Automated mapping of coastline from high resolution satellite images using supervised segmentation

Anne Puissant (1) , Sébastien Lefèvre (2) , Simon Rougier , Jean-Philippe Malet (3)

1 LIVE - Laboratoire Image, Ville, Environnement
2 SEASIDE - SEarch, Analyze, Synthesize and Interact with Data Ecosystems
3 IPGS - Institut de physique du globe de Strasbourg

In this article, we are dealing with the problem of coastline extraction in High and Very High Resolution multispectral images. Locating precisely the coastline is a crucial task in the context of coastal resource management and planning. According to the type of coastal units (sandy beach, wetlands, dune, cliff), several definitions for the coastline has to be used. In this paper a new image segmentation method, which is not fully automated but relies on a low intervention of the expert to drive the segmentation process, is proposed. The method combines both a marker-based watershed transform (a standard image segmentation method) and a supervised pixel classification. The user inputs only consist of some spatial and spectral samples which are defined depending on the coastal environment to be monitored. The applicability of the method is tested on various types of coastal environments in France.

Mots clés: Image segmentation; Weak supervision; High resolution optical imagery; Coastline extraction

Domaines: Traitement des images [eess.IV]

Anne Puissant, Sébastien Lefèvre, Simon Rougier, Jean-Philippe Malet. Automated mapping of coastline from high resolution satellite images using supervised segmentation. Fourth international conference on Geographic Object- Based Image Analysis, 2012, Rio de Janeiro, Brazil. pp.515-517. ⟨hal-00763489⟩