Average Color Difference Research Articles

Switchable Camouflage via Reflective Display

In nature, dynamic camouflage is performed by cephalopods and reptiles. Humans attempt to perform dynamic camouflage by employing display devices to show the surrounding background. In this work, a switchable camouflage device based on an electrophoretic display (EPD) is proposed. Color‐filter EPDs display colors by reflecting light through the color filters and black‐and‐white EPDs. The number of subpixels is found to be an important factor on color performance. To improve the poor saturation of color‐filter EPDs, the number of color filter subpixels is reduced. Compared with filters with three and four subpixels, a dual‐subpixel filter proposed in this work significantly improves the average saturation of red, green, and blue colors, with increases of 49% and 112%, respectively. Subsequently, the spectral characteristics of the color filter and black‐and‐white EPD are optimized by using genetic algorithm to reduce the average color difference between the display and the switchable target color, which can be reduced as low as 0.18. To visually demonstrate the color reproduction capability of the dual‐subpixel EPD, sample applications including the switchable vegetation and digital camouflages are designed and have a high degree of agreement with the background. In this work, an innovative and effective approach is introduced to dynamic camouflage.

Quantitative analysis of the color accuracy and reproducibility in digital textile printing: Discrepancies within color reproduction media

In this study, the overall color reproducibility of a digital textile printing (DTP) process was investigated within a wide range of color gamuts, and the design color factors influencing color reproduction in the initial design stage of DTP were analyzed. During the DTP process, digital image samples, display samples, and fabric samples, each representing 1296 colors, were created. The color reproduction ranges of these samples and the discrepancies in their lightness, chroma, hue, and overall color were analyzed. It was found that the display samples exhibited the widest color reproduction range, particularly in terms of lightness and chroma. In contrast, the fabric samples showed the narrowest color reproduction range. Furthermore, it was shown that color reproducibility was significantly reduced when the medium transitions from display to fabric during the DTP process, with an average overall color difference of 24.34 Δ E*ab being detected between the display and fabric samples. The low color reproducibility of this DTP process was particularly influenced by the K ratio and the L*10 and C*ab,10 values of the initial design. More specifically, the K ratio and the L*10 value of the initial design imparted significant effects on the lightness reproducibility, whereas the C*ab,10 value of the initial design had a significant impact all on the lightness, chroma, hue, and overall color reproducibilities.

Reconstructing hyperspectral images of textiles from a single RGB image utilizing the multihead self-attention mechanism

Hyperspectral images possess abundant information and play a pivotal role in enhancing the accuracy of color difference detection in textiles. However, traditional hyperspectral imaging methods necessitate costly equipment and intricate operational procedures. A novel deep learning model based on a multihead attention mechanism was proposed in this article to facilitate the extensive application of hyperspectral imaging technology in textile quality inspection. This model enabled the reconstruction of the hyperspectral information of plain weave textiles from a single RGB image. In this model, encoder-decoder architecture and pyramid pooling convolutional operations were employed to integrate multiscale features of plain weave cotton-linen textiles. This could capture details and contextual information in textile images more precisely, enhancing the accuracy of hyperspectral image reconstruction. Simultaneously, an attention mechanism was introduced to increase the model’s receptive field and improve its focus on key regions in the input image and feature maps. This resulted in a reduced weighting of redundant information during network learning, leading to an improved feature extraction capability of the network. Through these methods, successful reconstructions of plain weave textiles hyperspectral information from a single RGB image was achieved. Quantitative and qualitative tests were conducted on two datasets, namely, the NTIRE 2020 dataset and a self-made textile dataset, to evaluate the performance of the proposed method. The approach proposed in this article exhibited promising results on both datasets. Specifically, the reconstructed textile hyperspectral images achieved a root mean square error of 0.0344, a peak signal-to-noise ratio of 29.945, a spectral angle mapper of 3.753, and a structural similarity index measure of 0.955 on the textile dataset. In the reconstructed hyperspectral colorimetric experiment, the maximum value of average color difference was 2.641. These results demonstrate that the method can meet the requirements for textile color measurement applications.

