Human Action Tracking Guided by Key-Frames

The model-based approaches for tracking of human bodies in image sequences can be categorized into two types; tting model to body frame by frame, and accumulating estimated pose displacements in successive frames after model tting at the initial frame. The latter has an inherent drawback as accumulation of tracking errors while the one has a great advantage as small computational e orts compared with the former. This paper proposes a new method which can correct the tracking errors by propagation from tting model to body at a few key-frames. The propagation makes it possible to establish tracking of bodies under occlusion. Capturing the actor's motions in real old movies is presented.