-life color that closely matches human observation.

Learning bilinear models for two-factor problems in vision

Abstract: In many vision problems, we want to infer two (or more) hidden factors which interact to produce our observations. We may want to disentangle illuminant and object colors in color constancy; rendering conditions from surface shape in shape-from-shading; face identity and head pose in face recognition; or font and letter class in character recognition. We refer to these two factors generically as "style" and "content". Bilinear models offer a powerful framework for extracting the two-factor structure of a set of observations, and are familiar in computational vision from several well-known lines of research. This paper shows how bilinear models can be used to learn the style-content structure of a pattern analysis or synthesis problem, which can then be generalized to solve related tasks using different styles and/or content. We focus on three tasks: extrapolating the style of data to unseen content classes, classifying data with known content under a novel style, and translating data from novel content classes and style to a known style or content. We show examples from color constancy, face pose estimation, shape-from-shading, typography and speech.

A Comparison of the Multiscale Retinex With Other Image Enhancement Techniques

Abstract: The multiscale retinex with color restoration (MSRCR) has shown itself to be a very versatile automatic image enhancement algorithm that simultaneously provides dynamic range compression, color constancy, and color rendition. A number of algorithms exist that provide one or more of these features, but not all. In this paper we compare the performance of the MSRCR with techniques that are widely used for image enhancement. Specifically, we compare the MSRCR with color adjustment methods such as gamma correction and gain/offset application, histogram modification techniques such as histogram equalization and manual histogram adjustment, and other more powerful techniques such as homomorphic filtering and ''burning and dodging''. The comparison is carried out by testing the suite of image enhancement methods on a set of diverse images. We find that though some of these techniques work well for some of these images, only the MSRCR performs universally well on the test set.

Color constancy in variegated scenes: role of low-level mechanisms in discounting illumination changes

Abstract: For a visual system to possess color constancy across varying illumination, chromatic signals from a scene must remain constant at some neural stage. We found that photoreceptor and opponent-color signals from a large sample of natural and man-made objects under one kind of natural daylight were almost perfectly correlated with the signals from those objects under every other spectrally different phase of daylight. Consequently, in scenes consisting of many objects, the effect of illumination changes on specific color mechanisms can be simulated by shifting all chromaticities by an additive or multiplicative constant along a theoretical axis. When the effect of the illuminant change was restricted to specific color mechanisms, thresholds for detecting a change in the colors in a scene were significantly elevated in the presence of spatial variations along the same chromatic axis as the simulated chromaticity shift. In a variegated scene, correlations between spatially local chromatic signals across illuminants, and the desensitization caused by eye movements across spatial variations, help the visual system to attenuate the perceptual effects that are due to changes in illumination.

Data and Model-Driven Selection Using Color Regions

Abstract: A key problem in model-based object recognition is selection, namely, the problem of determining which regions in the image are likely to come from a single object. In this paper we present an approach that uses color as a cue to perform selection either based solely on image-data (data-driven), or based on the knowledge of the color description of the model (model-driven). Specifically, the paper presents a method of color specification in terms of perceptual color categories and shows its relevance for the task of selection. The color categories are used to develop a fast region segmentation algorithm that extracts perceptual color regions in images. The color regions extracted form the basis for performing data and model-driven selection. Data-driven selection is achieved by selecting salient color regions as judged by a color-saliency measure that emphasizes attributes that are also important in human color perception. The approach to model-driven selection, on the other hand, exploits the color and other region information in the 3d model object to locate instances of the object in a given image. The approach presented tolerates some of the problems of occlusion, pose and illumination changes that make a model instance in an image appear different from its original description. Finally, the utility of color-based selection is demonstrated by showing the extent of search reduction possible when color-based selection is integrated with a recognition system.

Analysis and improvement of multi-scale retinex

Abstract: The main thrust of this paper is to modify the multi-scale retinex (MSR) approach to image enhancement so that the processing is more justified from a theoretical standpoint. This leads to a new algorithm with fewer arbitrary parameters that is more flexible, maintains color fidelity, and still preserves the contrast-enhancement benefits of the original MSR method. To accomplish this we identify the explicit and implicit processing goals of MSR. By decoupling the MSR operations from one another, we build an algorithm composed of independent steps that separates out the issues of gamma adjustment, color balance, dynamic range compression, and color enhancement, which are all jumbled together in the original MSR method. We then extend MSR with color constancy and chromaticity-preserving contrast enhancement.