The impact of various substrates, ceramic shades, and brands on the ultimate color and masking capacity of highly translucent monolithic zirconia: an in vitro study.

This study aimed to examine the impact of substrates, ceramic shades, and brands on the color and masking ability of highly translucent monolithic zirconia (HTMZ) using CIELab and CIEΔE2000 metrics. A total of 156 1-mm thick HTMZ disks in shades A1, A2, and A3 were produced using Dental Direkt and Kerox zirconia brands. Four 3-mm thick substrates (nickel-chromium alloy, non-precious gold alloy (NPG), zirconia shade A2, and resin composite shade A2) were prepared. HTMZ disks were overlaid on these substrates, and color measurements were taken with a spectrophotometer. Color differences (ΔE) were analyzed using CIELab and CIEΔE2000 formulas. The influence of brand, shade, substrate, and their interactions on ΔE values was assessed with a General Linear Model (GLM) and LSD pairwise comparison test. Spearman's correlation test examined the relationship between CIELab and CIEΔE2000 values. Results indicated that ΔEab was significantly influenced by substrate type and shade, while ΔE2000 was also affected by the ceramic brand. Mean color differences across ceramic-substrate groups were within clinically acceptable and perceptible ranges (clinically perceptible: ∆Eab ≥ 1.3 and ∆E2000 ≥ 0.8; clinically acceptable: 0.8 < ∆E2000 ≤ 1.8 and 1.3 < ∆Eab ≤ 2.7), except for NPG, which had ΔE values exceeding the perceptible range (ΔE2000: 1.1 ± 0.11 to 1.8 ± 0.31; ΔEab: 1.61 ± 0.15 to 2.16 ± 0.36). A significant correlation (r = 0.974, P < 0.001) was found between ΔEab and ΔE2000. Various ceramic brands and shades led to notable ΔE variations, yet average color differences within all ceramic-substrate groups remained clinically acceptable. Both ΔEab and ΔE2000 were reliable methods with a strong correlation for measuring color differences.

Construction of a full color gamut color mixture model and a color prediction neural network model for rotor spinning melange yarn

This research is devoted to solving the bottleneck problem of spinning and color prediction of rotor spun melange yarn within the full color gamut range. Firstly, a full color gamut grid color mixture model with hue regulation range of 0–360°, saturation regulation range of 0-1 and lightness regulation range of 0-1 was constructed based on the preferred seven primary-color fibers through grid color mixing. Secondly, in the full color gamut grid color mixture model, 55 representative grid points were planned and corresponding melange yarns and fabrics were prepared based on grid point parameters, which served as the training set, testing set, and validation set for constructing neural network prediction models. Thirdly, based on the demand for color prediction of melange yarns, a color prediction model and a blending ratio prediction model with six sub-models are constructed by taking the reflectivity of the melange yarn fabric and the blending ratio of the primary-color fibers as input and output values. Finally, six prediction samples were produced to verify the predictive performance of the neural network prediction model. The results show that the average color difference between the predicted and measured colors is 1.06, and the average root mean square error value between the predicted and actual blending ratios is 0.48%. It indicates that the neural network prediction model based on the grid mixing of seven primary-color fibers can adapt to the prediction of the mixing color and blending ratio of primary-color fibers in the full color gamut. Due to the construction of six sub-models based on local hue regions, the prediction accuracy of the neural network prediction model has been greatly improved.

Colorimetric Evaluation of a Reintegration via Spectral Imaging-Case Study: Nasrid Tiling Panel from the Alhambra of Granada (Spain).

