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 212

In particular, it receives the requests sent by the agents for the creation of new agents.
Its role is to:
- take an inventory of the agent population;
- assign to the agents the resources necessary for them to function (allocation of
memory space);
- create a thread for each agent;
- provide a common time frame of reference.

The administrator has an information file for each agent that specifies its technical
features.

10.4. The control scheme

The objective is to distribute the vision processes among a population of agents.
The agents in the system have different roles and different behaviors with respect to
the information that is handled and produced. Because we are dealing with the analysis
of a sequence of images, we will distinguish two levels of control: intra-image control
and inter-image control.

10.4.1. The intra-image control cycle

The diagram of the principle of intra-image control is shown in Figure 10.3. It is
comprised of two exploration loops: spatial exploration and temporal exploration.

initialization

anchoring

spatial
exploration

temporal
exploration

movement

detection

focalization

decision

Figure 10.3. The intra-image control diagram is organized according to a spatial
exploration strategy and completed by a temporal analysis of the image

Spatial exploration of the image takes place in three phases:
- the first phase consists of searching areas in the image for samples, i.e. possible
attachment points for a given concept (road, field, vehicle, etc.). In concrete terms,
the goal is to search for the areas that comply with a set of clauses specified in the
knowledge base (a detailed description of this structure is given in section 10.5.1).
This is the localization phase;