Colour‐imager characterization by parametric fitting of sensor responses

This study describes a novel method for characterizing the colorimetric and photometric properties of three‐channel color imaging devices. The method is designed to overcome some undocumented aspects of the imager‐characterization problem: The effective spectral sensitivity profiles of the imager's color channels depend on the level of radiant input energy, and these profiles must be known in order to determine the true intensity‐response characteristics of the three channels. By fitting the response distributions of the three color channels explicitly with low‐dimensional models, the method takes these dependencies into account, and may, therefore, offer several advantages over other imager‐characterization methodologies, particularly where illuminant‐independent characterization is required. An application of the technique is detailed, in which a CCD camera is characterized using only the Macbeth ColorChecker and a number of artificial illuminants. © 2001 John Wiley & Sons, Inc. Col Res Appl, 26, 442–449, 2001     

Estimating spectral reflectance from camera responses based on CIE XYZ tristimulus values under multi‐illuminants

A new spectral reflectance estimation method based on CIE XYZ values under multi‐illuminants was proposed to obtain multi‐spectral images accurately by using digital still cameras. CIE XYZ values under multi‐illuminants were initially predicted from raw RGB responses by using a polynomial model with local training samples. Then, spectral reflectance was constructed from the predicted CIE XYZ values via the pseudo‐inverse method. Experimental results indicated that the new spectral reflectance estimation method significantly outperformed the traditional colorimetric characterization method without requiring extra training samples or greatly increasing computational complexities. © 2016 Wiley Periodicals, Inc. Col Res Appl, 42, 68–77, 2017     

Dictionary‐based estimation of spectra for wide‐gamut color imaging

With the widespread use of commercialized wide‐gamut displays, the demand for wide‐gamut image content is increasing. To acquire wide‐gamut image content using camera systems, color information should be accurately reconstructed from recorded image signals for a wide range of colors. However, it is difficult to obtain color information accurately, especially for saturated colors, if conventional color cameras are used. Spectrum‐based color image reproduction can solve this problem; however, bulky spectral imaging systems are required for this purpose. To acquire spectral images more conveniently, a new spectral imaging scheme has been proposed that uses two types of data: high spatial‐resolution red, green, and blue (RGB) images and low spatial‐resolution spectral data measured from the same scene. Although this method estimates spectral images with high overall accuracy, the error becomes relatively large when multiple different colors, especially those with high saturation, are arranged in a small region. The main reason for this error is that the spectral data are utilized as low‐order spectral statistics of local spectra in this method. To solve this problem, in this study, a nonlinear estimation method based on sparse and redundant dictionaries was used for spectral image estimation—where the dictionary contains a number of spectra—without loss of information from the low spatial‐resolution spectral data. The estimated spectra are represented by a mixture of a few spectra included in the dictionary. Therefore, the respective feature of every spectrum is expected to be preserved in the estimation, and the color saturation is also preserved for any region. Experiments performed using the simulated data showed that the dictionary‐based estimation can be used to obtain saturated colors accurately, even when multiple colors are arranged in a small region. © 2011 Wiley Periodicals, Inc. Col Res Appl, 2013     

Spectral color reproduction minimizing spectral and perceptual color differences

In this article, we are combining minimization criteria in the colorant separation process for spectral color reproduction. The colorant separation is performed by inverting a spectral printer model: the spectral Yule‐Nielsen modified Neugebauer model. The inversion of the spectral printer model is an optimization operation in which a criterion is minimized at each iteration. The approach we proposed minimizes a criterion defined by the weighted sum of a spectral difference and a perceptual color difference. The weights can be tuned with a parameter α ∞ [0, 1]. Our goal is to decrease the spectral difference between the original data and its reproduction and also to consider perceptual color difference under different illuminant conditions. In order to find the best α value, we initially compare a pure colorimetric criterion and a pure spectral criterion for the reproduction, then we combine them. We perform four colorant separations: the first separation will minimize the 1976 CIELAB color difference where four illuminants are tested, the second separation will minimize an equally weighted summation of 1976 CIELAB color difference with the four illuminants tested independently, the third colorant separation will minimize a spectral difference, and the fourth colorant separation will combine a weighted sum of a spectral difference and one of the two first colorimetric differences previously introduced. This last colorant separation can be tuned with a parameter in order to emphasize on spectral or colorimetric difference. We use a six colorants printer with artificial inks for our experiments. The prints are simulated by the spectral Yule‐Nielsen modified Neugebauer model. Two groups of data are used for our experiments. The first group describes the data printed by our printing system, which is represented by a regular grid in colorant space of the printer and the second group describes the data which is not originally produced by our printing system but mapped to the spectral printer gamut. The Esser test chart and the Macbeth Color Checker test chart have been selected for the second group. Spectral gamut mapping of this data is carried out before performing colorant separation. Our results show improvement for the colorant separations combining a sum of 1976 CIELAB color difference for a set of illuminants and for the colorant separation combining a sum of 1976 CIELAB color difference and spectral difference, especially in the case of spectral data originally produced by the printer. © 2008 Wiley Periodicals, Inc. Col Res Appl, 33, 494–504, 2008     