Color reintegration is a restoration treatment that involves applying paint or colored plaster to an object of cultural heritage to facilitate its perception and understanding. This study examines the impact of lighting on the visual appearance of one such restored piece: a tiled skirting panel from the Nasrid period (1238-1492), permanently on display at the Museum of the Alhambra (Spain). Spectral images in the range of 380-1080 nm were obtained using a hyperspectral image scanner. CIELAB and CIEDE2000 color coordinates at each pixel were computed assuming the CIE 1931 standard colorimetric observer and considering ten relevant illuminants proposed by the International Commission on Illumination (CIE): D65 plus nine white LEDs. Four main hues (blue, green, yellow, and black) can be distinguished in the original and reintegrated areas. For each hue, mean color difference from the mean (MCDM), CIEDE2000 average distances, volumes, and overlapping volumes were computed in the CIELAB space by comparing the original and the reintegrated zones. The study reveals noticeable average color differences between the original and reintegrated areas within tiles: 6.0 and 4.7 CIEDE2000 units for the yellow and blue tiles (with MCDM values of 3.7 and 4.5 and 5.8 and 7.2, respectively), and 16.6 and 17.8 CIEDE2000 units for the black and green tiles (with MCDM values of 13.2 and 12.2 and 10.9 and 11.3, respectively). The overlapping volume of CIELAB clouds of points corresponding to the original and reintegrated areas ranges from 35% to 50%, indicating that these areas would be perceived as different by observers with normal color vision for all four tiles. However, average color differences between the original and reintegrated areas changed with the tested illuminants by less than 2.6 CIEDE2000 units. Our current methodology provides useful quantitative results for evaluation of the color appearance of a reintegrated area under different light sources, helping curators and museum professionals to choose optimal lighting.

Drying Kinetics and Quality Attributes of Solar-Dried Red Peppers: A Comparative Study with Traditional and Industrial Methods

Red pepper is a valuable ingredient known for its abundance of vitamins and antioxidants. But, it usually needs to be dried for longer preservation. Hence, this research is aimed at examining the drying kinetics and quality attributes of dried red peppers utilizing various solar drying methods, in comparison with traditional open sun drying (OSD) and industrial laboratory thin layer dryers (LTLD). Analysis parameters employed include determining moisture content, measuring color properties, evaluating antioxidant capacity, analyzing capsaicinoid content, and assessing microbial presence. The drying process took place in a sunny environment with fluctuations in temperature and relative humidity as evidence of the dynamic conditions experienced within the solar dryers. It was observed that the logarithmic model was the most accurate in predicting moisture ratio over time, estimating a drying time of 25 hours to achieve 10% moisture content. The result demonstrated that direct solar dryers (DSD), indirect solar dryers (ISD), and mix mode solar dryers (MMSD) methods showed moderate changes in color parameters, with average ΔL∗, Δa∗, and Δb∗ values of -5.08, -23.71, and -13.62, respectively. The average overall color difference (ΔE) for these methods was 27.96. In addition, after comparing it to the LTLD method, which showed the highest content of phenolic compounds at 47.89%, MMSD displayed a slightly lower content of 44.71%. Similarly, MMSD exhibited higher levels of DPPH radical scavenging activity and ferric reducing power, measuring at 44.22% and 1163.75 μmol Fe2+/L, respectively. The capsaicin content remained relatively consistent across all drying methods, with MMSD, DSD, and LTLD demonstrating similar levels of approximately 31 mg/g. Although MMSD had slightly higher mold and yeast counts compared to LTLD, ISD, and DSD, it remained lower than OSD (less than 0.56 log10cfu/g). Furthermore, MMSD showed a lower total microbial count in comparison to other drying methods. These findings suggest that MMSD shows promise as a drying technique for preserving the phenolic compounds and antioxidant activity of the dried product.