Illumination-invariant video segmentation by hierarchical robust thresholding

Abstract: Many methods for video segmentation rely upon the setting and tuning of thresholds for classifying interframe distances under various difference measures An approach that has been used with some success has been to establish statistical measures for each new video and identify camera cuts as difference values far from the mean For this type of strategy the mean and dispersion for some interframe distance measure must be calculated for each new video as a whole Here we eliminate this statistical characterization step and at the same time allow for segmentation of streaming video by introducing a preprocessing step for illumination-invariance that concomitantly reduces input values to a uniform scale The preprocessing step provides a solution to the problem that simple changes of illumination in a scene, such as an actor emerging from a shadow, can trigger a false positive transition, no matter whether intensity alone or chrominance is used in a distance measure Our means of discounting lighting change for color constancy consists of the simple yet effective operation of normalizing each color channel to length 1 (when viewed as a long, length-N vector) We then reduce the dimensionality of color to two-dimensional chromaticity, with values which are in 01 Chromaticity histograms can be treated as images, and effectively low-pass filtered by wavelet-based reduction, followed by DCT and zonal coding This results in an indexing scheme based on only 36 numbers, and lends itself to a binary search approach to transition detection To this end we examine distributions for intra-clip and inter-clip distances separately, characterizing each using robust statistics, for temporal intervals from 32 frames to 1 frame by powers of 2 Then combining transition and non-transition distributions for each frame internal, we seek the valley between them, again robustly, for each threshold Using the present method values of precision and recall are increased over previous methods Moreover, illumination change produces very few false positives© (1997) COPYRIGHT SPIE--The International Society for Optical Engineering Downloading of the abstract is permitted for personal use only

The illumination-invariant matching of deterministic local structure in color images

Abstract: The availability of multiple spectral measurements at each pixel in an image provides important additional information for recognition. Spectral information is of particular importance for applications where spatial information is limited. Such applications include the recognition of small objects or the recognition of small features on partially occluded objects. We introduce a feature matrix representation for deterministic local structure in color images. Although feature matrices are useful for recognition, this representation depends on the spectral properties of the scene illumination. Using a linear model for surface spectral reflectance with the same number of parameters as the number of color bands, we show that changes in the spectral content of the illumination correspond to linear transformations of the feature matrices, and that image plane rotations correspond to circular shifts of the matrices. From these relationships, we derive an algorithm for the recognition of local surface structure which is invariant to these scene transformations. We demonstrate the algorithm with a series of experiments on images of real objects.

A Computational Approach to Color Illusions

Abstract: Tri-stimulus theory of color perception is not able to justify effectively some well known perception phenomena as color illusions and color constancy. Retinex theory, by Land and McCann, grounds color perception on a color space based on three lightness computed as relative reflectance along multiple exploration paths of the perceived scene. This paper considers in a new light Retinex theory, as a theory which tries to justify not only color constancy but also illusions arising from simultaneous contrast configurations. An improvement to Retinex computational model is presented in the paper, which selects Retinex computation paths by approximating a brownian path. The algorithm has been tested not only on traditional Mondrian patches, but also on natural pictures and photographs and on typical color illusion patches. The examples demonstrate the ability of the model to emulate human color perception behavior.

Visual learning and object verification with illumination invariance

Abstract: This paper describes a method for recognizing partially occluded objects to realize a bin-picking task under different levels of illumination brightness by using the eigenspace analysis. In the proposed method, a measured color in the RGB color space is transformed into the HSV color space. Then, the hue of the measured color, which is invariant to change in illumination brightness and direction, is used for recognizing multiple objects under different levels of illumination conditions. The proposed method was applied to real images of multiple objects under different illumination conditions, and the objects were recognized and localized successfully.

Role of remote adaptation in perceived subjective color.

Abstract: Light adaptation to illumination that is presented peripherally changes the subjective color of a central Benham disk stimulus. In our experiments we kept the peripheral illumination achromatic and remote (not even adjacent to the test stimulus). Using a high-frame-rate monitor, we produced the subjective color stimulus, to our knowledge for the first time, on a computer screen in emulation of the Benham disk programs. The resulting changes in the perceived subjective color were as follows: (1) Remote adapting illumination caused a dramatic shift in the perceived subjective color with a span from red to green; (2) there was a trade-off dependence between the area and the intensity of the remote adapting illumination with respect to the perceived color of the test stimulus; (3) the effect of the remote adaptation showed no interocular interaction. This finding suggests that the effect is elicited from a low-level stage in the visual pathway. In addition, we were able to approximate experimentally the spatial profile of the contribution of the remote illumination through the shift in the perceived color. We also found an opposite general trend of color shifts that occurred when either the central stimulus luminance or the remote illumination was increased. A suggested model for the reversed color shifts trend is discussed.

Illuminant Invariant Recovery of Surface Reflectances and Estimation of Colorimetric Values from Image Signals

Abstract: Computer simulation based on the finite-dimensional linear model for computational color constancy model has been carried out to estimate colorimetric values using image signals. We estimated the colorimetric values placing a reference object with known spectral reflectance in a scene under an unknown illuminant. It was found that prior knowledge of illuminants and surfaces of objects in a scene is not required for the estimation: any set of basis vectors to express the illuminants and any set of eigen vectors obtained by principal component analysis on data set of a color chart to represent the surfaces of different objects can be applied to the estimation.

