Perceptual analysis of distance measures for color constancy algorithms

Gijsenij, Arjan; Gevers, Theo; Lucassen, Marcel P.

doi:10.1364/josaa.26.002243

Cited by 83 publications

(73 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Any mismatch in the illuminant estimation will thus show up as a global color cast in the corrected image, whose color fidelity can be measured with the triad comparison method used in this study. That would help to perceptually evaluate the different algorithms, as reported previously [22]. We have shown in this paper that, for ellipsoidal chromatic distributions, illuminant changes are perceptually less well noticed when the direction of the illuminant change is parallel to the major axis of the chromatic distribution.…”

Section: Improvements To Color Constancy Algorithmssupporting

confidence: 68%

“…unique pairs]. We have previously used the triad comparison method in studies for perceptual evaluation of color constancy algorithms [22,39]. In those studies, the observers had the same task as in our study, namely, to indicate which of the two images in the bottom row best resembled the image on the top row.…”

Section: B Improvements In Methodologymentioning

confidence: 99%

“…In line with previous studies [21,22] we select a neutral reference illuminant (D65) and four chromatic illuminants, all composed with the CIE basis functions for spectral variations in natural daylight. In [21] it is investigated whether color constancy would be better for illuminant changes along the daylight locus than for illuminant changes perpendicular to it, but no experimental evidence was found that unambiguously supports this hypothesis.…”

Section: F Illuminants That Induce Equal Changes In the Reflected LImentioning

confidence: 99%

See 2 more Smart Citations

Effects of chromatic image statistics on illumination induced color differences

et al. 2013

Self Cite

View full text Add to dashboard Cite

We measure the color fidelity of visual scenes that are rendered under different (simulated) illuminants and shown on a calibrated LCD display. Observers make triad illuminant comparisons involving the renderings from two chromatic test illuminants and one achromatic reference illuminant shown simultaneously. Four chromatic test illuminants are used: two along the daylight locus (yellow and blue), and two perpendicular to it (red and green). The observers select the rendering having the best color fidelity, thereby indirectly judging which of the two test illuminants induces the smallest color differences compared to the reference. Both multicolor test scenes and natural scenes are studied. The multicolor scenes are synthesized and represent ellipsoidal distributions in CIELAB chromaticity space having the same mean chromaticity but different chromatic orientations. We show that, for those distributions, color fidelity is best when the vector of the illuminant change (pointing from neutral to chromatic) is parallel to the major axis of the scene's chromatic distribution. For our selection of natural scenes, which generally have much broader chromatic distributions, we measure a higher color fidelity for the yellow and blue illuminants than for red and green. Scrambled versions of the natural images are also studied to exclude possible semantic effects. We quantitatively predict the average observer response (i.e., the illuminant probability) with four types of models, differing in the extent to which they incorporate information processing by the visual system. Results show different levels of performance for the models, and different levels for the multicolor scenes and the natural scenes. Overall, models based on the scene averaged color difference have the best performance. We discuss how color constancy algorithms may be improved by exploiting knowledge of the chromatic distribution of the visual scene.

show abstract

Section: Improvements To Color Constancy Algorithmssupporting

confidence: 68%

Section: B Improvements In Methodologymentioning

confidence: 99%

Section: F Illuminants That Induce Equal Changes In the Reflected LImentioning

confidence: 99%

See 1 more Smart Citation

Effects of chromatic image statistics on illumination induced color differences

et al. 2013

Self Cite

View full text Add to dashboard Cite

show abstract

“…Plus, the maximum magnitude of velocity  m ax was set at 0.1. The proposed method was then evaluated based on the angular error between the estimated illuminant and the ground truth illuminant [27] .…”

Section: Experimental Evaluationmentioning

confidence: 99%

Illuminant Chromaticity Estimation via Optimization of RGB Channel Standard Deviation

Subhashdas¹,

Yoo²,

Ha³

2016

Journal of the Institute of Electronics and Information Enginee

View full text Add to dashboard Cite

The primary aim of the color constancy algorithm is to estimate illuminant chromaticity. There are various statistical-based, learning-based and combinational-based color constancy algorithms already exist. However, the statistical-based algorithms can only perform well on images that satisfy certain assumptions, learning-based methods are complex methods that require proper preprocessing and training data, and combinational-based methods depend on either pre-determined or dynamically varying weights, which are difficult to determine and prone to error. Therefore, this paper presents a new optimization based illuminant estimation method which is free from complex preprocessing and can estimate the illuminant under different environmental conditions. A strong color cast always has an odd standard deviation value in one of the RGB channels. Based on this observation, a cost function called the degree of illuminant tinge(DIT) is proposed to determine the quality of illuminant color-calibrated images. This DIT is formulated in such a way that the image scene under standard illuminant (d65) has lower DIT value compared to the same scene under different illuminant. Here, a swarm intelligence based particle swarm optimizer(PSO) is used to find the optimum illuminant of the given image that minimizes the degree of illuminant tinge. The proposed method is evaluated using real-world datasets and the experimental results validate the effectiveness of the proposed method. Keywords Ⅰ. IntroductionIn the real world, the perceived color of an object relies on the surface reflectance and light source color that illuminate the scene. However, the human eye can perceive an object color to be same, even when there is a change in the illuminant color. This ability of the human visual system to discount the influence of illuminant color is referred to as color constancy. In a computational approach, an implied assumption is that the influence of light source colorJournal of The Institute of Electronics and Information Engineers Vol.53, NO.6, June 2016 is static throughout a scene. Therefore, the scene illuminant is estimated first and the scene colors are then corrected using a chromatic adaptation model.The main objective of a color constancy algorithm is to estimate the scene illuminant. However, illuminant estimation is very difficult process and various approximations have been made on physical world scenes to cope up with that. State-of-the-art color constancy approaches can be categorized into three groups: 1) statistical-based methods, 2) learningbased methods, and 3) combinational-based methods [1～2] . The first type of algorithm is based on low level statistics or a physics based reflection model, the second type algorithm makes use of information extracted from the training phase to estimate the light source color, and the third type is formulated by combining statistical-based algorithms or choosing a superlative algorithm for a given image.The Max-RGB or white patch retinex algorithm is a well known statis...

show abstract

“…However, later it was shown [21] that the mean error does not provide an accurate statistical summary of the underlying error distribution and the median error was proposed instead. In a perceptual evaluation [17,19] of illuminant estimation algorithms the trimean error was introduced which gives the additional values of the error distribution (e.g. the top and bottom 25%).…”

Section: Introductionmentioning

confidence: 99%