In the process of intelligent inspection of ultra-high voltage transmission lines,it is crucial to extract buildings from satellite remote sensing images.In this paper,we propose a fine-grained feature fusion building extraction method,U-DeepNet,which integrates the advantages of Deeplab V3+and U-Net model to fully extract fine-grained features and ensure that densely arranged buildings can be accurately extracted and distinguished.By adding two different residual U-block (RSU),it increases the network's sensitivity to small targets and uses an attention mechanism with different di-mensions to pay reasonable attention to effective information.The experimental validation on the Massachusetts building dataset indicates the superiority of the proposed method.
维普
万方数据