Color arrestor pixels for high-fidelity, high-sensitivity imaging sensors

Abstract Silicon is the dominant material in complementary metal-oxide-semiconductor (CMOS) imaging devices because of its outstanding electrical and optical properties, well-established fabrication methods, and abundance in nature. However, with the ongoing trend toward electronic miniaturization, which demands smaller pixel sizes in CMOS image sensors, issues, such as crosstalk and reduced optical efficiency, have become critical. These problems stem from the intrinsic properties of Si, particularly its low absorption in the long wavelength range of the visible spectrum, which makes it difficult to devise effective solutions unless the material itself is changed. Recent advances in optical metasurfaces have offered new possibilities for solving these problems. In this study, we propose color arrestor pixels (CAPs) as a new class of color image sensors whose composite spectral responses directly mimic those of the human eye. The key idea is to employ linearly independent combinations of standardized color matching functions. These new basis functions allow our device to reproduce colors more accurately than the currently available image sensors with red-green-blue filters or other metasurface-based sensors, demonstrating an average CIEDE2000 color difference value of only 1.79 when evaluating 24 colors from the Gretag-Macbeth chart under standard illuminant D65. Owing to their high fidelity to the human eye response, CAPs consistently exhibit exceptional color reproduction accuracy under various spectral illumination compositions. With a small footprint of 860 nm height and 221 nm full-color pixel pitch, the CAPs demonstrated high absorption efficiencies of 79 %, 81 %, and 63 % at wavelengths of 452 nm, 544 nm, and 603 nm, respectively, and good angular tolerance. With such a high density of pixels efficiently capturing accurate colors, CAPs present a new direction for optical image sensor research and their applications.

Optimized combined spectral calibration model based on single-constant Kubelka–Munk theory for color prediction of pre-colored cotton fiber blends

There is a complex interaction between primary fibers in pre-colored fiber blends, which leads to the impracticality of Duncan’s additive theory in the single-constant Kubelka–Munk (KM-1) model, resulting in poor accuracy for color prediction. This paper builds an optimized combined spectral calibration model based on the KM-1 function to solve this issue. This proposed model involves colored and non-colored parts with trained spectral coefficients, and was established to achieve a good matching between the predicted and actual ratio of the absorption coefficient K to the scattering coefficient S, K/ S of samples for color prediction from the perspective of the spectrum. Experimentally, five primary cotton fibers were selected and prepared into fiber blends to demonstrate its validation, and its optimal number of training samples was found to be only 14, which was much less than the general training sampling method using half of all samples directly. Finally, a remarkable average color difference of only 0.63 CIEDE2000 units for 74 test specimens was achieved on this optimized model, which was significantly lower than that of the KM-1 model (∼1.16), two-constant KM model (∼1.03), Stearns–Noechel model (∼1.16) and Friele model (∼1.11). The results indicate that the optimized model behaves well and could be applied in the color prediction of pre-colored fiber blends.

Deep learning model for dynamic color design of all-dielectric metasurfaces.

Silicon nanostructure colors have rapidly developed in recent years, offering high resolution and a broad color gamut that traditional pigments cannot achieve. The reflected colors of metasurfaces are determined by the geometric structure of the unit cell and the refractive index matching layer parameters. It is evident that the design of specific colors involves numerous parameters, making it challenging to achieve through conventional calculations. Therefore, the tandem network instead of conventional electromagnetic simulation is natural. The forward part of the network incorporates feature cross terms to improve accuracy, enabling high-precision predictions of structural colors based on structural parameters. The average color difference between the predicted and actual color values in the L,a,b color space is 1.38. The network has been proven to accurately predict the refractive index and height of the refractive index matching layer during the dynamic tuning process. Additionally, the issue of the inverse network converging to incorrect solutions was addressed by leveraging the characteristics of the activation function. The results show that the color difference between the colors designed by the inverse network compared to the actual colors in the L,a,b color spaces is only 2.22, which meets the requirements for commercial applications.

Does the whitening dentifrice containing activated charcoal interfere with the properties of dental enamel? Microhardness, surface roughness and colorimetry analyzes.

