CIELAB Color Space Research Articles

Rotational Invariance Using Gabor Convolution Neural Network and Color Space for Image Processing

Convolutional neural networks (CNNs) are deep learning methods that are utilized in image processing such as image classification and recognition. It has achieved excellent results in various sectors; however, it still lacks rotation invariant and spatial information. To establish whether two images are rotational versions of one other, one can rotate them exhaustively to see if they compare favorably at some angle. Due to the failure of current algorithms to rotate images and provide spatial information, the study proposes to transform color spaces and use the Gabor filter to address the issue. To gather spatial information, the HSV and CieLab color spaces are used, and Gabor is used to orient images at various orientation. The experiments show that HSV and CieLab color spaces and Gabor convolutional neural network (GCNN) improves image retrieval with an accuracy of 98.72% and 98.67% on the CIFAR-10 dataset.

Effects of using composite flour containing wheat flour with different levels of green banana pulp flour on the quality of saj flatbread

There is an increasing trend in formulating food to contain dietary fibers and particularly resistant starch. Saj bread (a type of flatbread baked on a plate placed directly on fire) is a potential candidate to act as a vehicle for delivering resistant starch. This study aimed to investigate the effects of using composite flour containing wheat flour substituted with different levels (0, 5, 10, 15, and 20%) of green banana pulp flour "GBPF" on some physicochemical properties of flour (moisture, ash, wet and dry gluten content, gluten index, falling number, and farinograph parameters) and the quality of saj bread as measured by CIELAB color space, texture (stretchability and texture profile analysis "TPA"), and sensory properties. The texture of the saj bread was monitored during three days of storage. Composite flour moisture content and falling number were unaffected by wheat flour substitution with GBPF, while dry gluten content decreased significantly for composite flour containing 15% or more GBPF. With increasing wheat flour substitution level with GBPF, dough stability decreased. For saj bread color, the L* and b* values decreased with increasing substitution levels, while a* and ∆E*ab values increased. With increasing substitution levels, the stretchability of bread decreased, and all tested TPA parameters increased. With increasing saj bread storage time, the stretchability of bread decreased, and all TPA parameters increased except cohesiveness which decreased. Using composite flour improved bread taste and odor scores and decreased color acceptability scores. Texture and overall acceptability scores were not affected. In conclusion, GBPF can potentially substitute up to 20% wheat flour without negatively affecting saj bread quality.

Improved Preprocessing Strategy under Different Obscure Weather Conditions for Augmenting Automatic License Plate Recognition

Automatic license plate recognition (ALPR) systems are widely used for various applications, including traffic control, law enforcement, and toll collection. However, the performance of ALPR systems is often compromised in challenging weather and lighting conditions. This research aims to improve the effectiveness of ALPR systems in foggy, low-light, and rainy weather conditions using a hybrid preprocessing methodology. The research proposes the combination of dark channel prior (DCP), non-local means denoising (NMD) technique, and adaptive histogram equalization (AHE) algorithms in CIELAB color space. And used the Python programming language comparisons for SSIM and PSNR performance. The results showed that this hybrid approach is not merely robust to a variety of challenging conditions, including challenging weather and lighting conditions but significantly more accurate for existing ALPR systems.

Quality of White-Brined Nabulsi Cheese Made with Different Proportions of Sheep’s and Cows’ Milk

