2009 Volume 8 Issue 3 Pages 643-649
We have been developing a design support system to assist novice users as well as designers to suit to their taste or to those of expected consumers. Color coordination is considered to be an important factor to control the image of products, which means there is a strong relationship between a color coordination and imagewords. To accept wide variety of imagewords, we need a large number of color coordination as training examples. To reduce such a mental load, we focused on both intra-relationships among imagewords and inter-relationships between imagewords and color coordinates. In our method, a user firstly gives a 2D-imageword map to teach intra-relationships, i.e. subjective similarity, among them. Secondly, the user gives subjective scores to basic imagewords in relation to typical color coordinates. Applying multivariate analysis, we got the user's taste model on color coordinates. Thus, we can select suitable color coordinates from wide variety of imagewords. In our current experiment, we achieved good precision for the imagewords strongly associated with basic imagewords.