MESSIE: a multi expert system in computer vision. Application for aerial imagery interpretation

摘要

This work is in the context of the methodological development of audmultispecialist architecture called MESSIE . This paper presents theudsystem specifications for the interpretation of man-made structures in theudfield of aerial imagery . The main difficulty of such a system is theudknowledge expression necessary for the interpretation : among them,udstrategy, scene and objects knowledges. We show how the choosenudmodeling of physical objects described by four points of view allows toudlead to the specifieation of an architecture . These points of view areudgeometry and radiometry modeling (shape-aspect), spatial relation of theudobjet with the others (context) and the functionnality (semantic) . MESSIEudis a blackboard architecture organized with four types of knowledgeudbases that schematically are grouped in two hierarchical levels . The first level corresponds to the Scene and Strategy level and the secondudcorresponds to the specialists (one for each objet) . Each level works onlyudwith certain points of view . We show the modularity of a such approachudand the facility to use différent strategies . We illustrate this paper withudresults obtained with four specialists : Road, Building, Shadow andudVehicule for man-made objects and one specialist for natural objectud(river) .