Generating Harmonious Colors through the Combination of n-Grams and K-means

Abstract

Among the many approaches to studying color harmony tried so far, a relatively recent method is to leverage a large number of human-created and ranked color palettes, such as those hosted at colourlovers.com. Analysis of these large datasets could provide insights into the nature of color harmony but is usually overwhelming because of the sheer number of slightly differing colors. It is possible to quantize the colors in these color palettes to a manageable set of discrete colors without significantly affecting the harmony perception of the palette. Considering the quantized colors as words and the palettes as sentences, it is possible to form and compute the probabilities of n-Grams in the sentences. In this study, we create bigrams and trigrams from the corpus of highly ranked color palettes and use them to predict new color combinations. Respondents were asked to like or dislike the patterns colored with these color combinations. It was found that the new color combinations thus formed were almost as harmonious and pleasing as the originals.

Unlocking Visual Attraction: The Subtle Relationship between Image Features and Attractiveness

“…Through the nuanced blending of these colors, the entire spectrum becomes visible. This model finds extensive applications in both computer vision and photography [30].…”