A dense surface motion capture system for accurate acquisition of cloth deformation

This paper presents a system for accurate dense cloth motion capture. In the proposed pipeline cloth motion and dynamics are captured and reconstructed from a multi-view camera set-up. To allow precise tracking of a high number of surface points a tailored pattern is printed on the cloth. Several existing approaches make use of a printed pattern where point correspondence is determined using color-coded vertex neighbourhoods [3, 1]. In this paper we show that point correspondence can be improved and refined using a Laplacian mesh fitting process in the image domain.