A study on subjective evaluations of printed color images

Abstract Fuzzy measures and fuzzy integrals are applied to build an evaluation model of printed color images. First, subjective evaluation data of seven people are collected by pairwise comparisons. The subjects give 16 kinds of evaluations for each color proof. Then a two-layer evaluation model is proposed based on the result of factor analysis. In this study, the authors adopt Choquet's integral as a form of fuzzy integral because it has good properties compared with other forms of fuzzy integrals. A relaxation-method-like procedure has been devised to identify fuzzy measures of the two-layer model. After its effectiveness is confirmed with artificial data, the algorithm is applied to actual subjective evaluation data. This gives us results, revealing that the seven subjects are divided into two groups whose evaluation characteristics are structurally different from each other.