Spectral‐based color separation using linear regression iteration

The present article will introduce a very simple new method for spectral‐based color separation. This method inverts a Yule–Nielsen modified spectral Neugebauer model, utilizing its affine multilinearity in the 1/n‐space. By means of linear regression, a sequence of colorant combinations is constructed converging to a colorant combination that approximates the desired reflectance spectrum in the sense of the smallest RMS error. Each iteration step consists mainly of two simple matrix–vector multiplications. With the aid of various simulation experiments, investigations on the speed of convergence are conducted. © 2006 Wiley Periodicals, Inc. Col Res Appl, 31, 229–239, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/col.20211     

Irradiance‐independent camera color calibration

For a digital color camera to represent the colors in the environment accurately, it is necessary to calibrate the camera RGB outputs in terms of a colorimetric space such as the CIEXYZ or sRGB. Assuming that the camera response is a linear function of scene luminance, the main step in the calibration is to determine a transformation matrix M mapping data from linear camera RGB to XYZ. Determining M is usually done by photographing a calibrated target, often a color checker, and then performing a least‐squares regression on the difference between the camera's RGB digital counts from each color checker patch and their corresponding true XYZ values. To measure accurately the XYZ coordinates for each patch, either a completely uniform lighting field is required, which can be hard to accomplish, or a measurement of the illuminant irradiance at each patch is needed. In this article, two computational methods are presented for camera color calibration that require only that the relative spectral power distribution of the illumination be constant across the color checker, while its irradiance may vary, and yet resolve for a color correction matrix that remains unaffected by any irradiance variation that may be present. © 2013 Wiley Periodicals, Inc. Col Res Appl, 39, 540–548, 2014     

Recovering neugebauer colorant reflectances and Ink‐spreading curves from printed color images

Spectral reflection prediction models, although effective, are impractical for certain industrial applications such as self‐calibrating devices and online monitoring because their calibration requires specific color‐constant calibration patches. Using the CMYK Ink‐Spreading enhanced Yule‐Nielsen‐modified Spectral Neugebauer model (IS‐YNSN), we propose a method to recover the colorant reflectances (Neugebauer primaries), the ink‐spreading curves, and the Yule‐Nielsen n‐value using only tiles extracted from printed color images. There is no prior knowledge about the reproduction device. Thanks to a set of constraints based on principal component analysis and the relationships between composed Neugebauer primaries and the ink transmittances, good approximations of the Neugebauer primaries are achieved. These approximations are then optimized, yielding an accurately calibrated IS‐YNSN model comparable to the one obtained by classical calibrations. © 2013 Wiley Periodicals, Inc. Col Res Appl, 39, 216–233, 2014; Published online 23 February 2013 in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/col.21800     

Physics‐based spectral sharpening through filter‐chart calibration

The spectral overlap of color‐sampling filters increases errors when using a diagonal matrix transform, for color correction and reduces color distinction. Spectral sharpening is a transformation of colors that was introduced to reduce color‐constancy errors when the colors are collected through spectrally overlapping filters. The earlier color‐constancy methods improved color precision when the illuminant color is changed, but they overlooked the color distinction. In this article, we introduce a new spectral sharpening technique that has a good compromise of color precision and distinction, based on real physical constraints. The spectral overlap is measured through observing a gray reference chart with a set of real and spectrally disjoint filters selected by the user. The new sharpening method enables to sharpen colors obtained by a sensor without knowing the camera response functions. Experiments with real images showed that the colors sharpened by the new method have good levels of color precision and distinction as well. The color‐constancy performance is compared with the data‐based sharpening method in terms of both precision and distinction. © 2014 Wiley Periodicals, Inc. Col Res Appl, 40, 564–576, 2015     