Nabulsi cheese is the dominant white-brined cheese in Jordan and neighboring countries. Although this cheese is typically prepared using sheep’s milk, achieving large-scale production or attaining a specific quality involves the use of cows’ milk. Nabulsi cheese is kept in >18% brine for up to 1 year at room temperature (25 °C) and then soaked in water for desalting before eating. In this study, the effects of the milk type/combination used for production (100% sheep’s milk; 100% cows’ milk; 1:1, 3:1, or 1:3 sheep’s:cows’ milk mixture), desalting (salted vs. desalted cheese), and storage time (0, 6, and 12 months) on the cheese’s chemical composition directly after processing (in the salted form), texture (according to texture profile analysis “TPA”), and color (according to CIE lab color space) were assessed. The type or the combination of milk used does not affect the chemical composition of Nabulsi cheese, which was 43.35% ± 1.50 for moisture, 20.24% ± 0.59 for protein, 24.70% ± 1.53 for fat, 11.52% ± 0.60 for ash, 9.96% ± 0.44 for salt, and pH 6.43 ± 0.16. However, cheeses prepared using 100% sheep’s and 100% cows’ milk, respectively showed the highest (26.85%) and lowest (14.77%) yields. Desalting and milk type had the highest effects on cheese texture and color, respectively. Considering all tested parameters, mixing equal proportions of sheep’s and cows’ milk ensures that the Nabulsi cheese produced is similar to that produced from 100% sheep’s milk, albeit with increased cohesiveness (0.52 ± 0.07for cheese made from sheep’s milk and 0.59 ± 0.05 for 1:1 sheep’s: cows’ milk mixture).

Color and Chemical Composition of Timber Woods (Daniellia oliveri, Isoberlinia doka, Khaya senegalensis, and Pterocarpus erinaceus) from Different Locations in Southern Mali

Wood characteristics and properties are related to various factors connected to the biochemical processes that occur in the tree during wood formation, but also, to the interactions with the environmental conditions at the tree growing location. In addition to climatic factors, several investigations drew attention to the significance of the influence of other environmental parameters at the tree growing location. In this perspective, this work aimed to characterize the variation in color and chemical composition of timber wood from different locations in southern Mali, of trees growing under the same climatic conditions. To do so, a total of 68 grounded wood samples, from 4 timber wood species (Daniellia oliveri, Isoberlinia doka, Khaya senegalensis, and Pterocarpus erinaceus), were analyzed using CIELab color space and FTIR-ATR. Overall, the results indicated that the variation in wood color and chemical properties can be related to the local environmental conditions. Pterocarpus erinaceus presented significant differences between samples from the three areas according to the highest number of variables (color parameters, molecular composition determined by FTIR-ATR spectroscopy, and FTIR-ATR ratios). Daniellia oliveri and Khaya senegalensis, however, showed significant differences between areas of provenance for a lower number of variables. Isoberlinia doka, for its part, showed no significant differences and seems to be less sensitive to environmental factors. Furthermore, the results revealed that important differences exist between wood samples from Kati and Kéniéba.

Color change and thermochromism of sodalite, Na8[Al6Si6O24](I,S)2, synthesized by hydrothermal reaction

A sodalite material with a composition of Na8[Al6Si6O24](I,S)2 was synthesized from zeolite A via a hydrothermal reaction of a mixed solution of sodium iodide and sodium sulfide. The stoichiometric amount of NaI required to synthesize iodide sodalite (Na8[Al6Si6O24]I2) from zeolite A (Na12[Al12Si12O48]) was defined as a NaI content of 1.0, and synthesis was carried out using a NaI content of 0–25. The color of the obtained samples was almost white; however, the samples became blue or yellow following calcination. The blue color, which was attributed to the formation of S3− color centers similar to those in ultramarine, was obtained after calcination at 700–800 °C. However, samples calcined at 900 and 1000 °C were yellow. Raman spectroscopic analysis revealed that the formation of I2 molecules induced the color change to yellow. Among the prepared specimens, the yellow specimen prepared at 7.5 times the stoichiometric ratio of NaI exhibited the largest b* value in the CIELAB color space, which represents the blue–yellow component, with blue in the negative direction and yellow in the positive direction. No proportional relation was observed between the b* values and the amount of NaI added, suggesting that an appropriate I/S ratio existed for yellow coloration in sodalite. The yellow sample exhibited thermochromism with the absorption band broadening toward longer wavelengths and reddening at higher temperatures. This material has not been previously reported and is expected to have applications as a new host–guest material.

