Saliency‐based initialisation of Gaussian mixture models for fully‐automatic object segmentation

In meetings just prior to the 1997 AIC Congress in Kyoto, CIE TC1-37, chaired by M. Fairchild, established the CIE 1997 Interim Colour appearance Model (Simple Version), known as CIECAM97s. CIECAM97s was formally published in 1998 in CIE publication 131. CIE TC1-37 was dissolved shortly after publication of CIECAM97s at which time, a reportership, R1- 24 held by M. Fairchild, was established to monitor ongoing developments in color appearance modeling and notify CIE Division 1 if it became necessary to form a new TC to consider revision or replacement of CIECAM97s. In the four years between AIC Congresses, there has been much activity, both by individual researchers and within the CIE, aimed at furthering our understanding of color appearance models and deriving improved models for consideration. The aim of this paper is to summarize these activities, report on the current status of CIE efforts on color appearance models, and suggest what the future might hold for CIE color appearance models.

A comprehensive colour appearance model should be capable of predicting a wide range of viewing conditions including related and unrelated colours under a wide range of illuminance levels from mesopic to photopic region. In the past, colour appearance models can only predict the colour appearance for related colours in different viewing conditions. More recently, colour appearance models to predict unrelated colours have been developed and being improved constantly. This paper investigates the performance of different colour appearance models for unrelated colours. The three colour appearance modelswere compared including CAM15u, CAMFu and the latest CAM20u. An experimental dataset carefully accumulated by Fu et al. was used for evaluation. The results showed that the latest CAM20u performed best among the three models. CAMFu, as the oldest model, whose performance was much worse than other two models in predicting brightness and colourfulness, and performed well in predicting hue composition, while CAM15u gave a moderate performance in predicting brightness and colourfulness, but performed a little worse than the others for hue composition.

Tourmaline minerals are mainly produced in pegmatite, which is mainly produced in pegmatite.At present, the color evaluation of green tourmaline in China needs to be supplemented. In this paper, the CIECAM16 color appearance model is used to calculate the color appearance attribute parameters of green tourmaline samples, and the tristimulus XYZ of green tourmaline color measured by Colori5 colorimeter is used as the input data of CIECAM16 color appearance model to predict the color appearance attribute of green tourmaline under different conditions. The CIECAM16 color appearance model written by computer C++ language is used to calculate the color appearance attribute parameters under different illuminance levels under D65 light source, different environment (Average, Dim, Dark) conditions and different illumination light source color temperatures (D65, D50, A). Through analysis and calculation of the color appearance attribute parameters, it is found that that the increase of illuminance has a more intuitive effect on the visual brightness Q, while the change of ambient light and shade has a greater effect on the brightness than the change of light source. The A light source has a more obvious effect on the tourmaline of color than D65 and D50 light sources.

Some of the reasons for needing a color appearance model in imaging are given and the use of such a model in deriving a color reproduction index is described. There are three reasons why a model of color appearance is a necessary tool in color imaging systems. First, because the viewing conditions for images are often different at different stages of reproduction systems, it is necessary to allow for their visual effects; conventional colorimetry does not do this, but a model of color appearance can supply the necessary adjustments. Second, different display media typically have different color gamuts and a regime of gamut mapping is therefore often required; it is more appropriate to conduct this mapping in an appearance space. Third, it is desirable to be able to assess the quality of color reproductions in terms of a color reproduction index and a model of color appearance can provide a metric that is suitable as the basis of such an index.

Regarding the ocean background, the painted camouflage on the surface of targets effectively conceals them, providing a strategic defense against military reconnaissance. Accurately predicting the color appearance of both the target and background is crucial for assessing the camouflage’s performance. This study investigates the perceived color attributes (lightness, chroma, and hue) of marine targets against varying ocean backgrounds using memory matching assessments. Visual data reveal that background color slightly influences target hue perception through simultaneous contrast effects, while lightness and chroma remain unaffected. As the image-based color appearance model – iCAM offers a potential solution for extracting color appearance attributes from a complex RGB image, it has been adopted to predict the pixel-wise color appearance of the marine targets. Comparative validation shows iCAM substantially outperforms existing color appearance models in marine environments, establishing its superior capability for naval camouflage assessment and design. Further analysis indicates that chroma predictions are sensitive to changes in the filter kernel size of iCAM and the image size, while lightness and hue predictions remain stable. Notably, undersized kernels systematically underestimate perceived chroma, highlighting the critical importance of parameter optimization.

