Generic blackboard based architecture for data fusion

Information gathered by different knowledge sources from same data sets or scenes are often uncertain, inconsistent, incomplete or imprecise. Merging of redundant data can reduce imprecision and using multiple time segments can provide more complementary information. Fusion of such information can create a more consistent target recognition. The Generic BlackBoard (GBB) is the extension of Common LISP and CLOS, and provides an environment to user for developing the expert system with blackboard structure. Using the blackboard concept, we developed a data fusion system based on the Generic BlackBoard. The system was constructed by using the Allegro Common LISP under the Generic BlackBoard environment. The average approach was used to integrate extracted features on the feature fusion level. The majority vote approach was employed for implementing the decision fusion. These approaches were implemented under the presented data fusion system and tested with the real experimental data radiated by the underwater target. The goal is to obtain better descriptions of the radiation source and get more accurate classification of the target