The study evaluated the influence of whitening dentifrice containing activated charcoal on microhardness (MH), surface roughness (Ra) and colorimetry of tooth enamel. A total of 60 healthy bovine incisor teeth were used and divided into: G1 (regular non-whitening toothpaste), G2 (conventional whitening toothpaste), G3 (activated charcoal based whitening toothpaste) and G4 (10% carbamide peroxide gel - PC10). Groups G1, G2 and G3 underwent simulated toothbrushing for 14 days, while G4 received bleaching treatment for the same time. The readings of MH, Ra and colorimetry were performed before any intervention (T0 - baseline) and at the end of 14 days of the proposed treatments (T1). After confirming the normality of the data, the results of MH, Ra and colorimetry were subjected to 2-factor ANOVA for repeated measures (α = 5%). For MH, there were no statistical differences in G1, G2 and G3, only in G4. Considering Ra, a significant increase was observed in G2 and G3, with no statistical differences found in the other groups. Regarding colorimetry, the average color difference (ΔE) was greater in the G4 group (11.30 ± 4.31), even compared to the groups submitted to the whitening dentifrices: G2 (5.13 ± 2.75) and G3 (5.86 ± 3.66). It was concluded that the use of a whitening toothpaste containing activated charcoal caused deleterious effects on the enamel Ra, but did not affect the microhardness of the substrate, besides promoting a color change inferior to the regular non-bleaching toothpaste or PC10 gel. Key words:Whitening dentifrice, charcoal, roughness, microhardness.

A comparative study on the difference of color space conversion based on table look-up method and neural network

A comparative study on the difference of color space conversion based on table look-up method and neural network

Measuring visual and non-visual lighting metrics in building environments with RGB sensors

Today, a spectroradiometer can measure both visual lighting metrics (illuminance, chromaticity coordinates or color temperature) and non-visual lighting metrics (Circadian Stimulus Model (CS2021) by M. Rea et al. Melanopic Equivalent Daylight (D65) Illuminance (mEDI) in CIE S 026/E: 2018) very well. Unfortunately, their cost and size make them unsuitable for practical measurements after installation. And some commercially available miniature RGB sensors can measure some visual parameters, but the question is whether they can accurately determine both visual and non-visual characteristics for practical evaluation or real-time luminaire control. Therefore, in this work, the mixed light between incident daylight and typical LED (500 lx horizontal illuminance, 4000 K) in a typical office was measured using the CSS45 spectrometer and the AS 73211 RGB sensor. Then 410/516 spectra were used to process the RGB sensor. And the remaining 20% were used to verify the obtained correlations. The correlation coefficient R2 of 0.892 and the RMSE value of 15.436 lx for the relationship between the mEDI values of the RGB sensor and the spectrophotometer and their average color difference Δu'v' of 3.78.10−5 were very good evidence for the processing quality. For verification, the correlation quality parameter pair (R2/RMSE) of CS2021 and Ev also reach the very good values of 0.842/0.015 and 0.78/29.15, respectively. Based on these achievements, it can be confirmed that a commercially available RGB sensor can measure the visual and non-visual parameters of lighting systems accurately enough after installation if it is properly processed as recommended in this work.

Color characterization model for OLED displays with crosstalk effects

AbstractConventional models (e.g., a Gain‐Offset‐Gamma model) were developed based on the assumption of channel independence. Modern displays (e.g., organic light‐emitting‐diode OLED displays), however, sometimes have crosstalk effects among the three channels, making these conventional models have poor performance. In this article, we propose a 3D piecewise model. It divides the display gamma encoded RGB space (i.e., R′G′B′) into 27 subspaces and characterizes the crosstalk effects between two and three channels based on the measurements of the tristimulus values of 64 combinations of digital signal values. The average color difference ΔE00 of the 41 combinations between the model predictions and the measurements was around 0.34, in comparison to the average of 3.61 by the GOG model, across nine uncalibrated OLED displays. In addition, for displays with breakpoints on each channel, a revision is proposed to divide the subspaces for maintaining the accurate predictions. The proposed model was also validated by calibrating 94 uncalibrated OLED displays to the standard P3 RGB specifications, resulting an average color difference ΔE00 reduced from 3.96 to 0.72.

