Unique hue data for colour appearance models. Part I: Loci of unique hues and hue uniformity

Kaida Xiao, Sophie Wuerger, Chenyang Fu, Dìmosthenis KaratzasColor 2010 · DOI

TL;DR

This paper presents new psychophysical data on "unique hues" (the pure red, yellow, green, and blue sensations) from a large group of observers. They use this data to evaluate CIECAM02, a leading color appearance model, finding that its built-in unique hue predictions don't match their data and that the model struggles with "hue uniformity" (how perceived hue changes with lightness and saturation).

What Problem Does This Paper Address?

Accurately predicting how colors appear to humans under different viewing conditions is crucial for applications like cross-media color reproduction (e.g., ensuring a digital image looks the same on a screen, in print, and on a billboard). Color appearance models, like CIECAM02, aim to do this by translating physical color measurements (like RGB values) into perceptual attributes like lightness, chroma (saturation), and hue.

A key aspect of these models is their ability to represent "unique hues" – the four fundamental color sensations of red, yellow, green, and blue that are perceived as pure, without any admixture of another hue. These unique hues serve as important reference points. However, the specific hue angles used in CIECAM02 are based on the Natural Color System (NCS), and it's unclear if these accurately reflect how people actually perceive these pure hues, especially when considering variations in lightness and chroma. This paper addresses this by collecting new, large-scale psychophysical data on unique hues and using it to rigorously test CIECAM02's predictions and its assumption of hue uniformity.

Methods

The researchers conducted psychophysical experiments with 185 color-normal observers. They used a calibrated CRT display to present color patches. Observers were asked to identify the "unique hue" for red, yellow, green, and blue by selecting the patch that best represented that pure sensation, free from any admixture of other hues. This was done across various combinations of lightness and chroma.

The collected data (in CIE XYZ tristimulus values) was then transformed into CIECAM02's perceptual attributes (lightness, chroma, and hue angle). The researchers used this to:

  1. Evaluate the loci of unique hues: They compared the hue angles derived from their experimental data with the hue angles defined in CIECAM02 (which are based on NCS).
  2. Test hue uniformity: They analyzed how the perceived hue angle changed as lightness and chroma varied for each of the four unique hues within the CIECAM02 framework. This involved statistical analysis (ANOVA) and plotting hue angles against chroma and lightness.

Datasets

The paper introduces a newly collected psychophysical dataset of unique hue judgments from 185 color-normal observers. This dataset consists of unique hue settings across various lightness and chroma levels, measured on a CRT display under dark viewing conditions. This dataset is intended to serve as a benchmark for evaluating color appearance models. The paper also references existing datasets and models, such as the Natural Color System (NCS) hues, for comparison.

Key Insights and Ideas

  • NCS vs. Perceptual Reality: The study reveals a significant discrepancy between the unique hue angles used in CIECAM02 (derived from NCS) and the unique hue angles reported by their large observer sample. This suggests that the NCS-based unique hues may not be the most accurate representation of perceptual unique hues, particularly for color management applications where precise color specification is critical.
  • CIECAM02's Hue Uniformity Problem: The research demonstrates that CIECAM02 does not exhibit perfect hue uniformity. For most unique hues (yellow, green, and blue), the perceived hue angle changes significantly with variations in lightness and chroma. This is a critical finding because hue uniformity is a desirable property for color appearance models, especially for image reproduction and enhancement tasks where preserving hue is paramount.
  • Unique Red's Resilience: Interestingly, unique red appears to be relatively invariant under changes in lightness, showing better hue uniformity compared to the other unique hues within the CIECAM02 model.

Key Results

  • The hue angles derived from the experimental data were inconsistent with the Natural Color System (NCS) hues used in CIECAM02.
  • Significant disagreements were found in testing hue uniformity for CIECAM02 across all four unique hues, with unique red being the exception, showing invariance under lightness changes.
  • Observer variability (measured in CIEDE2000 units) was found to be lower than previously reported in some studies, suggesting a relatively reliable dataset. Inter-observer variability was highest for red (2.30 ΔΕ00), while intra-observer variability was highest for yellow (1.07 ΔΕ00).

Limitations and Future Work

  • CRT Display Limitations: The experiments were conducted on a CRT display. While calibrated, CRTs have inherent limitations compared to emissive displays or real-world objects, which might influence the generalizability of the findings to all viewing media.
  • Dark Viewing Conditions: The study focused on dark viewing conditions. The authors acknowledge that ambient illumination can affect unique hue settings, and this was part of a more extensive series of experiments not fully detailed in this paper.
  • Model Modification: The authors conclude that CIECAM02 needs to be modified to accurately reflect the hue data of color-normal observers, particularly concerning the loci of unique hues and hue uniformity. Future work would involve developing and testing revised versions of the model.
  • Data Availability: The paper states that the unique hue data (CIE XYZ) will be made available, which is a valuable contribution for future research and model development.

Summary updated

Abstract

Abstract Psychophysical experiments were conducted to assess unique hues on a CRT display for a large sample of colour‐normal observers (n = 185). These data were then used to evaluate the most commonly used colour appearance model, CIECAM02, by transforming the CIEXYZ tristimulus values of the unique hues to the CIECAM02 colour appearance attributes, lightness, chroma and hue angle. We report two findings: (1) the hue angles derived from our unique hue data are inconsistent with the commonly used Natural Color System hues that are incorporated in the CIECAM02 model. We argue that our predicted unique hue angles (derived from our large dataset) provide a more reliable standard for colour management applications when the precise specification of these salient colours is important. (2) We test hue uniformity for CIECAM02 in all four unique hues and show significant disagreements for all hues, except for unique red which seems to be invariant under lightness changes. Our dataset is useful to improve the CIECAM02 model as it provides reliable data for benchmarking. © 2010 Wiley Periodicals, Inc. Col Res Appl, 2011
BibTeX
@inproceedings{xiao2010unique,
  author = {Kaida Xiao and Sophie Wuerger and Chenyang Fu and Dìmosthenis Karatzas},
  title = {{Unique hue data for colour appearance models. Part I: Loci of unique hues and hue uniformity}},
  booktitle = {Color},
  year = {2010},
  doi = {10.1002/col.20637},
}
Figure