COLIBRI Fuzzy Model: Color Linguistic-Based Representation and Interpretation

Colors are omnipresent in today’s world and play a vital role in how humans perceive and interact with their surroundings. However, it is challenging for computers to imitate human color perception. This paper introduces the Human Perception-Based Fuzzy Color Model, COLIBRI (Color Linguistic-Based Representation and Interpretation), designed to bridge the gap between computational color representations and human visual perception. The proposed model uses fuzzy sets and logic to create a framework for color categorization. Using a three-phase experimental approach, the study first identifies distinguishable color stimuli for hue, saturation, and intensity through preliminary experiments, followed by a large-scale human categorization survey involving more than 1000 human subjects. The resulting data are used to extract fuzzy partitions and generate membership functions that reflect real-world perceptual uncertainty. The model incorporates a mechanism for adaptation that allows refinement based on feedback and contextual changes. Comparative evaluations demonstrate the model’s alignment with human perception compared to traditional color models, such as RGB, HSV, and LAB. To the best of our knowledge, no previous research has documented the construction of a model for color attribute specification based on a sample of this size or a comparable sample of the human population (n = 2496). Our findings are significant for fields such as design, artificial intelligence, marketing, and human-computer interaction, where perceptually relevant color representation is critical.

💡 Research Summary

The paper introduces COLIBRI (Color Linguistic‑Based Representation and Interpretation), a fuzzy‑logic‑driven color model that explicitly aligns computational color representation with human visual perception and linguistic categorization. Recognizing that traditional models such as RGB, HSV, and CIELAB treat color as a set of fixed numerical coordinates, the authors argue that these models fail to capture the gradual, context‑dependent, and language‑influenced nature of how people actually perceive and name colors.

To build COLIBRI, the authors conducted a three‑phase experimental program. Phase 1 involved preliminary psychophysical tests to determine the smallest just‑noticeable differences in hue, saturation, and intensity. Using these thresholds, a set of distinguishable color stimuli was generated within the HSI space, chosen for its semantic correspondence to human descriptors (Hue, Saturation, Intensity). Phase 2 was a large‑scale human categorization survey with 2,496 participants (including gender balance and a subset of color‑blind individuals). Participants were shown 1,071 color patches and asked to provide free‑form names as well as select from a predefined list, allowing the collection of both spontaneous and constrained linguistic data. The responses were aggregated per color channel, yielding empirical probability distributions for each linguistic category.

From these distributions the authors derived fuzzy membership functions. Rather than relying on simple triangular or Gaussian shapes, they employed a hybrid of trapezoidal and conical functions that better fit the observed non‑linear spread of human judgments, especially around fuzzy boundaries such as “blue‑cyan” or “red‑orange”. Phase 3 introduced an adaptive mechanism: a neural‑network‑based feedback loop continuously refines the fuzzy partitions as new data arrive, thereby accommodating contextual factors like lighting conditions, cultural background, or individual differences.

Evaluation compared COLIBRI against RGB, HSV, and CIELAB on a color‑discrimination task where the ground truth was the human response distribution. Mean‑square error (MSE) between model predictions and human data was substantially lower for COLIBRI across all three dimensions. The advantage was most pronounced at linguistic transition zones, where traditional models produce abrupt “step” changes, whereas COLIBRI yields smooth membership gradients that mirror human perception.

Additional analyses explored gender effects and color‑vision deficiencies. The dataset revealed statistically significant gender‑based naming patterns (e.g., males more often used “blue” while females favored “cyan” or “turquoise”), and color‑blind participants exhibited broader membership functions in low‑saturation regions, confirming that COLIBRI can model perceptual variability across sub‑populations.

Beyond the core model, the authors propose a new family of “Soft Color Models” in which a single color may belong simultaneously to multiple linguistic categories with varying degrees of membership. This paradigm shift opens avenues for applications that require nuanced color reasoning, such as design automation, marketing palette selection, and human‑computer interaction interfaces where verbal color instructions are common.

In summary, COLIBRI represents a significant methodological advance: it combines a massive, demographically diverse human dataset with sophisticated fuzzy set theory and adaptive learning to produce a color model that is demonstrably closer to human perception than any existing standard. The paper outlines future work on real‑time video processing, cross‑cultural extensions, and integration with device‑dependent color management pipelines, positioning COLIBRI as a foundational tool for perceptually aware color computing.