CIC30 FINAL PROGRAM

Luminance underestimates the brightness of chromatic visual stimuli. This phenomenon, known as the Helmholtz-Kohlrausch effect, is due to the different experimental methods—heterochromatic flicker photometry (luminance) and direct brightness matching (brightness)—from which these measures are derived. This paper probes the relationship between luminance and brightness through a psychophysical experiment that uses slowly oscillating visual stimuli and compares the results of such an experiment to the results of flicker photometry and direct brightness matching. The results show that the dimension of our internal color space corresponding with our achromatic response to stimuli is not a scale of brightness or lightness.

The concept of color space has served as a basis for vast scientific inquiries into the representation of color including colorimetry, psychology, and neuroscience. However, the ideal color space that can both model color appearance attributes and color difference as a uniform Euclidean space is still not yet available. In this work, based on the alternative representation of independent one-dimensional color scales, the brightness and saturation scales for five Munsell principal hues were collected via partition scaling, where the MacAdam optimal colors served as anchors. Furthermore, the interactions between brightness and saturation were evaluated using maximum likelihood conjoint measurement. For the average observer, saturation as constant chromaticity is independent of luminance changes, while brightness receives a small positive contribution from the physical saturation dimension. This work further supports the feasibility of representing color as multiple independent scales and provides the framework for further investigation between other dimensions.

In order to reproduce the colour appearance between real scene and images on self-luminous display, this study conducted a series of psychophysical experiments using threshold method. Three types of real scenes were built up in a lighting room, including painting, fruit and vegetable, skin colour chart. Sixteen adapting conditions were designed including four CCTs (3000K, 4500K, 6500K, 8000K) and 4 luminance levels (10lux, 100lux, 500lux, 1000lux). Four displays with different sizes were studied. The result indicated that the colour appearance of real scene and the image on the display were different, especially for low CCT and luminance level. The contents of scene and size of display didn’t show a significate impact. The prediction performance of CIECAM16 was tested, and a revised formulation was proposed with high accuracy.

Automotive cockpits are becoming more and more digital day by day and this is evident by the increase in the number of displays inside the cockpit. With the advent of OLED displays, automotive cockpits not only have LCD but have started having mixed display technologies. A recent example of such a cockpit display is the MBUX Hyperscreen in Mercedes EQS.

A general situation is most mid ranged cars from Renault includes several displays inside the cockpit. These displays are placed at different locations inside the cockpit and go through various changes in external environmental condition during day and night. A vehicle cockpit experiences a wide range of illumination change during the day, from 35 kLux in bright sunlight to a few Lux during the night. The displays are generally used in high luminance range during the day and very low luminance during the night.

Traditional color difference formulae like the CIEDE76 or CIEDE2000, the latter being the current industry standard, are defined for a specific set of evaluation conditions. For the myriad of different conditions that the displays undergo inside the cockpit, there isn’t a recommended color difference formula which can be used to quantify the color difference between any two chosen displays inside the cockpit. An attempt has been made to include various real time parameters involving color difference evaluation between two displays, one of which is an LCD and other OLED. The motive of this study is to find out which color difference formula is the most representative of the perceived color difference between two mixed technology displays for a group of observers. The metric used for this the study is the CIE recommended STRESS index. The outcome of this research is to serve as a reference regarding the choice of color difference metric by display manufactures and OEM suppliers.

The development of various new imaging systems introduces new viewing conditions that did not exist in the past. Mixed reality systems, which capture the real environment and render the captured image on display with virtual objects superimposed, require more immersive and realistic feeling than virtual and augmented reality systems, especially when users just put on the headsets. Therefore, the rendering shown on the display needs to be carefully adjusted to match the appearance in the real environment. In this study, we specifically focus on reproducing the overall color tone of the real environment under different ambient illumination colors by shifting the white point of the display. The human observers viewed a real environment under different ambient illumination conditions, in terms of CCT and chromaticities, and evaluated the rendering of the captured scene with 44 white points. The results clearly suggested that the display white point should be adaptive to the ambient illumination color, especially when the ambient illumination had a CCT below 4000 K, to provide a good user experience.