Color opponency and discrimination

Abstract: The rationale for color opponency is investigated. A principle of optimal discrimination of visual information is introduced: If two visual variables are kept fixed, the third one, left free, must vary as much as possible. We prove that, if satisfied, the principle requires the existence of two chromatic functions, which are under the reference source E (corresponding to the equal-energy spectral distribution function), orthogonal with respect to a certain inner product, so that given one function, the other one is determined, apart from a multiplicative constant. Experimental chromatic functions are recovered. Two almost-illuminant-independent chromatic variables, η and ζ, are constructed from the achromatic variables in the companion paper.

Color Linear Model

Abstract: In this paper, procedures for creating an effective linear model to represent surface spectra are presented. The model is derived by considering spectral data and the human visual characteristic that depends on wave lengths. Two human visual weighting functions (HVWF) are derived from human visual characteristic. The basis functions of the linear model for the surface reflectance are selected by minimizing least square error in approximating the spectral data weighted by the HVWF. The linear model is shown to perform better than conventional linear models for color constancy, the surface identification related to object recognition, and the characterization of a scanner and a camera.

Non-supervised chromatic illuminant: corrector for autonomous robots

Abstract: One of the well-known problems in colour image interpretation is the colour-constancy problem. Autonomous robots that use colour information to select objects or landmarks can be deceived in presence of heavy coloured illuminants. Classic chromatic filtering presupposes detailed information about light source characteristics, but this is not always possible. The presence of emergency lights or different kinds of light sources can heavily influence object colour. Retinex theory, by Land and McCann (1971), can resolve these problems. This theory gives color perception on a color space based on three brightness computed as relative reflectance along multiple exploration paths of the perceived scene. This paper considers the application of this theory in order to allow automatic colour detection in autonomous robots. The algorithm has been tested on simple coloured scenes illuminated with different light sources. The results obtained are compared.

A method and apparatus for improving a digital image

Abstract: not available for EP0901671Abstract of corresponding document: US5991456A method of improving a digital image is provided. The image is initially represented by digital data indexed to represent positions on a display. The digital data is indicative of an intensity value Ii(x,y) for each position (x,y) in each i-th spectral band. The intensity value for each position in each i-th spectral band is adjusted to generate an adjusted intensity value for each position in each i-th spectral band in accordance with where S is the number of unique spectral bands included in said digital data, Wn is a weighting factor and "*" denotes the convolution operator. Each surround function Fn(x,y) is uniquely scaled to improve an aspect of the digital image, e.g., dynamic range compression, color constancy, and lightness rendition. The adjusted intensity value for each position in each i-th spectral band is filtered with a common function and then presented to a display device. For color images, a novel color restoration step is added to give the image true-to-life color that closely matches human observation.

Color Constancy for Scenes with Varying Illumination (Color Constancy with Varying Illumination)

Abstract: We present an algorithm which uses information from both surface reflectance and illumination variation to solve for color constancy. Most color constancy algorithms assume that the illumination across a scene is constant, but this is very often not valid for real images. The method presented in this work identifies and removes the illumination variation, and in addition uses the variation to constrain the solution. The constraint is applied conjunctively to constraints found from surface reflectances. Thus the algorithm can provide good color constancy when there is sufficient variation in surface reflectances, or sufficient illumination variation, or a combination of both. We present the results of running the algorithm on several real scenes, and the results are very encouraging.

Shedding Some Light on the Blue Vein Enigma

Abstract: Alex Vitkin takes a look at an interesting visual mystery—why veins appear blue. He finds his answer in retinex theory.

Spalvų suvokimo konstantiškumas: empirinis tyrimas ir matematinis modelis

Abstract: The method to estimate the color constancy is suggested. The experimental results confirm the hypothesis of color constancy linear model. The linear operator calculated from the chromaticity coordinates of three chips under two different illuminations. The analysis of this linear operator is performed to clear up the physiological mechanism.

Color constancy—color recovery from biased illumination

Abstract: In color vision, it is of great importance to steadily extract color descriptors under various illuminative conditions. This is called color constancy. In general, the reflected light of an object is dependent on two components which are the illumination of the environment and the surface reflectance of the object. If we don't have enough information about the scene, we could not correctly say the color that we see is real or biased by the ambient light. Thus, all we can do is to feel it in global view. To estimate the surface reflectance even when the spectral distribution of the ambient light is unknown, an algorithm consisting of a finite-dimensional model, a homomorphic model, a statistical model, and a recovery model, is proposed. Some experiments conducted under different illuminative conditions confirm that the proposed algorithm is feasible.

Color image normalization through illuminant recovery

Abstract: The information in a color image is always a function of the illuminating source, the geometry, the reflectance properties of the object and the characteristic of the camera. Separating the influence of the spectral distribution of the illumination and the reflectance properties of the object is known as the color constancy problem. Successful separation is important for vision and pattern recognition tasks, quality control in the graphic arts and image database applications. We describe an approach to the color constancy problem which is based on statistical assumptions about the distribution of colors. It uses the eigenvector system of the logarithmic spectra in a large database of color samples and employs methods from robust statistics to recover the illumination spectrum. We illustrate the performance of the algorithm with a simulation in which the effect of the illumination by the standard A-source is eliminated.