Classification Method Based on Multilayer Tree-Structured Neural Network for Remote Sensing Images of Tianjin Development Area

Classification has been widely applied in images information mining. In this paper, the classification methods based on multilayer tree-structured neural network for remote sensing images of Tianjin development area is presented as an extended version of the classification algorithms in neural network, which has a dynamic tree-structure generated during the training process and its´ growth speed can be controlled through the weight of network, and the user can get the hierarchical results through changing the weight of network in different steps.