High dynamic range (HDR) technology enables a much wider range of luminances – both relative and absolute – than standard dynamic range (SDR). HDR extends black to lower levels, and white to higher levels, than SDR. HDR enables higher absolute luminance at the display to be used to portray specular highlights and direct light sources, a capability that was not available in SDR. In addition, HDR programming is mastered with wider color gamut, usually DCI P3, wider than the BT.1886 (“BT.709”) gamut of SDR. The capabilities of HDR strain the usual SDR methods of specifying color range. New methods are needed.

A proposal has been made to use CIE LAB to quantify HDR gamut. We argue that CIE L* is only appropriate for applications having contrast range not exceeding 100:1, so CIELAB is not appropriate for HDR. In practice, L* cannot accurately represent lightness that significantly exceeds diffuse white – that is, L* cannot reasonably represent specular reflections and direct light sources. In brief: L* is inappropriate for HDR. We suggest using metrics based upon ST 2084/BT.2100 PQ and its associated color encoding, IC_TC_P.

Various uniform color spaces and color appearance models were mainly developed for characterizing stimuli under a low dynamic range condition. Real scenes in daily life, however, are commonly high dynamic range (HDR), containing highlights with luminance beyond the diffuse white, whose color appearance characterization was never investigated in the past. This study was carefully designed to investigate the color appearance characterization of highlights in HDR scenes, covering extremely wide ranges of diffuse white luminance (up to 11000 cd/m²), stimulus luminance (up to 49000 cd/m²), stimulus chromaticities (reach Rec. 2020 gamut), and scene luminance contrast (up to 72045). The observers viewed two stimuli, including one highlight and one dark stimulus, in a viewing booth, and were asked to adjust the color appearance of another stimulus, so that the color differences between the adjusted stimulus to each of the other two stimuli appeared the same. The results clearly showed that none of the existing models, including the one (i.e., IC_tC_p) that was recently designed for HDR scenes, has a good performance. The models using a power function to characterize the non-linear compressive responses of the human visual system (i.e., CIELAB and IPT) had a slightly better performance. The findings provided some guidance for performing tone mapping and chroma/saturation adjustments, and clearly suggest the necessity to carry out further work to develop a better model for HDR scenes.

Many psychophysical experiments have shown that color discrimination thresholds vary as a function of the color center under consideration, MacAdam ellipses being a prime example. Color discrimination thresholds are usually modelled either as ellipses in chromaticity space or ellipsoids in 3D color space. Various color difference models, such as those of Luo et al., have been developed based on fits to the data obtained from psychophysical experiments (e.g., Li et al.). However, such models do not explain why the color difference thresholds vary across color space as they do. Funt et al. show that the uncertainty created by metamer mismatching provides a very interesting explanation. Extending that theory as described below enables it to predict, not only the ellipsoid volumes, but the shapes and orientations of the ellipsoids as well, thereby providing further evidence of its validity.

Imaging is enabled by the limitations of the human visual system, which is blind to certain physical differences. The trichromacy of human vision, e.g., allows for very different materials to be combined in a way that results in the same signals triggered by the eye’s cones. As a result a print can elicit the same response as a display, or a projection can yield colors like those in a painting. The limits of spatial acuity too allow for discrete patterns, e.g., those resulting from halftoning, to appear continuous. This paper turns its attention to the limits of color difference perception in stimuli with a very small subtense, such as thin lines or fine features of 3D objects. A first set of psychophysical data, obtained in an on-line visual experiment, indicates a dramatic relaxation of perceptibility thresholds when comparing very thin with thicker lines. The second half of the paper then presents printer imaging pipeline strategies that take advantage of these experimental findings to successfully render fine lines while taking advantage of the more limited sensitivity with which their specific colors are perceived.

An experiment was carried out to investigate separation and CMF effects on colour-difference evaluation using display colours. In total, 1120 sample pairs around 5 CIE recommended colour centres were assessed 20 times using the grey-scale method. Sample pairs were selected to have colour-difference of 4 and 8 CIELAB units, include or exclude separation between two colours on a pair, have four fields of view (FoVs), 2º, 4º, 10º and 20º. The experiment results were used to test 3 colour-difference equations or uniform colour spaces, CIELAB, CIEDE2000 and CAM16-UCS. For separation (S) sample pairs, CIEDE2000 performed the best, followed by CAM16-UCS and CIELAB the worst. For no-separation (NS) sample pairs, all models gave worse performance than separation (S) sample pairs. The parametric formula derived earlier was verified to predict colour-difference for sample pairs to have no-separation line. Five colour matching functions (CMFs), CIE 1931, CIE 1964, CIE 2006-2º, CIE 2006-10º and 2006-4º were tested and the results indicated very small CMF effect on calculating colour-difference.

