Abstract: In the area at the junction of land and sea in remote sensing images,due to the complex and changeable environment,the spectra of the boundary between land and sea are not clearly distinguishable,which makes it difficult to accurately determine the location of the coastline.To solve this problem,this paper constructs a coastline extraction model (EGOM) that integrates an edge detection neural network and grey theory.The model uses a multi-scale module group SEM,which can effectively capture multi-scale features.With the help of cross-resolution local fusion and secondary fusion mechanisms,the edge prediction map is optimized.Finally,a pseudo-edge culling strategy based on grey theory is introduced.Experimental results show that the model has an average offset index of 18.89 m,a root mean square error of 21.05 m,and can effectively remove pseudo-edges.Its overall performance is better than several other coastline extraction methods.

Key words: remote sensing imagery, shoreline extraction, edge detection neural network, grey theory, pseudo-edges