Lévy–Cauchy arithmetic optimization algorithm combined with rough K-means for image segmentation

Rough K-Means (RKM) is a well-known unsupervised clustering algorithm based on rough set logic that is utilized in a wide range of applications. However, when dealing with complex problems like image segmentation, it is frequently trapped in local optima during execution, resulting in undesirable segmentation results. To handle the issue in a realistic computing time, this study develops a Lévy–Cauchy Arithmetic Optimization Algorithm (LCAOA), an enhanced form of AOA, for performing rough clustering. The Levy flight and Cauchy distribution help in exploration and exploitation, respectively, in the proposed LCAOA. Therefore, well-balanced exploration and exploitation have been incorporated into LCAOA, which is a major problem of classical AOA. Opposition-based learning is also incorporated into LCAOA to maintain an efficient population during the optimization process. As the segmentation efficacy is somewhat dependent on the selection of color spaces caused by the non-illumination of regions, the suggested method employs the CIELab color space. The suggested method is compared to conventional and Nature-Inspired Optimization Algorithms (NIOA)-based state-of-the-art image segmentation techniques over traditional color images, color pathology images, and leaf images. The proposed clustering methodology outperforms all other examined clustering algorithms, according to the results of the experiments. For example, proposed LCAOA-based rough clustering gives average Feature Similarity Index (FSIM) values of 0.9513, 0.9688, and 0.9769 for traditional color images with 4, 6, and 8 clusters, respectively. The proposed technique is associated with an average FSIM value of 0.9525 for cluster number 2 in images of oral pathology. Lastly, for leaf images, the proposed approach yields a mean FSIM value of 0.9759 with an accuracy of greater than 97% for cluster number 2.

MACGRAN-IJ: AN EFFICIENT METHOD TO ANALYZE DIMENSIONS AND COLOR OF WHEAT GRAINS

The dimensions, shape, and color of wheat grain are important in plant breeding, although their manual measurement is difficult and time-consuming. While it is true that digital image analysis (DIA) with specific software has made it easier, some are semiautomatic, exclude color, and require the digitization of individual grains. The goal of this study was to compare the precision, accuracy, and efficiency of MacGran-IJ versus SmartGrain, GrainScan, and manual caliper measurement. The morphological characteristics of durum wheat (Triticum durum Desf.) and bread wheat (Triticum aestivum L.) grains were measured using DIA with MacGran-IJ, including color (in RGB and CIELab color spaces), size (mm2), lateral length (LLG, mm), ventral length (VLG, mm), lateral (LWG, mm) and ventral (VWG, mm) width, perimeter (mm), and circularity. Length, LWG, and VWG were measured with a digital caliper. The time (s) taken in each method was recorded. Size, LLG, VLG, and VWG were the same between methods for both species, but the LWG measurement obtained with vernier showed differences between the three programs (p ≤ 0.01), as well as low accuracy and precision. VWG obtained manually had R2 values of 0.92, 0.91, and 0.89 (p ≤ 0.01) for MacGran-IJ, SmartGrain, and GrainScan, respectively. The ventral and lateral color red of wheat grain differed between MacGran-IJ and GrainScan; the lateral part of the bread wheat grain showed differences for red, green, and blue (p ≤ 0.05). DIA in CIELab color required 11.57 s, and in RGB it needed 7.25 s (p ≤ 0.01). Manual measurement of 50 wheat grains required 1528.07 s, whereas SmartGrain took 101.84 s, MacGran-IJ 45.4 s, and GrainScan 54.11 s (H = 36.59, p ≤ 0.01). MacGran-IJ is an automated, simple, and efficient method for accurately measuring the size, length, and width of wheat grains using digitized images