Usually images reproduced on different devices have different colors. In some settings, it is desirable for images reproduced on various devices to have a similar color impression. The degree of similarity of color impression among this set of images can be defined as consistent color appearance (CCA). If the CCA is high, colors appear similar (or consistent). To evaluate whether the colors have CCA, it is necessary to develop a metric to describe CCA. Although several color difference metrics, such as CIEDE2000, have been often used to evaluate color difference, they do not necessarily represent subjective differences in color impression. To solve this problem, the concept of a color trendline has been proposed. The degree of deviation from a trendline might be used to evaluate CCA, i.e., a set of prints from different printers whose colors have a smaller deviation from the trendline would have more consistent colors. In this study, we conducted several psychophysical experiments to verify this approach. Color patches were used to determine whether deviations from a trendline correlates with CCA. Psychophysical testing was used to find the CCA between sets of natural images and these results are used to determine whether deviations from a set of trendlines can explain the degree of CCA.

For a few years, the automotive industry has produced new cars with continuous changing display models, this by combining display sizes, forms, and technologies all together to bring new experiences to the users. In this paper we will present the color matching solution implemented for a new car display system where regular LCD display technology and LED lighting tiles are mixed together. The solution we proposed is based on accepted display model providing the transformation device RGB space to CIE XYZ independent color space and in reverse. The approach we followed choose the LCD display as reference, then transformation matrix is derived to modify the RGB LED control values. Once the color matching operation is applied, the color difference between the two display areas is greatly reduced.

Recently-proposed modifications to the CIECAM16 color appearance model require an update to its corresponding uniform color space, CAM16-UCS, in order to ensure that the formulas continue to predict the available color difference data. Theoretical and statistical inconsistencies in the current CAM16-UCS formulas are also discussed and addressed by the proposed revisions. The STRESS metric is used to derive new formulas for CAM16-UCS and to evaluate the performance of these formulas in comparison to existing uniform color spaces or color difference formulas on a common color difference dataset.

In colorimetry and color appearance modeling it is assumed that chromatic adaptation is reversible. Recent experimental results suggest that this might not be the case and that bidirectional models of chromatic adaptation might be needed. This paper describes a series of experiments designed to collect very-high precision corresponding colors data for sensory chromatic adaptation to test the hypothesis that chromatic adaptation is reversible for individual observers. The results indicate that there are small, but statistically significant, differences in corresponding colors due to changes in the previous state of adaptation. However the effect sizes are small and the number of repeated observations required to detect these differences is very large. Thus it is concluded that these differences are, while interesting, not problematic for practical colorimetry. In addition the application of the vk20 chromatic adaptation model for predicting such effects is further explored and its 15000K reference adaptation state is validated.

In this paper, we propose a noise-robust pulse wave estimation method from NIR video. Pulse wave estimation in the near-infrared region is expected to be applied to non-contact monitoring in dark areas. Conventional method cannot consider noise when performing estimation, so the accuracy of pulse wave estimation in noisy environment is not very high. This may adversely affect the accuracy of heart rate and other data obtained from pulse waves. Therefore, the objective of this study is to perform pulse wave estimation robust to noise. The Wiener estimation method was used in this study. The Wiener estimation method is a simple linear computation that can consider noise. The proposed method is expected to enable non-contact and accurate estimation of pulse wave from near-infrared video images. Experimental results show that the proposed method estimates the pulse wave more robustly to noise than the conventional method. Furthermore, the heart rate was estimated from the estimated pulse wave and the proposed method was able to obtain a value closer to the ground truth.

