River shoreline extraction and bank collapse recognition based on Chinese GF satellite

Extracting river information from high-resolution satellite remote sensing images is of great significance for monitoring and warning of river bank collapses. This article takes the example of the bank collapse with a length and depth of about 100m in Xiaopan of Jiayu County, Hubei Province in December 2021, and establishes a water extraction model based on DeepLabV3+image semantic segmentation technology. The application ability of domestic GF No.1 and No.2 satellites in identifying river bank collapses is discussed. The research results indicate that the DeepLabv3+model can extract water edge information well and the extraction results are reliable. By comparing long-term image images, it can reflect the situation of bank collapse on both sides of the river. The resolution of domestically produced Gaofen-1 and Gaofen-2 satellites is sufficient for accurate identification of larger scale landslides, and further exploration is necessary to enhance their ability to identify smaller scale landslides.