Determining key colors from a design perspective using dE‐means color clustering

• Published in: Color research and application 2023-01, Vol.48 (1), p.69-87
• Main Authors: Gijsenij, Arjan, Vazirian, Marjan, Spiers, Peter, Westland, Stephen, Koeckhoven, Pim
• Format: Article
• Language: English
• Subjects: Algorithms; Clustering; Color; color design; color palettes; color themes; Design; design styles; Mathematical analysis; mood board images
• ISSN: 0361-2317; 1520-6378
• DOI: 10.1002/col.22817

Designing a mood board is a creative tool designers often employ at the start of a new (interior) design project. The colors of a mood board represent important information about the envisioned interior designs. In this article, we focus on digital mood boards and we propose a novel method to determine the mood boards' key colors from a design perspective using dE‐means color clustering. This proposed algorithm includes a fixed initialization to overcome the non‐deterministic nature of traditional k‐means, and a merging step to ensure all colors in the calculated palette are at least a dE‐threshold apart. Quantitative and qualitative results demonstrate that the proposed algorithm outperforms existing methods in determining the key colors from a design perspective. Additionally, we introduce a novel method to visualize the calculated color palettes that takes into account the contribution of individual pixels to the construction of the calculated color palettes. Finally, we demonstrate how the proposed algorithm can be used to characterize color palettes for various design styles like modern, industrial and art deco. This figure shows an example mood board and the key colors from a design perspective, as extracted by the proposed method.