Underwater Image Enhancement Based on the Improved Algorithm of Dark Channel

Enhancing underwater images presents a challenging problem owing to the influence of ocean currents, the refraction, absorption and scattering of light by suspended particles, and the weak illumination intensity. Recently, different methods have relied on the underwater image formation model and deep learning techniques to restore underwater images. However, they tend to degrade the underwater images, interfere with background clutter and miss the boundary details of blue regions. An improved image fusion and enhancement algorithm based on a prior dark channel is proposed in this paper based on graph theory. Image edge feature sharpening, and dark detail enhancement by homomorphism filtering in CIELab colour space are realized. In the RGB colour space, the multi-scale retinal with colour restoration (MSRCR) algorithm is used to improve colour deviation and enhance colour saturation. The contrast-limited adaptive histogram equalization (CLAHE) algorithm defogs and enhances image contrast. Finally, according to the dark channel images of the three processing results, the final enhanced image is obtained by the linear fusion of multiple images and channels. Experimental results demonstrate the effectiveness and practicality of the proposed method on various data sets.

Preferred skin reproduction of different skin groups

To accurately reproduce preferred skin colour is an important goal in digital image colour reproduction. A psychophysical experiment was conducted to specify the preferred skin colour for different skin types. Ten original facial images were captured to cover different skin types, including Caucasian, Chinese, South Asian, African, different genders and different ages. For each original image, 49 rendered images, uniformly sampled within the skin colour ellipsoid in CIELAB colour space, were used to morph the skin colours. Thirty observers from each of three ethnic backgrounds, Caucasian, Chinese and South Asian, participated in the experiment to investigate ethnic differences. Ellipsoid models were developed to specify preferred skin colour regions and centres for each original image. These results can be used to improve skin colour reproduction of colour imaging products, for example on mobile phones, for different skin types.

Study on the Chromogenic Mechanism of Violet-Red to Orange-Red Spinel

In this paper, the chemical composition and spectroscopy of 38 violet-red to orange-red spinel samples were analyzed by energy dispersive X-ray fluorescence spectrometer and UV-visible spectrophotometer, based on CIE LAB color space, to explore the effects of different chromogenic ions on spinel color transition. The results show that the chroma C* is mainly controlled by the variation of color coordinates a*, and the hue angle h° is mainly controlled by the variation of color coordinates b*. Spinel can be divided into iron-bearing spinel and chromium–vanadium-bearing spinel according to their chromogenic elements, where Fe is mostly present in the form of Fe2+, which is mainly replaced with Mg2+ in tetrahedral voids, while Cr3+ and V3+ are mainly replaced with Al3+ in octahedral voids, and the level of Al content indicates the degree of isomorphism in spinel. The lightness L* of spinel decreases with increasing Cr content, and the hue h° decreases with increasing Fe and increases with increasing V content.

Supramolecular Study of the Interactions between Malvidin-3-O-Glucoside and Wine Phenolic Compounds: Influence on Color.

Supramolecular study of the interactions between the major wine anthocyanin, malvidin-3-O-glucoside (Mv3G) and different wine phenolic compounds (quercetin 3-O-β-glucopyranoside (QG), caffeic acid, (-)-epicatechin, (+)-catechin, and gallic acid) has been performed at two different molar ratios (1:1 and 1:2) in acidic medium where flavylium cation predominates (pH ≤ 2). Color variations have been evaluated by differential colorimetry using CIELAB color space. These studies have been complemented with isothermal titration calorimetry assays and molecular dynamics simulations. The color of Mv3G flavylium cation is modified by the interaction with QG toward more bluish and intense colors. Interaction constants between the anthocyanin and the different phenolic compounds were obtained, ranging from 9.72 × 108 M-1 for QG to 1.50 × 102 M-1 for catechin. Hydrophobic interactions and H-bonds are the main driving forces in the pigment/copigment aggregation, except for the interactions where caffeic acid is involved, in which hydrophobic interactions acquire greater preponderance.

