Conference article

This paper presents a statistics-based interactive evolutionary computation (IEC) method for color scheme search. Color schemes are utilized in an enormous range of items such as websites; clothing; advertising media; and housewares. However; people who do not have sufficient skill or knowledge of colors need to devote considerable time and effort to a creating color scheme. Currently; artists’ color schemes are freely available from websites. However; obtaining an appropriate color scheme from a large data set is difficult for novice users. To overcome this problem; we rely on a statistics-based interactive genetic algorithm (IGA). Use of this IGA is expected to reduce computing costs compared with conventional IEC approaches and to take overall color scheme impressions into account. These contributions enable to realization of the kansei-based color search system in real time. In addition; we introduce four similarity search (SS) functions (hue; saturation; brightness; and color differences) to facilitate the convergence of a color scheme search. The combination of a statistics-based IGA and four SS functions allows users to easily and effectively find their desired color schemes. To investigate the performance of the proposed method; we conducted two experiments and confirmed that the implemented application allows users to obtain a desired color scheme in less than 48 s. In addition; we also confirmed that the proposed method can provide some favorable recolored illustrations in less than 52 s.

Color Scheme Search; Interactive Evolutionary Computation; Statistics; Color Transfer