Color-normal subjects sometimes disagree about metameric matches involving highly structured SPDs, because their cone fundamentals differ slightly, but non-negligibly. This has significant implications for the design of light sources and displays, so it should be estimated. We propose a broadly applicable estimation method based on a simple adaptive "front-end" interface that can be used with any selected standard color appearance model. The interface accepts, as input, any set of color matching functions for the individual subject (for example, these could be that person's cone response functions) and also the associated tristimulus values for the test stimulus and also for the reference stimulus (i.e. reference white). The interface converts this data into tristimulus values of the form used by the selected color appearance model (which could, for example, be X, Y, Z), while also carrying out the needed transform, which is based on an estimate of the subject's likely previous long-term adaptations to their unique cone fundamentals. The selected standard color appearance model then provides color appearance data that is an estimate of the color appearance of the test stimulus, for that individual subject. This information has the advantage of being interpretable within that model's well-known color space. The adaptive front end is based on the fact that, for any selected input SPD and the subject's unique color matching functions, there can be many different SPDs that are metameric for that individual. Since observer-to-observer color perception differences are minimized for spectrally smooth SPDs, smooth metamers predict color appearances reasonably accurately.

Detecting colour differences in warp‐knitting fabric is essential to ensure its quality. Machine vision algorithms are commonly used for automatic colour difference detection. However, the current algorithm directly extracts RGB (red, green, blue) colour stimulus values from the image without considering the image's colour attributes, leading to a significant error in colour difference detection. To solve this issue, a colour appearance model is introduced into the field of colour difference detection for warp‐knitted fabric. Initially, a colour appearance model based on CAM16 is established, and the critical parameters are calculated to obtain the colour appearance attribute of the fabric. The colour difference calculation formula is then constructed in the uniform colour space of CAM16‐UCS. A template selection algorithm based on principal component analysis was designed to select warp‐knitted cloth images with standard colours. Subsequently, a cloth colour difference detection algorithm was developed using the cloth colour profile model. The performance of the colour difference formula based on the colour profile model was evaluated using the PF/3 method. To compare the CIELab, CMC(2:1), and CIEDE2000 colour difference formulas, standardised residual sum of squares was used. The results indicated that the colour difference formula based on the colour‐appearance model is about 5.32% different from the visual colour difference perceived by the human eye. However, it can perform as well as the CMC(2:1) colour difference formula, which is widely used in the textile industry.

Application of the CIECAM02 colour appearance model to predict the effect of gamma on the colours viewed on CRT monitors

In this study, Swedish Natural Color System (NCS) unique hue data were used to evaluate the performance of unique hue predictions by the CIECAM02 colour appearance model. The colour appearance of 108 NCS unique hue stimuli was predicted using CIECAM02, and their distributions were represented in a CIECAM02 ac–bc chromatic diagram. The best‐fitting line for each of the four unique hues was found using orthogonal distance regression in the ac–bc chromatic diagram. Comparison of these predicted unique hue lines (based on the NCS data) with the default unique hue loci in CIECAM02 showed that there were significant differences in both unique yellow (UY) and unique blue (UB). The same tendency was found for hue uniformity: hue uniformity is worse for UY and UB stimuli in comparison with unique red (UR) and unique green (UG). A comparison between NCS unique hue stimuli and another set of unique hue stimuli (obtained on a calibrated cathode ray tube) was conducted in CIECAM02 to investigate possible media differences that might affect unique hue predictions. Data for UY and UB are in very good agreement; largest deviations were found for UR. © 2014 Wiley Periodicals, Inc. Col Res Appl, 40, 256–263, 2015