In this study, we introduced a measurement method for gloss unevenness as a function of reflectance angles. Gloss is one of the most important qualities of materials, and it is evaluated subjectively by the intensity of reflected light and gloss unevenness. People can estimate the texture of a material using gloss unevenness, which is found in the small peripheral part of the gloss area and can easily be recognized by observing the reflected light while moving the object or eyes. However, it is not easy to photograph and quantify gloss unevenness. One reason is the area where gloss unevenness is observed is a small area near specular reflection. Second, the appearance of gloss unevenness changes depending on the reflection angle. We developed a measurement apparatus to measure gonio-photometric gloss unevenness. We introduced two solutions: a wide-area gloss unevenness measurement technology using telecentric imaging and a rotating mirror optical system that defects reflected light at an angular resolution of 0.02°. We analyzed three materials—mirror, plastic, and paper—and proposed three indicators as a quantitative evaluation method for gloss: intensity of reflected light at the specular angle, full width at half maximum (FWHM) of Bidirectional Reflectance Distribution Function (BRDF), and gloss unevenness image at the FWHM.

Glare introduces a complex scene-depended transformation of the array of “All Scene Luminances” making a different spatial pattern in the array of light on all receptors, called “Retinal Contrast”. The spatial convolution of “All Scene Luminances” with Vos and van den Berg’s CIE 1999 Glare Spread Function calculates high-resolution arrays of “Retinal Contrasts”. The results show that uniform-luminance scene segments become low-slope gradients that are nearly invisible, or invisible. Visual inspection of these arrays is misleading. Plots of calculated “Retinal Contrast” values, histograms, and other numerical techniques are needed to analyze the effects of glare. Pseudocolor Look-up Tables (LUT)s are very helpful in visualizing the complexity of glare’s spatial transformation that controls the amount of light falling on rods and cones.

This article studies Lightness Illusions that contain two identical scene-luminance segments that are identified as the “Regions-Of-Interest”(ROI). Following receptor responses, neural spatial processes generate a second spatial-image transformation that leads to appearances. Contrast, Assimilation, and Natural Scene Illusions demonstrate [Appearance ≠ scene luminance]. Analysis of Illusion’s patterns of light on receptors shows that: Contrast Illusions, Edwin Land’s B&W Mondrian, Adelson’s Checkershadow all exhibit Glare’s Paradox. Namely, that vision’s second neural transformation overcompensates the effects of glare. Illusions’ GrayROIs appear darker despite large amounts of glare light on receptors. GrayROIs that appear lighter have smaller amounts of glare. Assimilation Illusions adds light to GrayROI’s that appear lighter. The combination of intraocular glare and Lightness Illusions shows complex-spatial-image-processing transformations following receptor responses in normal scenes.

Materials with special appearance properties such as goniochromatic materials require complex bidirectional measurements to properly characterise their colour and gloss. Normally, these measurements are performed by goniospectrophotometers which are expensive and not commonly available. In this paper a flexible imaging system composed of a snapshot multispectral camera and a light source attached to a robotic arm, is used to obtain HDR BRDF measurements of patinas commonly used in cultural heritage objects. The system is evaluated by comparing the results to those of a commercially available goniospectrophotometer. It is found that with a known uncertainty, the system is capable of producing accurate measurements of samples with a roughness equal or lower than 6.19 μm. For roughnesses higher that 12.48 μm, the accuracy of the system decreases. Moreover, it is found that the size and orientation of the region of interest plays a great influence on the precision of the imaging system.

Hue linearity is critically important to uniform color spaces and color appearance models. Past studies investigating hue linearity only covered relatively small color gamuts, which was generally acceptable for conventional display technologies. The recent development of HDR and WCG display technologies has motivated the development of new color spaces (e.g., IC_TC_P and J_za_zb_z). The hue linearity of these new color spaces, however, was not verified for the claimed HDR and WCG conditions, due to the lack of constant hue loci data. In this study, an experiment setup was carefully developed to produce HDR and WCG conditions, with the stimulus luminance of 3400 cd/m² and the diffuse white luminance of 1000 cd/m² and the stimulus chromaticities almost covering the Rec. 2020 gamut. The human observers performed a hue matching task, adjusting the hue of the test stimulus, with a hue angle step of 0.2°, at various chroma levels to match that of the reference stimulus at 21 different hues. The derived constant hue loci were used to test the various UCSs and suggested the need to improve the hue linearity of these spaces.

