This paper presents a system for animating customized virtual humans using motion parameters estimated from multi-view image sequences. The advantage of our method is that the subject does not have to wear markers or other devices. We offer a detailed analysis of the singularities that might arise during tracking humans using multiple cameras and we provide an analysis of the criteria that allow the ordering of the cameras that provide the most information for tracking. We demonstrate our system using image sequences requiring the recovery of complex threedimensional motionsinthe presence of significant occlusion, and we present a quantitativeanalysis of the accuracy of the estimated motion parameters.
Ioannis A. Kakadiaris, Dimitris N. Metaxas