Color appearance of images on self-luminous displays, such as LCD monitors, depends on two sources of illumination: scene illumination within the image itself and ambient illumination in the viewing environment. Color appearance models (CAMs) attempt to predict how perception changes as ambient illumination changes. At their best they provide viewers with an illumination-independent viewing experience. However, existing CAMs are complex and application-specific. Thus they are poorly suited to adaptive displays. For example, existing adaptive displays provide only achromatic adaptation, limiting the complexity of CAMs and minimizing tuning. We propose a new model based on von Kries adaptation, the theory on which most CAMs are based upon. It takes into account the mixed adaptation state produced by self-luminous displays. Compared to existing CAMs, our model is simpler and more efficient, yet shares the generic features of more complex CAMs.

Fifty-four spectral reflectances of skin colors (facial pattern of Japanese women) were measured and analyzed by the principal component analysis. The results indicate that the reflectance spectra can be estimated approximately 99% by only three principal components. Based on the experimental results, it is shown that the spectral reflectance of all pixels in skin can be calculated from the R, G, B signals of HDTV (High Definition Television) camera.Computer simulations of color reproduction in skin color have been developed in both colorimetric color reproduction and color appearance models by von Kries, LAB and Fairchild. The obtained color reproduction in facial pattern and skin color patch under different illuminants are analyzed. Those reproduced color images were compared by memory matching method. The difference of appearance models is described and the difference of appearance between skin color patch and facial pattern is also discussed and analyzed.

Many successful models for predicting attention in a scene involve three main steps: convolution with a set of filters, a center-surround mechanism and spatial pooling to construct a saliency map. However, integrating spatial information and justifying the choice of various parameter values remain open problems. In this paper we show that an efficient model of color appearance in human vision, which contains a principled selection of parameters as well as an innate spatial pooling mechanism, can be generalized to obtain a saliency model that outperforms state-of-the-art models. Scale integration is achieved by an inverse wavelet transform over the set of scale-weighted center-surround responses. The scale-weighting function (termed ECSF) has been optimized to better replicate psychophysical data on color appearance, and the appropriate sizes of the center-surround inhibition windows have been determined by training a Gaussian Mixture Model on eye-fixation data, thus avoiding ad-hoc parameter selection. Additionally, we conclude that the extension of a color appearance model to saliency estimation adds to the evidence for a common low-level visual front-end for different visual tasks.

In 1879, in post-civil-war America, Henry George published a very popular, and still referenced, book entitled “Progress and Poverty: An Inquiry into the Cause of Industrial Depressions and of Increase of Want with Increase of Wealth: The Remedy.” The title of this paper is an obvious play off of George's title. George suggested that the so-called “Gilded Age”, which was supposed to be a time of economic growth and reform for everyone in the nation, was not for many as the increase in wealth for some was accompanied by increased poverty for many others. George also had several creative solutions to these problems. When looking over the research and application of color appearance models (and fundamental colorimetry) across the 20 years of Color (and) Imaging Conferences, it becomes clear that there are also cycles of increased progress (wealth) followed by increased poverty (want) that somewhat parallel George's main thesis. In general, when new capabilities are developed, their shortcomings are quickly and sometimes vociferously elucidated and it can appear that the want of the shortcomings outweighs the wealth of the advances. In some situations that is indeed the case. This paper traces some examples of that path and reaches a destination describing one view of the status quo of color science.

Two daylight-balanced colour negative films, one professional and one amateur product, have been exposed to a standard subject using approximately Planckian light sources with a wide range of colour temperatures. Prints have been made to achieve a standard grey reproduction from each negative and have also been prepared using a commercial automatic printer calibrated with the negative, in each set, exposed using professional electronic flash. Subjective investigations of print acceptability are used to investigate the influence of the different spectral sensitivity distributions of the two negative films. Print acceptability is also related to a Colour Reproduction Index calculated from the standard subject, using the Hunt colour appearance model. Conclusions are drawn concerning the applicability of different subjective techniques in determining print acceptability, the effect of negative sensitivities and the applicability of the Hunt colour appearance model to problems of print acceptability.

Recognition of traffic signs based on their colour and shape features extracted using human vision models