Fabrication of fabric-like bacterial cellulose/activated carbon membrane for filtering layer applications of face masks by using textile padding method

PurposeBacterial cellulose (BC) is an ideal alternative filtering material. However, current functionalization approaches for BC have not been fully discovered industrially as well as academically applying textile processing. This study aims to create a sustainable fabric-like membrane made of BC/activated carbon (AC) for applications in filtration using textile padding method, to protect people from respiratory pandemics.Design/methodology/approachFabric-like BC is first mechanically dehydrated then AC is loaded via a textile padding step. The finishing efficacy, properties of fabric-like BC/AC and NaOH pretreatment are analyzed and characterized by scanning electron microscope (SEM), field emission scanning electron microscope (FE SEM), X-ray diffraction (XRD), CIELab color space, color strength (K/S), nitrogen adsorption-desorption isotherm including Brunauer–Emmett–Teller (BET) specific surface area and Barrett–Joyner–Halenda (BJH) pore size and volume.FindingsThis research results in a fabric-like BC/AC with pore diameters of 3.407 ± 0.310 nm, specific surface area of 115.28 m2/g and an efficient scalable padding process, which uses 8 times less amount of chemical and nearly 30 times shorter treating duration than conventional methods.Practical implicationsOur globe is now consuming an alarming amount of non-degradable disposable masks resulting in massive trash buildup as a future environmental problem. Besides, current disposable masks requiring a significant upfront technological investment have posed challenges in human protection from respiratory diseases, especially for countries with limited conditions. By combining a sustainable material (BC) with popular padding method of textile industry, the fabric-like BC/AC will offer sustainable and practical values for both humankind and nature.Originality/valueThis research has offered an effective padding process to functionalize BC, and a unique fabric-like BC/AC membrane for filtration applications.

Carbureto de cálcio na antecipação e na uniformização do amadurecimento de frutos de Cajá-manga

Abstract Fruit ripening promoted by the exogenous application of ethylene analogs, such as calcium carbide, has commercial advantages. Thus, the knowledge of the responses of fruits treated with ethylene-inducing agents is essential to optimize the use of these substances. This work aimed to evaluate the influence of exposure to calcium carbide on the anticipation and standardization of postharvest ripening of cajá-manga fruits. Physiologically mature fruits were exposed to calcium carbide for 24 hours at concentrations 0, 20, 40, 80, and 110 g m-3. The fruits were stored at a temperature of 28±2 °C and evaluated at 0, 2, 4, and 6 days for the loss of fresh mass, color of the epidermis and pulp given by the CIELAB color space, titratable acidity, soluble solids content, the ratio between soluble solids content and titratable acidity, and vitamin C content. Cajá-manga fruits treated with different concentrations of calcium carbide had their ripening anticipated without compromising their characteristics. The concentrations of 20, 40, and 80 g m-3 of calcium carbide allowed the anticipation and standardization of fruit ripening within four days during storage, while for the highest concentration (110 g m-3), complete maturation was accelerated, occurring between two and four days of storage.

Comparative Evaluation of Colour Difference using Different Digital Photography Equipment to Standardise Colour Assessment in Dental Photography

