Knowledge-integrated stepwise optimization making modal for spatial feature mining and its application in remote sensing image classification

Extending the method of Gaussian mixture modeling and decomposition (GMDD), a new feature mining method named step wise optimization model (SOMM) with genetic algorithms (GA) is proposed in this paper. This method is used in the extraction of tree-like hierarchical structure of unknown feature distributions in feature space. To approximate reality accurately, the integration of SOMM-GA with symbolic geographical knowledge is essential in the feature mining and classification of remote sensing images. A knowledge-integrated SOMM-GA model that combines the power of SOMM-GA and logic reasoning of rule-based inference is proposed. In addition to conceptual and technical discussions of the model in detail, it is tested in a practical application in some districts.