Ultimately we want the ontologies/vocabularies
We need to know which information is worth adding and how to represent it.
We don't need to add information that is already available. Much image metadata is already available in the resource itself.
As with Delivery Context, we anticipate a core ontology and an extension mechanism.