Introduction: Photography has always been thought to be an important part of dentistry. Since the introduction of digital technology; imaging has become simpler and more accessible. Colour matching is critical for the success of restorative and prosthetic dental treatments, also communication with the dental laboratory is still subjective, and further research is needed on this aspect. Aim: To compare the colour difference of digital photography equipment to standardise colour assessment in dental photography. Materials and Methods: A single blind clinical study was conducted after Clinical Trial Registry of India (CTRI) registration and informed consent, and a total of 22 participants were enrolled in accordance with the selection criteria. A spectrophotometer was used to determine the CIELAB ( Colour space defined by International Commission on Illumination) values of the right central incisor for each participant. They were then photographed with five different photography equipment along with a grey reference card. Group A: Canon 1300D Digital Single Lens Reflex (DSLR) with 100 mm lens and pop-up flash (N=22). Group B: Canon 1300D DSLR+100 mm lens+70 GSM white tissue paper (N=22). Group C: Canon 1300D DSLR+ 100 mm lens with ring flash (N=22). Group D: Canon 1300D DSLR+100 mm lens+ring flash+diffuser (N=22). Group E: Iphone 11 (N=22). CIELAB values were obtained for all the groups using adobe Photoshop software after white balancing. Delta E was calculated by comparing CIELAB values of each group with the spectrophotometer values. The data was tabulated and analysed with one way ANOVA and Post-hoc test using IBM SPSS 20 software with p<0.05 considered statistically significant. Results: Mean value of Delta E was least in group D (DSLR+ ring flash with diffuser) (5.033) followed by group B (DSLR+ pop up flash with white tissue paper) (6.57), group A (DSLR+pop-up flash) (10.70), group E (Iphone11) (10.74) and highest in group C (DSLR+ ring flash) (11.32). Conclusion: Group D (DSLR+ ring flash with diffuser) was determined to be closest to the standard spectrophotometric values for colour assessment in dental photography

Possible effects of the MC4R Asp298Asn polymorphism on pig production traits under ad libitum versus restricted feeding.

The missense mutation Asp298Asn in the melanocortin 4 receptor (MC4R) is associated with daily gain or fatness in pigs (Sus scrofa domesticus). However, to the best of our knowledge, no study has directly compared the effects of the polymorphism between different feeding levels, even though diet plays a vital role in the swine industry. To explore possible differences, data from 439 mostly commercial hybrids fattened ad libitum and 119 commercial hybrids fattened with restricted feed ration were collected. The recorded traits were average daily gain (ADG), feed conversion ratio (FCR), carcass weight (CW), dressing percentage (DP), lean meat content (LM), backfat thickness (BFT), lean cuts weight, and meat quality parameters such as pH, temperature, drip loss, and CIELAB colour space. The general linear model revealed that the overall effect of MC4R was not statistically significant, but significant differences (p < 0.05) were found in ADG, FCR, CW, DP, LM, and BFT. In the ad libitum category, the AA genotype (298Asn/298Asn) tended to be the most favourable for growth-related traits, with the lowest LM, which is consistent with previous findings. In the restricted category, on the other hand, GA heterozygotes (298Asp/298Asn) achieved the best performance in terms of growth, whereas AA homozygotes showed the worst performance. Therefore, these results raise the possibility of an interaction between MC4R and the feeding level.

A Novel Method for Calibration of Digital Soil Images Captured under Irregular Lighting Conditions.

Soil color is commonly used as an indicator to classify soil and identify its properties. However, color-based soil assessments are susceptible to variations in light conditions and the subjectivity of visual evaluations. This study proposes a novel method of calibrating digital images of soil, regardless of lighting conditions, to ensure accurate identification. Two different color space models, RGB and CIELAB, were assessed in terms of their potential utility in calibrating changes to soil color in digital images. The latter system was determined to be suitable, as a result of its ability to accurately reflect illuminance and color temperature. Linear regression equations relating soil color and light conditions were developed based on digital images of four different types of soil samples, each photographed under 15 different light conditions. The proposed method can be applied to calibrate variations in the soil color obtained by digital images, thus allowing for more standardized, objective, and accurate classification and evaluation of soil based on its color.