Applying image‐based color palette for achieving high image quality of displays

In the highly competitive display market, manufacturers continuously develop new technologies to improve the image quality of displays. However, color measurement and visual assessment are time‐consuming to production lines. A new method to measure and improve color quality of the displays automatically therefore, is urgently needed to the manufacturers. This article proposes a familiar color correction strategy to optimize the colors of different displays by means of creating an image‐based color palette which enables color correction for familiar objects (e.g., facial skin, blue sky, or green grass) in the multidisplay systems. To produce the image‐based color palette, the 8‐bit RGB value of each pixel in an image is transformed to L*d*n* (lightness/dominant color/nondominant color) color channels, and the dominant‐color regions in an image are subsequently extracted from the dominant color (d*) channel. The memory color data of familiar objects can be set in reference monitor in advance to determine the dominant color (d*) channel. Then a series of palette colors are generated around a displayed image. The color palette will be displayed as a target for two‐dimensional colorimeter shooting to obtain the measured color data. The familiar color correction model was established based on a first‐order polynomial regression to achieve a polynomial fit between the measured color data and the reference color data on the color palette. The proposed method provides a solution to correct familiar colors on a displayed image, and maintains the original color gamut and tone characteristic in the multidisplay systems simultaneously. It is possible to achieve the preferred intent of the displayed images by using the proposed familiar color correction method. © 2012 Wiley Periodicals, Inc. Col Res Appl, 39, 154–168, 2014     

Investigation of metallic color perception using real‐world materials

Metallic colors have a unique appearance of glossiness with features such as highlights, contrast, and reflections on their surface, and therefore, metallic objects are very attractive to humans. Especially, gold, silver, and copper colors are familiar metals used as decorative materials, coins, and other furnishings. However, the mechanism and condition of metallic perception have not been fully investigated. There are a few studies for investigating metallic perception using rendered patches or images, but there is no study using real‐world objects. In our previous study, we developed a simple representation technique that made real objects appear to be made of gold by projecting a solid color onto a target nonmetallic object. By using the representation technique, in this study, we have further investigated the perception of metallic appearance such as gold, silver, and copper using real‐world materials, and analyzed the difference between these metallic perceptions. Our results indicate that the perception of the metallic object is different for gold, silver, and copper. Our new findings are as follows: the glitter required for the perception of gold and silver becomes an obstacle to the perception of copper; the metallic perception reveals that learning experience might be strongly affecting; and luminance adjustment is sensitive to the perception of metallic objects.     

Near‐lossless compression methods for spectral images

In this work, several near‐lossless compression methods for spectral images have been analyzed and compared. These methods are based both on the principal component analysis (PCA) and on the choice of a minimum number of spectral points, selected following different criteria. The analysis, initially carried out on 14 National Physical Laboratory tiles of certified colour, has been extended to some spectral images of paintings taken at the National Gallery of Parma (Italy). The comparison of the results with those obtained by applying the PCA analysis shows that the best method indicated as “method of a few significant points” allows reducing the spectral image size of a factor of 10 without loss of spectral and colour information. © 2010 Wiley Periodicals, Inc. Col Res Appl, 2010     

Media‐dependent color appearance modeling based on artificial neural networks

In this study, we tried to consider various color appearance factors and device characterization together by visual experiment to simplify the across‐media color appearance reproduction. Two media, CRT display (soft‐copy) and NCS color atlas (hard‐copy), were used in our study. A total of 506 sample pairs of RGB and HVC, which are the attributes of NCS color chips, were obtained according to psychophysical experiments by matching soft copy and hard copy by a panel of nine observers. In addition, a set of error back‐propagation neural networks was used to realize experimental data generalization. In order to get a more perfect generalizing effect, the whole samples were divided into four parts according to different hues and the conversion between HVC and R_HVCG_HVCB_HVC color space was implemented. The current results show that the displays on the CRT and the color chips can match well. In this way, a CRT‐dependent reproduction modeling based on neural networks was formed, which has strong practicability and can be applied in many aspects. © 2006 Wiley Periodicals, Inc. Col Res Appl, 31, 218–228, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/col.20209     

Spectral‐reflectance‐factor deconvolution and colorimetric calculations by local‐power expansion