In the latter half of the 1980s, PM2.5 pollution in Beijing became a serious problem, and there were concerns about health hazards. It was expected that China's emissions must be reduced from 2013 to 2016, and the lockdown effect of Covid-19 would bring about an end, but it is still reluctant to regulate CO2 emissions. Again, in Beijing in November 2021, a visibility of 500 m or less has been observed, then road traffic is dangerous in addition to health.

After that, the center of pollution has moved from India to Mongolia, and now Nepal, Qatar and Saudi Arabia. The situation is still serious in developing countries.

Image restoration to remove the effects of haze and fog has been a long-standing concern of NASA, and their original Visual Servo has been put into practical use. Though the mainstream moved to the technique based on atmospheric physics. He et al.'s Dark Channel Priority (DCP) logic has had a certain effect on heavily polluted PM 2.5 scenes , but there is a limit to the restoration of detailed visibility. The observed images are affected by two spatial inhomogeneities of 1) atmospheric layer and 2) illumination.

As a countermeasure, we have improved DCP process with the help of Retinex and introduced the veil coefficient as reported in CIC24. Recently, a variety of improvements in single image Dehazing, using FFA-net, BPP-net, LCA-net, or Vision-based model are in progress. However, in each case, visibility of details is still a common problem.

This paper proposes an improvement in detail visibility by (1) joint sharpness-contrast preprocess and (2) adjustment in Dehaze effect with veil coefficient v.

Lastly, we challenge numerical evaluation of improvement in detail visibility by the two ways of attenuation of high-frequency Fourier spectrum and the expansion rate of the color gamut.

MA (Material Appearance) is a perceptual phenomenon that our brain deciphers from the retinal image. What features of retinal image are most closely related to the stimulus inside the visual cortex of V1 ~ V5? The function of V1 is the most well-studied. V1 has the function of seeing fine in the fovea and rough in the periphery, and is mathematically described by LPT (Log-Polar Transform). Since LPT samples the retinal image at a higher rate in the fovea but at lower rate peripherally, the color information tends to gather in center of V1. Paying attention to this LPT features in V1, we reported a novel method to transfer MA from one to another scenes. After LPT, PCM (Principal Component Matching) is applied to match the color distribution between source and target scenes. By just showing the target scene as an example, our previously reported LPT-PCM model can transfer the MA of target to that of source without any a priori information. However, this model had drawbacks such as changes in appearance depending on the background margins and unpredictable results for the scenes consisting of multiple color clusters. This article explores measures to overcome such drawbacks and discusses the applicability of proposed LPT-PCM. Finally we propose a new numerical index to evaluate the similarity between the target and the transferred with the examined samples.

Objects in nature have diverse appearances, and appearance is one of the elements constituting the unique visual aspect of an object. However, previous studies have shown that when an object is represented as a digital image, its appearance can change compared to the real object depending on the material. In this study, the focus was on the “bumpiness” of the object surface, whereby a method was proposed to edit the bumpiness of the object in the image. First, the statistics obtained from images were analyzed in relation to the sensibility value of the bumpiness perceived by humans. Since the statistics on the original image could not fully explain the perception of bumpiness, analyses were conducted on multiscale images. The results suggested that the mean, standard deviation, and top, defined as the average of the luminance values within the top 10 [percentile], were highly influential as cues for bumpiness perception. The results also indicated that the components in the range of 5–65 [cycles per image-width] were highly influential. Based on these analysis results, a method is proposed to edit bumpiness perception by modulating components in the low and medium frequency bands. The effectiveness of the proposed method is demonstrated by image modulation experiments on objects of various materials.

Sparkle from snow is a common phenomenon in Nature but not well studied in the literature. We perform a statistical study on digital snow images captured in-situ to analyze sparkle events by using contrast and density of sparkle spots descriptors. The method for measuring sparkles by Ferrero et al. is adapted, tested, and verified to the case of snow. The dataset is divided into three categories representing the type of snow acquired: dense snow, fresh snow, and old snow. Our analysis highlights the link between the sparkle of snow, the nature of snow, and its grain structure. Sparkle could thus be a feature used for snow classification.