Effects of Exogenous Substances Treatment on Fruit Quality and Pericarp Anthocyanin Metabolism of Peach

In this study, the two peach cultivars ‘Baifeng’ and ‘Weiduanmihong’ were used as experimental materials, and their fruits were sprayed with different concentrations of L-glutamic acid, brassinolide, and sucrose to study the effects of these three exogenous substances on fruit quality and anthocyanin metabolism of peaches. The results showed that the appearance quality (average single fruit weight, fruit firmness, and peel color difference), nutritional quality (soluble solids, soluble sugar, titratable acid, anthocyanins, total phenols, flavonoids, etc.), peel anthocyanin-related enzyme activity, and related gene expression of ‘Baifeng’ and ‘Weiduanmihong’ peaches treated with three different exogenous substances were different from those of the control. Higher-concentration treatments could significantly improve the appearance of peach fruit, the nutritional quality of peach fruit and the activity of anthocyanin-related enzymes in peel, as well as promote the expression of related genes. Treatment with 400 mg/L L-glutamic acid significantly promoted the average fruit weight of ‘Baifeng’ peaches. Treatment with 800 mg/L L-glutamic acid significantly promoted the increase in PAL enzyme activity and the expression of PpPAL and PpF3H in the two peach varieties and significantly promoted the expression of anthocyanin metabolism genes PpF3′H and PpGST1 in ‘Baifeng’ peach peel and anthocyanin metabolism gene PpUFGT in ‘Weiduanmihong’ peach peel. Treatment with 34 mg/L sucrose significantly increased the fruit firmness of ‘Baifeng’ peaches and the soluble sugar content of ‘Weiduanmihong’ peaches. Treatment with 51 mg/L sucrose significantly promoted the increase in flavonoid content and PpUFGT expression in ‘Baifeng’ peach fruit and significantly promoted the expression of anthocyanin metabolism genes PpDFR and PpANS in ‘Weiduanmihong’ peach peel. Treatment with 0.6 mg/L brassinolide significantly promoted the increase in soluble solids (TTS), soluble sugar, anthocyanin, total phenol content, PAL enzyme activity, UFGT enzyme activity, and the expression of anthocyanin metabolism genes PpDFR and PpMYB10.1 in ‘Baifeng’ peach fruit, and it significantly increased the average single fruit weight, fruit hardness, anthocyanin content, and UFGT enzyme activity of ‘Weiduanmihong’ peach fruit and promoted the expression of anthocyanin metabolism genes PpF3H and PpGST1 in ‘Weiduanmihong’ peach peel. The comprehensive effect of 0.6 mg/L brassinolide treatment on improving peach fruit quality and increasing anthocyanin content produces the best results and could be popularized in production practices.

Construction of grid color mixture model of seven primary-color and modified Stearns-Noechel color matching algorithm for color prediction of full-color-gamut rotor melange yarn

In this paper, the full-color-gamut grid color mixture model containing 601 grid points is constructed by ternary double coupling blending of seven primary-color fibers, and the spinning method of full-color-gamut melange yarn is given by combining with three-channel NC rotor spinning technology. A modified S-N color prediction model was constructed by selecting 55 uniformly distributed grid points for yarn and fabric production from the full-color-gamut grid color mixture model as samples for solving the reflectance conversion coefficients. On this basis, the method of predicting the color value of a melange yarn based on its primary-color fiber composition and blending ratio and predicting the primary-color fiber composition and blending ratio based on the color value of a melange yarn using the parameters of the nearest sample grid point is proposed, and six samples with different blending ratios in six color mixing regions of the full-color-gamut grid color mixture model are designed for validation. The results showed that the average color difference between the predicted color and the actual color of the melange yarn is 1.15, the predicted primary-color fiber composition of the melange yarn is consistent with the actual composition, and the average error between the predicted blending ratio and the actual blending ratio is 3.95%. The method proposed in this paper can effectively predict the color value and blending ratio of melange yarn.

