RBF neural tree networks for multi-class classification in remote sensing

In this paper, the algorithm and realizing procedures of the RBFNN used in classification of remote sensing image were discussed, and a training algorithm based on Adaptive Global Distance Fast Cluster (AGDFC) and a tree-like hierarchical RBFNN constructing algorithm were. Then, the case of practical application of remote sensing land cover classification in Tai'an region was presented. Through comparing with MLC, classification process and results were synthetically analyzed. Experimental results show that RBF neural tree networks approach has more advantages in training time, network structure, classification precision, etc.