Knowledge-based 3-D object recognition

The organisation of a system for 3-D object recognition by using a knowledge base, is presented. Globally, such a system consists of three major modules, a supervisor which coordinates the other components in order to derive reconstruction of an entire scene, a sensory module and a matchings module. The supervisor communicates with the user of the system through a user interface unit for interactive exchange of information purposes. The second module denoted as sensory provides this information. The third module, indicated as the matching module consists of two components namely the structural and the geometrical matching processes. In our design of the system the structural matching module forms the main idea of the 3D-reconstruction therefore it has the major effort of consideration. The implementation of the supervisor and the structural matching module adopt a knowledge base approach which is explained in detail