Publication,
Research,
Docs and Info,
KAMEDA
When a person participate in the activities in the real world, he/she can watch the activities in his/her own way. In the case a user is in a remote place, the user would do the same as if he/she were in the place. This means that each user in the remote place is assigned to an active camera in the target space. If we can reconstruct the target space in 3D space, this could be possible with virtual cameras[5]. However, considering image quality, this is impractical. Therefore the problem here is how to get the users' favorite video images under the constraint that the number of the active cameras is limited.
Since it is difficult to see multiple video images simultaneously and it needs broad band width of the network in transmitting multiple video images for one user, the user is assumed to watch only one video image. A video image selection process for each user should be prepared which satisfies the request of the user under the physical constraint of the active cameras such as locations, pan/tilt angles, and camera ranges.
Each user proclaims his/her favorite imaging rule in advance by specifying several camera-works at each dynamic situation. Since the number and position of the cameras are limited, the mediation is necessary to satisfy as many users as possible.