A Synergic Expert-Neural Network System to Identify Relocatable Targets using Multi-Sensor Fusion
Small Business Information
Lnk Corp., Inc.
6811 Kenilworth Avenue, Riverdale, MD, 20737
Dr. Srinivasan Raghavan
AbstractThe central focus of this proposal is on building an automatic target recognition system for identifyi-ng relocatable targets by integrating information from multiple sensors using a synergistic framework of neural networks and exprt systems. This frairework is ideally suited for 3..ricorporating human expertise of both perceptual and qualitative reasoning processes . From our prior experience in this area (NADC N62269-90-C-0567), we find the key to an efficient solutLon lies in using a composite architecture conbining an e)q.rt system with a hybrid neural subsystem. We propose a th ^ elevel processing system for identi- fying the relocatable targets. Th- first level of %is system addresses issues related to data representa^Lon and registration ljetween sources . The second level achieves feature extraction and parbLal recognition results using neural networks. Die third level consists of a decision maki-ng expert system wLth fuzzy logic reasonlng to achieve a collectLve decision from the multi.ple sources using an object-oriented representaLLon of the target.
* information listed above is at the time of submission.