A neural network algorithm and its prediction model towards the full color phase mixing process of colored fibers

Aiming at the demand of color matching techniques in the spinning process, a neural network prediction model is constructed in this research study, and the gridded full color phase mixing space of colored fibers is used as the sample space. Subsequently, 30 grid points are employed as training samples, while another 30 grid points are adopted as testing samples, in which the parameters of the input, hidden, and output layers are optimized. Additionally, the neural network prediction model is constructed by training samples, and validated by testing samples. Lastly, a neural network prediction model is applied to implement the prediction of color and mixing ratios for any point within the full color phase mixing model. Through the assessment of the testing samples, the predicted results for the colors of the grid point samples showed an average color difference of 1.29 (minimum was 0.22 and maximum was 2.97); the forecasts for the mixing ratios of the colored fibers were that the range of the mean absolute error for the mixing ratios of individual samples was from 0.01% to 0.18%, and the mean absolute error for the mixing ratios of all samples was 0.21%. The experimental results indicated that the proposed neural network model has a relatively advanced prediction accuracy.

The influence of workpiece surface texture on visual measurement of roughness

Abstract In view of the universal applicability of color information index and the ability of spectrum index to measure small roughness, this paper studies the influence of machined surface texture on roughness machine-vision measurement. Based on the micro-topography of different types of texture surface, the correlations between the average color difference, five typical spectral indices and the surface roughness parameters of two types of texture representative workpiece milling and grinding are focused on in the discussion. On this basis, the influences of the angle of light source and camera relative to the surface normal of the sample and the sample surface texture direction on roughness measurement are further discussed, and the reasons for the difference in index to characterize different process roughness are analyzed from the mechanism. The experimental results show that the different surface textures have different effects on light scattering, which in turn affects the sensitivity of the index to the roughness parameters.

Inverse design of a vanadium dioxide based dynamic structural color via conditional generative adversarial networks

Dynamic structural color provides an additional time dimension in contrast to the static one, enriching the information load and functionalities. As a phase change material, vanadium dioxide offers great opportunities to implement dynamic structural color as its insulator-metal transition. Nevertheless, the multiple states also place a barrier to the efficient design of the structure configurations. This work firstly reports the dynamic structural color inverse design of asymmetric Fabry-Pérot cavity through a parameter-based conditional generative adversarial networks approach. The proposed structure attains a gamut as large as 117% of sRGB in the insulator state and can produce a 5% color coverage variation via the phase change of VO2 layer. By using the trained conditional generative adversarial networks, the inverse design accuracy with the average color difference ΔE of 0.98 is achieved. A monochromatic pattern is designed by the trained networks to demonstrate different color dynamics of the various structures.

Hyperspectral reconstruction from a single textile RGB image based on the generative adversarial network

Hyperspectral images are capable of significantly increasing the accuracy of textile color measurement because of their rich information. However, hyperspectral imaging generally requires expensive equipment and complex operations. If the hyperspectral information can be reconstructed based on a single RGB image, it can facilitate the widespread application of hyperspectral imaging technology, such as in textile color measurement. In this paper, a deep learning model was proposed for hyperspectral reconstruction of cotton and linen fabrics based on the conditional generative adversarial network. According to this model, the encoder–decoder structure and spatial pyramid convolution pooling operation were adopted to fuse multi-scale features for the prevention of mode collapse. Atrous convolution was introduced to increase the receptive field to adapt to the fabric texture information, and the hyperspectral information of the fabric from a single RGB image was reconstructed. The quantitative and qualitative tests verified that the method in this paper had good results. The root mean square error and peak signal-to-noise ratio were 0.0271 and 31.372, respectively, for reconstructed fabric hyperspectral images; the highest average color difference [Formula: see text] in the reconstructed hyperspectral colorimetry experiment was obtained as 2.755. Thus, the proposed method can meet the common application requirements of color measurement.