IMPACT OF TEMPERATURE AND ULTRAVIOLET RADIATION ON CHANGES OF COLOUR OF FIR AND SPRUCE WOOD

This study deals with the investigation of impact of temperature and ultraviolet (UV) radiation on spruce wood (Picea abies(L.) H. Karst.) and fir wood (Abies albaMill.) colour changes. Samples of investigated woods species were loaded by temperatures of 110, 130, and 150°C and UV radiation (with 253.7 nm wavelength and 40 W m-2intensity) during 72, 168, 336 and 672 hours. Colour changes were evaluated in the CIE Lab colour space. Theneural network for prediction of both colour coordinates and total colour difference of spruce and fir wood was trained by data regarding exposure conditions (temperature, UV radiation and time) and by obtained results. Coefficient of determination (R2) of the neural network was above 0.99 for training, validation and testing. Average colour coordinates (±standard deviation) of the spruce and fir wood before exposure were L* = 80.08 ± 3.70,a* = 7.55 ± 2.13, b* = 21.56 ± 1.79, L* = 80.46 ± 1.91, a* = 6.84 ± 0.97, and b* = 18.90 ± 1.26, resp. Total colour differences after thermal loading were in the interval from ΔEab* = 3.76 ± 1.95 (spruce wood at 110°C) to ΔEab* = 45.37±1.46 (fir wood at 150°C). Total colour differences of both wood species exposed by UV radiation were approximately in intervals from ΔEab* = 12 to 13 (after 72 h) up to ΔEab* = 16 to 20 (after 168 to 672 h). Obtained results proven that both temperature and UV radiation have significant impact on the colour changes of the investigated woods.

The “Geometry” of Matyushin’s Color Triads: Mapping Color Combinations from the Reference Book of Color in CIELAB

This study analyzes the color combinations composed by the Russian avant-garde artist Mikhail Matyushin and his disciples for the Reference Book of Color (1932), summarizing their experimental research on the psychophysiology of color perception. Having extracted 90 colors from the 34 hand-painted charts of the book’s first edition, we mapped these colors into CIELAB and CIELCh color space to identify their chromatic characteristics (hue, lightness, and chroma) and their combination criteria. We demonstrate graphically that each of Matyushin’s color triads has a similar “geometry” in CIELAB color space and on the color circle, with the “intermediary” color much closer to the environment background color than to the main color. We conclude that the revealed patterns of these relationships are identical in all four sections of the Reference Book and reflect underlying principles of Mathyshin’s theory of color as well as fundamental aspects of human perception. The analysis presented in this paper will facilitate opportunities for artists, architects, designers, and other color professionals to use Matyushin’s original color charts and incorporate Matyushin’s style in creating their own triads.

Color-concept associations reveal an abstract conceptual space

People have systematic associations between concepts and colors, and concepts can be represented as distributions of association strengths across color space (Schloss et al., 2018; Rathore et al., 2018). These patterns of color-concept associations are highly structured, which may be critical in explaining how people flexibly draw conceptual inferences about colors across contexts (Mukherjee et al., 2021). Here we consider the structure of the semantic space defined by color-concept associations, evaluating three hypotheses. First, the structure may be arbitrary, so that items evoking similar color associations are conceptually unrelated. Second, it may be redundant with the conceptual structure apparent in natural language, as estimated via word-embeddings computed from large corpora. Third, it may express a representational space important to human cognition but distinct from those latent in natural language. To adjudicate these possibilities, we selected 30 concepts spanning 6 semantic domains (clothes, emotions, scenes, fruits, directions, times-of-day) and collected association ratings between each concept and 58 colors sampled across CIELAB color space. Pairwise dissimilarities between concepts in this color-concept association space could be well captured by embedding them in a reduced 3-dimensional space, where concepts with similar color-concept associations were embedded near each other. A comparison of these embeddings to those extracted from natural-language corpora revealed highly distinct conceptual spaces. To determine whether the color-concept space is reflected in mental representations of word meanings, we used a triadic comparison task to embed the same items based on direct judgments of semantic similarity. These correlated more strongly with color-concept similarities, suggesting that the color-concept associations express aspects of semantic knowledge beyond color that elude word-embeddings derived from language corpora, supporting our third hypothesis. The results provide an initial basis for understanding abstract conceptual relations in human cognition through the lens of visual properties of concepts — in this case, color-concept associations.