A MULTI-AGENT APPROACH TO SELF-ORGANIZING VISION SYSTEMS
Abstract:
We present a new multi-agent system architecture for modelling self-organizing computer vision systems, that provide a great degree of exibility: Firstly, the vision systems can be easily incorporated into complex applications; secondly, the vision systems are able to adapt dynamically to dierent (possibly changing) tasks and environmental conditions; thirdly, the architecture provides the ability to handle dierent competitive information; and lastly, the addition of new processing modules can be done without rebuilding the complete system. We describe in detail the basic concepts of the proposed multi-agent system including the agent architectures, the control strategies as well as the communication language. The multi-agent system architecture is demonstrated on the basis of a distributed object recognition system modelled as a society of autonomous agents.
Citations
| 8 | ASIC: An architecture for social and individual control and its application to computer vision – Boissier, Demazeau - 1994 |
| 4 | A purposive computer vision system: a multiagent approach – Bianchi, Rillo - 1997 |
| 4 | An architecture of object recognition systems for various images based on multi-agent – Yanai, Deguchi - 1998 |
| 4 | Recognition of partially occluded objects through fuzzy invariant indexing – Graf, Knoll, et al. - 1998 |
| 3 | Mavi: a multi-agent system for visual integration – Boissier, Demazeau - 1994 |
| 2 | Framework for implementing self-organized task-oriented multisensor guidance – Scheering, Knoll - 1998 |