The experimental data of the spectral‐reflectance factor are considered as dependent on the instrument‐spectral‐bandwidth function in order to perform their deconvolution and to compute the tristimulus values. The deconvolution is performed by local‐power expansion. In the case that the spectral‐bandpass dependence regards only the spectral transmittance of the monochromator, the goodness of this technique is evaluated by simulation (1325 reflectance factors of the Munsell samples are considered as trial functions) and compared with other usual techniques: Stearns and Stearns method for bandpass error, ASTM‐weighting function interpolation, and Venable‐ASTM weighting function. The zero order of the deconvoluted spectral‐reflectance factor can be related to the Stearns and Stearns method for bandpass error. With respect to any other technique, the second‐order deconvolution, for the CIE standard illuminants, gives color differences lower by a factor 0.1 or more for a bandpass Δλ = 10 nm, color differences lower by a factor 0.3 or more for a bandpass Δλ = 20 nm and, for the CIE fluorescent illuminants, color differences generally lower. © 2000 John Wiley & Sons, Inc. Col Res Appl, 25, 176–185, 2000     

Assessing color performance of whole‐slide imaging scanners for digital pathology

The color performance of two commercial whole‐slide imaging (WSI) scanners was compared against the ground truth and a hypothetical monochrome scanner. Three biological tissue slides were used to test the WSI scanners. A multispectral imaging system was developed to obtain the color truth of the biological tissue slides at the pixel level. The hypothetical monochrome scanner was derived from the color truth as a lower bound for comparison. The CIEDE2000 formula was used to measure color errors. Results show that color errors generated by the modern commercial WSI scanner, the legacy commercial WSI scanner, and the monochrome WSI scanner are in the range of [8.4, 13.0], [18.0, 26.33], and [17.4, 17.6] ΔE₀₀, respectively. The legacy commercial WSI scanner was outperformed by not only the modern commercial WSI scanner but also by the hypothetical monochrome scanner.     

Comparison of different monoptic and dichoptic color‐matching functions

The objective of this article is to analyze different color matching functions (CMFs) obtained with three (650, 530, and 460 nm) and four primary colors (650, 565, 513, and 460 nm), using both monoptic and dichoptic central vision. This strategy helps to clarify (i) lack of additivity of brilliance; (ii) shift in maximum sensitivity peaks of CMFs when experimental conditions change; (iii) variations in luminance for the same reason; (iv) strong metamerism of the mixtures; and (v) differences of chromatic opponence between monoptic and dichoptic vision. The results obtained reflect two important facts: marked stability of the visual system, which allows the experimental conditions analyzed to be solved with an equal degree of success, and plasticity based especially on the balance of retinal illumination, which was maintained at an average of 40 trolands. The results obtained bring to mind an assertion made by MacAdam to the effect that the law of additivity when applied to luminance is not applied to measurements of brightness. Perceptively, brightness is not additive, and so CMFs should not be considered as significant functions in computing tristimulus values R, G, and B. © 2005 Wiley Periodicals, Inc. Col Res Appl, 30, 416–426, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/col.     

A non‐PC look at principal components

Two principal‐component methods are used in color science. For a given data set of spectra, one method finds the best‐fitting subspace about the mean spectrum, and the other finds the best‐fitting subspace about the zero spectrum. The first of these was originally developed for illuminants and the second for reflectance analysis. Yet there seems to be no strong argument for choosing one method over the other, in either case. Hence it is urged that each of us declares which one we are using, even if making that discrimination is considered “non‐PC” (i.e., not “politically correct”). © 2002 Wiley Periodicals, Inc. Col Res Appl, 28, 69–71, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/col.     

Camera characterization for color research

In this article we introduce a new method for estimating camera sensitivity functions from spectral power input and camera response data. We also show how the procedure can be extended to deal with camera nonlinearities. Linearization is an important part of camera characterization, and we argue that it is best to jointly fit the linearization and the sensor response functions. We compare our method with a number of others, both on synthetic data and for the characterization of a real camera. All data used in this study is available online at www.cs.sfu.ca/～colour/data . © 2002 Wiley Periodicals, Inc. Col Res Appl, 27, 152–163, 2002; Published online in Wiley Interscience (www.interscience.wiley.com). DOI 10.1002/col.10050     

Diffuse reflectance‐factor measurements of rose petals

The diffuse reflectance factor for different colored rose petals is measured as a function of wavelength using a high resolution optical spectrometer. The tristimulus values, the CIE chromaticity coordinates, the dominant wavelength and purity, the CIE whiteness index, the tint index, the CIE 1976 LAB coordinates, as well as CIELAB hue‐angle and chroma are reported. The data on diffuse reflectance factor are presented in the 390?800 nm range at intervals of 10 nm. Using the data, one can generate the perceived color of the roses and the color coordinates in different illuminating light sources and environments. The present data will be useful for the color characterization of flowers, realistic rendering of flowers in computer graphics, color photography, and in the development of filters for color photography. © 2010 Wiley Periodicals, Inc. Col Res Appl, 2011