Color identification method for fashion runway images: An experimental study

Abstract

AbstractFashion runway images are one of the most important sources for color research. In this study, an experimental study was conducted to develop the optimal color identification method for fashion images under uncertain conditions. For this purpose, five different experimental schemes were designed for comparison. Experts' visual assessment, PANTONE Color Palettes Generator (PCPG), and K‐means method are the existing methods. Experts' visual assessment combined with PCPG and experts' visual assessment combined with K‐means method are the new method developed in this study. The color measuring instrument extracted the color of the standard sample by the actual clothing. Five methods extracted the color of the contrast sample from the clothing image, and the color difference was calculated. The data were processed using Microsoft Excel and the statistical program SPSS23.0. Descriptive analyses, One‐way ANOVA and least‐significant difference multiple comparisons analysis were conducted to analyze the color difference data. The results of data analysis showed that experts' visual assessment combined with K‐means method was the most accurate method with ΔE*ab 1.41–4.58, mean <4, and SD <1. The experimental results were verified and the accurate method was stable. These results are of great importance for the research of fashion colors. They help to improve the accuracy of fashion color database.