Developing a sortable color palette through geometric approaches

Abstract

Color is the primary visual element that is initially perceived in media, and humans use it to express and convey their thoughts and emotions. A color palette is composed of a limited number of colors and is frequently utilized in a variety of visual industries, such as design, fashion, cosmetics, architecture, art, etc. Rather than using a single color, it is preferable to employ a color palette comprised of numerous colors. This is because richer semantics may be transmitted through the interaction between colors as well as their distinct properties. This dissertation proposes three approaches to color palette processing that can be applied to various visual media. First, we introduce an automated color palette extraction from movies, which are heavily edited videos in order to redeem the previous simple color extraction. Second, we introduce a flow-based color sorting algorithm for the intuitive visualization of multiple palettes. Third, we introduce a palette similarity measurement to facilitate the comparison and retrieval of palettes from various visual sources. Additionally, we demonstrate several improved applications of color palettes such as palette interpolation, retrieval, and image recoloring based on the provided palette processing methods.