The purpose of this study was to develop a more accurate method for analyzing fashion image digital colors in uncertain observation conditions. The standard color values used in this experiment were obtained from 10 pieces of actual clothing provided by their designers. The comparison color values were created from fashion images using human visual perception, software color analysis, and a combination of human visual perception and software color analysis. The other 10 color values were determined using the method developed in this study. The data were processed using Microsoft Excel 2013 and the statistical program R 3.4.2. Descriptive analyses, 1-way anova tests, and a multiple comparisons post-test were conducted using false discovery rates. The results showed that a combination of human and software analysis produced more accurate results than either alone. Delta E-ab* 3-6 was found to be the minimum color difference tolerance between actual clothing and fashion image.