Advances in computer assisted image interpretation

In this paper we present a semi-automatic aerial image interpretation system. The knowledge, to be incorporated into the system in order to identify the best solution for the otherwise under-constrained problem of image interpretation, is structured in a logical way and distributed over a set of knowledge sources in a blackboard system. Each knowledge source applies the knowledge representation method which is best suited for its type of knowledge. In the instances where the knowledge represented in the system is insufficient, the system will suffer from an optical illusion and the human supervisor will have to correct the solution stored on the blackboard. The main goal of the system is to free the human image interpreter from the routine object vectorization part of the work and allow him/her to focus on the interpretation of the presence of specific objects at certain positions.