Spatially varying bidirectional reflectance distribution functions (SVBRDFs) play an important role in appearance modeling of real-world surfaces. Automatic capture of these surface properties is highly desirable, but many techniques only partially capture these properties or use complicated setups to do so. Micro surface roughness variations are especially difficult to capture using image-based methods. In this paper, we propose a novel approach towards estimating the complete SVBRDFs of surfaces using a portable projector-camera system made of standard consumer-grade components. Our approach uses insights about the relations between the illumination and viewing geometry and captured image statistics to estimate surface reflectance properties. Our technique should be of great value to practitioners seeking to model and render the geometric and reflectance properties of complex real-world surfaces.

Virtual production stages with LED walls utilize illumination, display, and camera equipment which was not designed with this use case in mind. Because the spectral sensitivity of a camera is different from a human observer, a device specific calibration is required. Furthermore, the illumination spectrum emitted by the display contains large gaps in the cyan and yellow wavelength ranges and is dissimilar to the light sources for which cameras are designed. This causes object colors to be reproduced by the camera in an unnatural manner, making cinematographers hesitant to use LED walls as their primary light source.

In this paper, a display calibration and camera color correction workflow for LED wall virtual production stages is proposed. A linear color correction matrix and the spectrum of a multi-channel LED fixture are jointly optimized to better reproduce object colors simultaneously illuminated by an LED display and the multi-channel fixture as they would appear under high CRI (Color Rendering Index) light sources. An alternative color correction method using root polynomials is found to further improve color reproduction. It is shown that the camera’s response to the display can be characterized by a linear 3×3 matrix and the display can be calibrated using the inverse of the color correction, allowing for a color accurate reproduction of a virtual environment.

In recent years, image processing has proven to be a great tool to help document, preserve, and restore art pieces, especially visual art. One example is using colorization techniques when the original color information of the image has been lost or is unavailable. To expand on that, we used Ghent Altarpiece as the study case. One of the panels of this polyptych has been lost, and only an archival photograph exists. Using the state-of-the-art colorization method, we want to digitally restore what has been lost in its original form. In this work, we proposed a pipeline that consists of a colorization transformer (ColTran) trained on the captured patch-based imaging data of the Ghent Altarpiece. We evaluated the proposed pipeline and addressed its strengths and weaknesses. Moreover, we presented planned future steps and improvements for this project.

Spot photometers measure the luminance that is emitted or reflected from a small surface area in a physical environment. Because the measurement is limited to a “spot,” capturing dense luminance readings for an entire environment is impractical. In this paper, we provide preliminary results demonstrating the potential of using an off-the-shelf commercial camera to operate as a 360° luminance meter. Our method uses the Ricoh Theta Z1 camera, which provides a full 360° omnidirectional field of view and an API to access the camera’s minimally processed RAW images. Working from the RAW images, we describe a calibration method to map the RAW images under different exposures and ISO settings to luminance values. By combining the calibrated sensor with multi-exposure high-dynamic-range imaging, we provide a cost-effective mechanism to capture dense luminance maps of environments. Our results show that our luminance meter performs well when validated against a significantly more expensive spot photometer.

In this paper, we study individual quality scores given by different observers for various image distortions (saturation, contrast, and color quantization) at different levels. We created a database that contains a total of 232 images, derived from 21 pristine images, three distortions, and five levels. The database was rated by 31 participants collected through an online platform. The study shows that observers have distinguishable patterns with respect to different distortions. Using quadratic regression models, we visualized the behavior patterns of different groups of observers. The database and the individual scores collected are publicly available and can be further used for quality assessment research.

We propose a new anisotropic diffusion process for removing noise from MRI images without distorting the edges. The method is based on a simple principle: any diffusion that increases a gradient at neighbouring pixels should be prohibited. From this principle, we deduce an inequality that allows diffusion along the edges but not across them. We introduce promising results using synthetic data with various types of noise as well as real MRI scans.

