Using neural networks for urban area classification in satellite images

摘要

This work presents a new approach to the satellite image classification process. An artificial neural network has been applied for the identification of urban areas mapped in satellite images. The results presented here are based on images obtained from the Landsat satellite using the Thematic Mapper scanner and a land cover map produced under the CORINE Program. The evaluation of the results has been conducted using a point-to-point comparison with a topographic map. The test areas include the metropolitan areas of Warsaw and Cracow.