2D 이미지 기반 형상 변형을 이용한 개인 맞춤형 인체 하반신 모델링

Abstract

A realistic and accurate human body model is being used as a powerful tool for designing for a human-centered product as an international trend in ergonomics, game industry, and automobile industry. In this respect, there have been many efforts and studies to obtain the realistic and accurate human body model. But there are limitations on each type of data required to make this model. Therefore, this study proposes a novel method for creating a personalized human body models using two 2D images and a human template model. When the template model is determined through simple individual information such as the user’s height and weight, the personalized human body model is created by the following four steps; i) Image processing and silhouette extraction, ii) Defining characteristic points and creating key-sections, iii) Width deformation through matching with body silhouette, and iv) Creating a sweep-based human body model. Unlike existing methods of generating the human body models through 3D scan data, the proposed method can provide convenience to the user by requesting simple information. In addition, the method can improve accuracy of the model through deforming the width by matching with the body silhouette. Finally, the study demonstrated the validity of the method by comparing the result model generated by the purposed method with the model created through the 3D scan data.