Multi-Agent Methods: An Example of an Architecture and its Application for the Detection, Recognition and Identification of Targets - Information Fusion in Signal and Image Processing - page 215

The usage of a sequence of images is carried out by a loop that operates in a set
time interval. For each time increment, the intra-image process is started over again
by way of the Concept agent.
The objective is then to use the past to our benefit, and by past we mean the infor-
mation gathered from each image, in order to avoid going through all of the processes
for each new image. Thus, under the fixed camera hypothesis, we have to consider the
following two situations depending on the kinematics associated with the concept:
- for a fixed concept, it seems wise to keep the satisfactory samples obtained for a
given image. These samples will be copied at the time t +1in the following image.
This information copying phase is controlled by a validation threshold that specifies
the minimum authorized confidence. Based on these copied samples, a new statistical
analysis is conducted by the Detection agents in order to locally adjust the segmenta-
tion parameters;
- for a concept in movement, it seems wise to start over a complete analysis. Thus,
nothing is copied.
After describing the intra- and inter-image control cycles, which made it possible
to present the various agents of the systems and their roles, we are now going to
describe in greater detail the information handled by the agents. We will focus on
their features, on how they are organized, as well as on possible causality relations.
10.5. The information handled by the agents
Each agent uses and provides information. All of these elements of information
have to be organized so as to allow other agents to have access to them. We will
distinguish two types of elements of information in our system:
- descriptive and operational knowledge that provides a prior representation of the
different concepts present in the application;
- the information gathered by agents during the analysis of the image sequence.
This distinction is important since the representation modes associated with these
elements of information are distinct and it is precisely the composition of these infor-
mation structures that we are going to describe with the knowledge base and the world
model.
10.5.1. The knowledge base
The knowledge base is specified in order to provide agents with the information
required for their execution. It defines, among other things, the field of the process,
the initial objective and provides an operational description of the concepts.