Automatic monitoring of land cover in sub-urban areas by X-Band VHR cosmo-skymed imagery

With the launch of the latest generation satellites, a huge amount of data acquired with high spatial and temporal resolution is allowing scientists to constantly monitor the environment, allowing timely intervention when necessary. In particular, the use of VHR SAR images, such as those taken by the COSMO-SkyMed constellation, can be extremely helpful to assess the damage caused by natural disasters, as well as for monitoring the urbanization process. However, managing such a large archive of data implies the development of automatic techniques for data processing and information extraction. This study focuses on the implementation of an automatic processing chain based on neural networks for mapping and monitoring the changes of land use in sub-urban areas, from X-band Spotlight and Stripmap COSMO-SkyMed imagery. Land cover maps provided by Multi Layer Perceptron Neural Networks, and preliminary change detection results achieved by means of a Neural Architecture for HIgh-Resolution Imagery (NAHIRI), are shown and discussed