FHSI: Toward More Human-Consistent Color Representation

Abstract

In this paper, we propose a novel approach toward the development of a perceptual color space, FHSI, which stands for “Fuzzy HSI," because it is based on the fuzzification of the well-known HSI color space. FHSI represents a set of fuzzy colors obtained by partitioning the gamut of feasible colors in the HSI model corresponding to standardized linguistic tags. In fact, color categorization was performed on the basis of personal judgments of humans collected by way of an online survey. This approach helps to significantly enhance color matching and similarity searches by producing more intuitive and human-consistent output for users. The introduced method has potential for use in various color image applications involving query processing, for example, in the coordination of online apparel shopping.