This paper presents an estimation method of subjective interpretation and attentive graphical descriptions by constructing a user model. This model is based on the relationship between graphical descriptions such as Lab color system, direction, position, and repeated pattern and adjective words for office space images. A statistical method of principal component analysis and multiple-regression analysis is applied to analyze and construct the relationship. We introduce an auto-correlation model for repeated pattern like texture. Our psychological experiments indicate that the four graphical descriptions are significant for the relationship and our estimation method is applied for individual users and gets about 80% correct answer.
抄録全体を表示