When Color Constancy Goes Wrong: Correcting Improperly White-Balanced Images

@article{Afifi2019WhenCC,
  title={When Color Constancy Goes Wrong: Correcting Improperly White-Balanced Images},
  author={Mahmoud Afifi and Brian L. Price and Scott D. Cohen and M. S. Brown},
  journal={2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2019},
  pages={1535-1544}
}
This paper focuses on correcting a camera image that has been improperly white-balanced. [] Key Method Our method is enabled by a dataset of over 65,000 pairs of incorrectly white-balanced images and their corresponding correctly white-balanced images. Using this dataset, we introduce a k-nearest neighbor strategy that is able to compute a nonlinear color mapping function to correct the image's colors. We show our method is highly effective and generalizes well to camera models not in the training set.

Figures and Tables from this paper

Color Temperature Tuning: Allowing Accurate Post-Capture White-Balance Editing
TLDR
This work proposes an imaging framework that renders a small number of “tiny versions” of the original image, each with different WB color temperatures, that can significantly outperform existing solutions targeting post-capture WB editing.
Three-Color Balancing for Color Constancy Correction
TLDR
A three-color balance adjustment for color constancy correction that not only outperforms white balancing but also has almost the same performance as Cheng’s method with 24 target colors.
Deep White-Balance Editing
  • M. Afifi, M. S. Brown
  • Computer Science
    2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2020
TLDR
A deep neural network (DNN) architecture trained in an end-to-end manner to learn the correct white balance for sRGB images that are rendered with the incorrect white balance is introduced.
How does Color Constancy Affect Target Recognition and Instance Segmentation?
TLDR
A novel augmentation strategy is proposed to address the negative impact of incorrect WB by expanding the training datasets in both color transformation and synthetic occlusion and it is shown that this approach clearly improves the performance of DNNs in detection and segmentation tasks with small objects and objects with heavily occluded backgrounds.
Auto White-Balance Correction for Mixed-Illuminant Scenes
TLDR
A unique departure from conventional AWB, this method does not require illuminant estimation, as is the case in traditional camera AWB modules, and proposes to render the captured scene with a small set of predefined white-balance settings.
A Multi-Hypothesis Approach to Color Constancy
TLDR
This work proposes a Bayesian framework that naturally handles color constancy ambiguity via a multi-hypothesis strategy and provides state-of-the-art accuracy on multiple public datasets (up to 11% median angular error improvement) while maintaining real-time execution.
A Multi-Hypothesis Classification Approach to Color Constancy
TLDR
This work proposes a Bayesian framework that naturally handles color constancy ambiguity via a multi-hypothesis strategy and provides state-of-the-art accuracy on multiple public datasets (up to 11% median angular error improvement) while maintaining real-time execution.
Cross-Camera Convolutional Color Constancy
TLDR
C5 achieves state-of-the-art accuracy for cross-camera color constancy on several datasets, is fast to evaluate on a GPU or CPU, and requires little memory, and thus is a practical solution to the problem of calibration-free automatic white balance for mobile photography.
INTEL-TAU: A Color Constancy Dataset
TLDR
This dataset, called INTEL-TAU, contains 7022 images in total, which makes it the largest available high-resolution dataset for illumination estimation research, and is coherent with the new General Data Protection Regulation (GDPR).
What Else Can Fool Deep Learning? Addressing Color Constancy Errors on Deep Neural Network Performance
  • M. Afifi, M. Brown
  • Computer Science
    2019 IEEE/CVF International Conference on Computer Vision (ICCV)
  • 2019
TLDR
This paper explores how strong color casts caused by incorrectly applied computational color constancy - referred to as white balance (WB) in photography - negatively impact the performance of DNNs targeting image segmentation and classification and proposes a novel augmentation method that can emulate accurate colorconstancy degradation.
...
...

References

SHOWING 1-10 OF 46 REFERENCES
Beyond White: Ground Truth Colors for Color Constancy Correction
TLDR
It is shown that under certain illuminations, a diagonal 3*3 matrix is capable of correcting not only neutral colors, but all the colors in a scene.
Combining gray world and retinex theory for automatic white balance in digital photography
  • E. Lam
  • Computer Science
    Proceedings of the Ninth International Symposium on Consumer Electronics, 2005. (ISCE 2005).
  • 2005
TLDR
This paper presents an effective technique that combines Gray world assumption and retinex theory together, while preserving the strength of the two methods, and shows that this approach is a viable alternative to the two existing methods.
Convolutional Color Constancy
  • J. Barron
  • Computer Science
    2015 IEEE International Conference on Computer Vision (ICCV)
  • 2015
TLDR
The problem of color constancy is reformulate as a 2D spatial localization task in a log-chrominance space, thereby allowing us to apply techniques from object detection and structured prediction to the colorconstancy problem.
Illuminant estimation for color constancy: why spatial-domain methods work and the role of the color distribution.
TLDR
It is shown that spatial information does not provide any additional information that cannot be obtained directly from the color distribution and that the indirect aim of spatial-domain methods is to obtain large color differences for estimating the illumination direction.
Color Homography: Theory and Applications
TLDR
The surprising result that homographies are the apposite tool for relating image colors of the same scene when the capture conditions—illumination color, shading and device—change is shown.
Shades of Gray and Colour Constancy
TLDR
It is shown that Max-RGB and Grey-World are two instantia-tions of Minkowski norm, and that for a large cali-brated dataset L6 norm colour constancy works best over-all (the authors have improved the performance achieved by a sim-ple normalization based approach).
Bayesian color constancy revisited
TLDR
This paper introduces a new tool in the form of a database of 568 high-quality, indoor and outdoor images, accurately labelled with illuminant, and preserved in their raw form, free of correction or normalisation, which shows that automatic selection of grey-world algorithms according to image properties is not nearly so effective as has been thought.
Color Correction Using Root-Polynomial Regression
TLDR
This paper proposes a new polynomial-type regression loosely related to the idea of fractional polynomials which it is called root-PCC (RPCC), a simple (low complexity) extension of LCC that enhances color correction performance on real and synthetic data.
Modeling Radiometric Uncertainty for Vision with Tone-Mapped Color Images
TLDR
It is observed that this uncertainty varies substantially across color space, making some pixels more reliable than others in tone-mapping, and a model for this uncertainty and a method for fitting it to a given camera or imaging pipeline are introduced.
Color Constancy
  • M. Ebner
  • Art, Computer Science
    Computer Vision, A Reference Guide
  • 2014
TLDR
This book provides a comprehensive introduction to the field ofcolor constancy, describing all the major color constancy algorithms, as well as presenting cutting edge research in the area of color image processing.
...
...