We consider the problem of estimating surface-spectral reflectance with a smoothness constraint from image data. The total variation of a spectral reflectance over the visible wavelength range is defined as the measure of smoothness. A penalty on roughness, equivalent to smoothness, is added to the performance index to estimate the spectral reflectance functions. The optimal estimates of the spectral reflectance functions are determined to minimize a total cost function consisting of the estimation error and the roughness of the spectral functions. An RGB camera and multiple LED light sources are used to construct the multispectral image acquisition system. We model the observed images using spectral sensitivities, illuminant spectrum, unknown spectral reflectance, a gain parameter, and an additive noise term. The estimation algorithms are developed for the two estimation methods of PCA and LMMSE. The optimal estimators are derived based on the least-square criterion for PCA and the mean squared error minimization criterion for LMMSE. The feasibility of the proposed method is shown in an experiment using three mobile phone cameras. It is confirmed that the optimal estimators improve the accuracy for both original PCA and LMMS estimators.

In this study, we used a custom-built Optical See-Through Augmented Reality (OST AR) system to conduct a psychophysical experiment to determine the preferred gamma and black level for high naturalness perception in OST-AR. We utilized 6 different fruit stimuli and 11 different backgrounds to do this experiment. We used two-way ANOVA to analyze the data and concluded that only the effect of different fruits on virtual objects’ gamma preference for high naturalness is considered statistically significant. Surprisingly, all ANOVA analyses indicate background’s color does not contribute to observers’ gamma or black level preference for naturalness. We found that gamma preference has a strong correlation with the average lightness of the virtual stimuli. There is no clear correlation between chroma, hue, and gamma preference in terms of naturalness perception. This finding suggests that the background can be ignored in future imaging pipelines emphasizing high naturalness appearance in Augmented Reality.

Colors, generally, have effects on human interpretations that can manifest in a variety of emotions, reactions, and behaviors. The objective of this study is to understand the reasoning behind the choices of pill colors in relation to expected efficacy of drugs, as well as the color associations made by participants. The research was conducted at several university campuses in USA, UAE, Croatia, Kosovo, and China, and focused on different age brackets, gender, ethnic backgrounds, educational levels, and pill usage frequency. Understanding the reasoning and color associations helps us better comprehend the expected efficacy of drugs, and can therefore support pharmaceutical companies in designing and manufacturing drugs, thereby maximizing the potential effect on patients’ adherence rates.

Many image-editing tasks are carried out in the gradient domain. Suppose that for an image I the gradient ∇I consists of a pair of fields (p;q); then some image “reintegration” scheme is tasked with converting derivative fields (p;q) back to imagespace I; typically, a Poisson equation solver is used for this task. But what if we have altered (p;q) so that this pair (p;q) is no longer integrable? Then we have to project onto integrable gradient data that will indeed reintegrate to an approximation of the original image. For example, we may wish to alter (p;q) so as to emphasize or de-emphasize some aspects of the image, e.g. ameliorating wrinkles in skin images (or indeed enhancing them in the case of ageing a face image).

Here, we propose a new gradient kernel that retains part of the original image, regularising the reintegration back into the image domain. We compare our approach with the Screened-Poisson method which includes a term λ times a “screen” term that moves the solution image back closer to the input image. Effectively, we are doing a similar adjustment, but we show that the results are a good deal better than using Screened-Poisson, which tends to overly blur the output. Moreover, in Screened-Poisson one must choose a value for λ, which may be different for every image – here we determine that our new kernel method does not need to adapt to each image yet delivers comparable or better results.

The use of polarization while trying to keep the digital color reproduction accuracy at its finest is very challenging due to how polarization is interacting and affecting the light spectrum itself and due to the quality of the used polarization materials. Our study on RGB imaging and color reproduction’s fidelity with and without polarization shows that a cross circular polarization (on a camera lens and light source) will have a major impact on how a linear grayscale, whether it has a semi-glossy or matt finishing, would be reproduced in contrast to no polarization at all. A major loss in deep black shades in the case of a semi-glossy grayscale is unmistakable. In addition to a noticeable shift in both lightness and Chroma components regardless of the grayscale’s finishing but depending rather on the used color target for correction. DE00 could not paint the full picture about color fidelity despite its low conformant reported values. Whereas, a closer visual inspection of the color components separately (lightness and Chroma) reveals color reproduction problems caused by polarization.

