Publication,
Research,
Docs and Info,
KAMEDA
People usually gather together and do various activities in a certain fixed space. For instance, there are conferences, amusement events such as sports and concerts, lectures in schools, and business activities.
It technically becomes possible to watch such an on-going activity from a remote place thanks to power-up of computers and speed-up of transmission lines.
Video image is considered to be a typical presentation media and is suitable to convey visual information of what happens in the real space.
Video cameras are set to surround the real space to sense the activities there and can change their direction and zooming parameter dynamically. Therefore, the dynamic camera work is a key issue in our research.
Formulated camera works in movies[2] and computer supported movie makers[3][4] have been proposed, but they assume that screenplays or shooting scripts are given in advance and they do not support live camera control. As our case does not allow us to have such scripts, we use concept of dynamic situation instead.
We propose a method of live video imaging for multiple users, which is in a field of multimedia content generation from the real space. This research is located as a visualization part of the CDV framework[1].
In our method, each user is asked to write his/her imaging rule which represents his/her favorite way of imaging an activity. To describe an imaging rule, we define the dynamic situation of the real space and camera-works. The representation of them is the point of this paper.
A dynamic situation is a description of what happens in the real space. Therefore, an activity is described by a sequence of the dynamic situations.
The dynamic situation is linked to the camera-works in the imaging rules of the users and is described by the situation features derived from sensor data, i.e. video images.
The camera-work is concerned how to image an object. It consists of three elements; object, imaging direction, and zooming degree of the camera. With this mechanism, the live video imaging is realized.
In addition, for multi-user environment, every user has his/her own favorite video imaging, which conflicts with a constraint of the location and the number of the cameras. Therefore, mediation process is necessary.
We set a lecture room in Kyoto University as the target real space . This has two reasons. One is that the activities in the real space are relatively simple. The other is that lecture is one of the most important activities in the university, and there is a strong request for students to participate in a lecture from remote places.
We explain the imaging rule and the representation of dynamic situation in Section 2. In Section 3, our proposing multi-user live video imaging method is presented. Section 4 describes experiments and shows the result.