In a vision system, every task needs that the operators to apply should be « well chosen » and their parameters should be also « well adjusted ». The diversity of operators and the multitude of their parameters constitute a big challenge for users. As it is very difficult to make the « right » choice, lack of a specific rule, many disadvantages appear and affect the computation time and especially the quality of results. In this paper we present a multi-agent architecture to learn the best operators to apply and their best parameters for a class of images. Our architecture consists of three types of agents: User Agent, Operator Agent and Parameter Agent. The User Agent determines the phases of treatment, a library of operators and the possible values of their parameters. The Operator Agent constructs all possible combinations of operators and the Parameter Agent, the core of the architecture, adjusts the parameters of each combination by treating a large number of images. Through the reinforcement learning mechanism, our architecture does not consider only the system opportunities but also the user preferences.

1. R. Clouard, A. Elmoataz & F. Angot, "PANDORE : une bibliothèque et un environnement de programmation d'opérateurs de traitement d'images", Rapport interne du GREYC, Caen, France, Mars 1997.
2. B. A. Draper, J. Bins, and K. Baek, "ADORE: Adaptive Object Recognition". Videre, 2000. 1(4): p. 86-99.
3. B. Nickolay, B. Schneider, S. Jacob, "Parameter Optimization of an Image Processing System using Evolutionary Algorithms" 637-644. CAIP 1997.
4. G. W. Taylor, "A Reinforcement Learning Framework for Parameter Control in Computer Vision Applications" Proceedings of the First Canadian Conference on Computer and Robot Vision (CRV'04), IEEE 2004.
5. F. Sahba, H. R. Tizhoosh, M. Salama. "Application of reinforcement learning for segmentation of transrectal ultrasound images" BMC Medical Imaging 2008, 8:8.
6. I. Qaffou, M. Sadgal, A. Elfazziki:"A Reinforcement Learning Method to adjust Parameters of Vision Operators", Sixth International Conference on Intelligent Systems: Theory and Application. 23-29, Rabat 2010.
7. I. Qaffou, M. Sadgal, A. Elfazziki:"A Reinforcement Learning approach to adjust parameters of texture segmentation", Wotic'09, Abstract p 53. Agadir 2009.
8. RS. Sutton, AG. Barto: "Reinforcement Learning" Cambridge, MA: MIT Press; 1998.
9. Watkins CJCH, Dayan P: "Q-Learning". Machine Learning 1992, 8:279-292.
10. R Haroun. "Segmentation des tissus cérébraux sur des images par résonance magnétique". Master's thesis, Université des sciences et de la technologie Houari Boumediène, 2005.
11. S. Chabrier, H. Laurent, C. Rosenberger, Y. J. Zhang, "Supervised evaluation of synthetic and real contour segmentation results", European Signal Processing Conference (EUSIPCO) 2006.
12. DO Minh Chau, "Évaluation de la segmentation d'images". Rapport final TIPE. Institut de la francophonie pour l'informatique. Nanoï 2007.

Computer Vision Reinforcement Learning Multi-agent System Parameter Adjustment Operator Selection Q-learning Segmentation