Colour correction is the process of converting camera dependent RGB values to a camera independent standard colour space such as CIE XYZ. Regression methods — linear, polynomial, and root-polynomial least-squares — are traditionally used to solve for the colour correction transform. More recently neural net solutions for colour correction have been developed.

This paper begins with the observation that the neural net solution — while delivering better colour correction accuracy compared to the simple (and widely deployed) 3×3 linear correction matrix approach — is not exposure invariant. That is to say, the network is tuned to mapping RGBs to XYZs for a fixed exposure level and when this exposure level changes, its performance degrades (and it delivers less accurate colour correction compared to the 3x3 matrix approach which is exposure invariant). We develop two remedies to the exposure variation problem. First, we augment the data we use to train the network to include responses for many different exposures. Concomitantly, the trained network is robust to a changing exposure. Second, we redesign the network so, by construction, it is exposure invariant.

Experiments demonstrate that, by adopting either approach, Neural Network colour correction can be made exposure invariant.

An image dataset of multiple-illumination scenes with three forms of ground-truths, including human perceptual ground truths, is presented, together with benchmarking results of single- and multi-illumination color constancy algorithms. The results demonstrate the need for new benchmarks for color correction algorithms based on the spectral and spatial non-uniformities of chromatic adaptation in complex scenes.

An object’s color is affected by the color of the light incident upon it, and the illuminant-dependent nature of color creates problems for convolutional neural networks performing tasks such as image classification and object recognition. Such networks would benefit from illuminant-invariant representation of the image colors. The Laplacian of the logarithm of the image is introduced as an effective color invariant. Applying the Laplacian in log space makes the input colors approximately illuminationinvariant. The illumination invariance derives from the fact that finite-difference differentiation in log space is equivalent to ratios of neighboring pixels in the original space. For narrow-band sensors, rationing neighboring pixels cancels out their shared illumination component. The resulting color representation is no longer absolute, but rather is a relative color representation. Testing shows that when using the Laplacian of the logarithm as input to a Convolutional Neural Network designed for classification its performance is: (i) approximately equal to that of the same network trained on sRGB data under white light, and (ii) largely unaffected by changes in the illumination.

Smart image editing is drawing attention and a wide range of edit operations have been investigated. We address the problem of creating new image versions where light conditions and object colors can be altered while maintaining physical coherence across the scene. We propose a baseline framework comprised of a surreal dataset with a large Ground-Truth on light effects and a set of basic deep architectures relying on intrinsic decomposition. Our proposal is evaluated for image relighting and outperforms the state-of-the-art on the previous VIDIT dataset. The codes and datasets are available: https://github.com/ liulisixin/ImageEditingSI

Planck’s law and Wien’s approximation of this law are both widely used to calculate the spectral radiation of a black-body based on its color temperature. The Wien approximation is a slightly simpler equation and has the advantage that the logarithm of a Wien light can be written as a linear sum of two basis vectors plus an offset (a fact that is exploited in some computer vision algorithms).

In this paper, we show that the Wien formulation can, in general, be used to approximate Planckian lights assuming there is a mapping function taking Planckian to corrected Wien temperature. Significantly, we show that a correction function f() exists and for the range of color temperatures of interest the Wien spectrum calculated for f(T) has a very similar shape to the actual spectrum of a Planckian light with temperature T. We find that defining f() as a polynomial-type function models to a good extent the relationship between the color temperatures of Planckians and their closest Wien-Planckians lights both in terms of the angular error between their two respective spectral functions and their projections to u’v’ coordinates.

Illuminant estimation is critically important in computational color constancy, which has attracted great attentions and motivated the development of various statistical- and learning-based methods. Past studies, however, seldom investigated the performance of the methods on pure color images (i.e., an image that is dominated by a single pure color), which are actually very common in daily life. In this paper, we develop a lightweight feature-based Deep Neural Network (DNN) model—Pure Color Constancy (PCC). The model uses four color features (i.e., chromaticity of the maximal, mean, the brightest, and darkest pixels) as the inputs and only contains less than 0.5k parameters. It only takes 0.25ms for processing an image and has good cross-sensor performance. The angular errors on three standard datasets are generally comparable to the state-of-the-art methods. More importantly, the model results in significantly smaller angular errors on the pure color images in PolyU Pure Color dataset, which was recently collected by us.