Contents of this page are obsolete. This page is preserved and stored at this URL just from historical viewpoint. Original URL was http://www.mm.media.kyoto-u.ac.jp/members/kameda/...
Please visit www.kameda-lab.org for recent information. (2002/12/06, kameda@ieee.org)

Publication, Research, Docs and Info, KAMEDA


next up previous
Next: Difference Image Up: Node Mode and Previous: Node Mode and

Node Mode

Suppose you observe a motion of a human body from the camera viewpoint. Human body parts can be classified into 3 modes: moving mode, stationary mode, and occlusion mode. To correspond these modes with the model, we classify the nodes in the model in the same way. The characteristics of these node modes are described below.

  1. Moving Mode
    A part corresponding to the node labeled the motion mode can be seen from the camera. A difference image is available to estimate the joint angles of the node.

  2. Stationary Mode
    A part corresponding to the node labeled the stationary mode can be seen from the camera. As the node is not moving, there is no area of it in a difference image.

  3. Occlusion Mode
    A node in this mode can not be seen from the camera. In this case, we cannot get any information about the joint rotational angles from the images. So, we apply the inertia assumption to predict the joint angles.

Let be a node set at time t where the nodes are in the motion mode, and be a set of the stationary mode nodes and be a set of the occlusion mode nodes.



Yoshinari Kameda
Mon Apr 7 14:56:47 JST 1997