FROM BODY MEASUREMENTS TO HUMAN PERCEPTION OF BODY SHAPES: MODELING USING INTELLIGENT TECHNIQUES

摘要

This paper presents a method for modeling the relationship between abstract human perception describing body shapes and concrete body measurements. Two modeling procedures are proposed. In the first procedure, the experts generate a list of sensory descriptors describing human body shapes and then evaluate the virtual human bodies using these descriptors. For each evaluator, a decision tree is then created to model the relationship between each sensory descriptor and the body measurements. In the second procedure, the experts evaluate the relationship between these sensory descriptors and a number of abstract fashion themes such as "sporty" and "attractive" without taking into account specific human bodies. This conceptual relationship given by different experts is modeled using a fuzzy cognitive map. Finally, the relationship between the abstract fashion themes and the body measurements for all the evaluators is obtained using the composition operation of fuzzy relations. Using the proposed method, we can effectively estimate the human perception of body shapes, expressed by a set of keywords, using the corresponding body measurements. These keywords can be further integrated into a knowledge base of garment design for developing personalized new garment products.