Databases Reference
In-Depth Information
Figure 1: Conceptual architecture of RWDB
•
Multi-Media DataBase Engine (MMDBE)
•
Cyber World Reconstructor for retrieval (CWR)
Among them, the RWM has an essential role to recognize the object and the motion
from captured data and to extract “abstract” (or index) data for database retrieval.
For this purpose, we introduce “Mediator” to describe object's and motion's
characteristics with small amount of data. The mediator is an approximated model
based on the meaning of the objects, with semi-automatic fitting on the sampled
data from real world. For example, as to figure of human head, we use a cylinder
with 10 layer, polygonal column
1)
. Model of a human head has the additional
information that the third layer from top is the height of eyes. Also, it is also
possible to supply position information that in the parameter of these sections
which vertex represents the left and right eye or the mouth. This way it becomes
possible to supply general information about a certain type of parts, and integrate
the individual parts data into a meaningful model representing each instance.
Additionally, this paper describes the RWDB concept, and is focusing on
Modeling aspect of real objects. The (shape) Mediator is proposed as an
optimum solution. Also, “event” concept is discussed from the viewpoint of
“location analysis” of objects. In this paper, two more mediators are discussed:
Structure Mediator and Motion Mediator. These Mediator data can be also
detected semi-automatically from real world. Finally, we will show the concrete
database schema to represent relationships between three types of Mediators
and real world data.
