키넥트와 통계 모델에 기반한 특정 개인의 인체 외형 모델링

Abstract

In this paper, we propose a novel modeling framework to generate a subject-specific body shape model using kinects and a statistical model. The modeling framework consists of the following three phases: statistical model construction, rough deformation and fine deformation. In first phase, we construct a statistical body shape model based on 3D whole-body scans. The statistical body shape model is roughly deformed by processing rough deformation with specific subject"s body shape captured by kinects in second phase. Finally, exact body shape of the subject is created in fine deformation phase. In this phase, we propose a new method to guarantee the matching process, between scanned 3D point cloud and the statistical model, to be accurate. Our approach have produced a realistic and detailed body shape model having a high modeling accuracy satisfying specified subject"s body geometry.