Satellite image interpretation requires the assignment of sets of objects (or pixels) that share certain attribute values to object categories. This procedure requires expert intervention and knowledge. An approach has been developed that formalizes expert knowledge in the image interpretation procedure with ontologies. Ontologies provide a definition of object categories and associated attribute values that are known to represent these object categories. A classic ontology has the limitation that the definitions of object categories and their properties need to be crisp, i.e. not overlapping. Practical tests showed that less rigid definitions of class properties make the ontology-based approach more flexible and adaptable to different study areas and satellite images. This paper presents the extension of the ontology-based approach with fuzzy rules and discusses the advantages of this extension.