Feature-mapping data fusion
Fuxi Wang, John Litva, Titus K. Y. Lo, Éloi Bossé, Howard W. H. Leung · IEE Proceedings - Radar Sonar and Navigation · 1996
A centralised plot-level data fusion technique, which is based on a neural network, is presented. A self-organised feature-mapping technique, which learns from sensor observations, is used to integrate the data from several sensors with various unknown measurement accuracies. Topological neighbourhood formulation among networks for integrating data is described. Computer simulations show that neural data fusion enjoys many advantages over maximum likelihood fusion and other ad hoc fusion methods.