An Anatomic Human Body For Motion Capture

Motion capture is the process of capturing the live motion of a person in order to animate a virtual character. Until now the approach providing the best results was marker recognition from multiple camera views (in the infrared range). However, it still requires significant human intervention in the post-processing phase due to weaknesses in the marker tracking. In the framework of the MOCA project we propose a motion capture methodology based on a 3D anatomic human model. This model encompasses a precise description of the skeleton mobility associated with an approximated envelope. Its first objective is to ensure the best matching between the real performer and the articulated structure supporting the animation description. The second objective is to allow the prediction of the 3D location and the visibility of markers.