In order to apply Satellite Remote Sensing (RS) to mining areas, some key issues should be solved. Based on an introduction to relative studying background, related key issues are proposed and analyzed oriented to the...In order to apply Satellite Remote Sensing (RS) to mining areas, some key issues should be solved. Based on an introduction to relative studying background, related key issues are proposed and analyzed oriented to the development of RS information science and demands of mining areas. Band selection and combination optimization of Landsat TM is discussed firstly, and it proved that the combination of Band 3, Band 4 and Band 5 has the largest information amount in all three-band combination schemes by both N-dimensional entropy method and Genetic Algorithm (GA). After that the filtering of Radarsat image is discussed. Different filtering methods are experimented and compared, and adaptive methods are more efficient than others. Finally the classification of satellite RS image is studied, and some new methods including classification by improved BPNN(Back Propagation Neural Network) and classification based on GIS and knowledge are proposed.展开更多
基金Under the auspices of the Research Foundation of Doctoral Point of China(No.RFDP20010290006).
文摘In order to apply Satellite Remote Sensing (RS) to mining areas, some key issues should be solved. Based on an introduction to relative studying background, related key issues are proposed and analyzed oriented to the development of RS information science and demands of mining areas. Band selection and combination optimization of Landsat TM is discussed firstly, and it proved that the combination of Band 3, Band 4 and Band 5 has the largest information amount in all three-band combination schemes by both N-dimensional entropy method and Genetic Algorithm (GA). After that the filtering of Radarsat image is discussed. Different filtering methods are experimented and compared, and adaptive methods are more efficient than others. Finally the classification of satellite RS image is studied, and some new methods including classification by improved BPNN(Back Propagation Neural Network) and classification based on GIS and knowledge are proposed.
