Separation of sea and land in SAR images using texture classification

The coastline mapping is a crucial point for coastal studies and cartography. In the most general case, coastline evolution is slow enough to allow regular upgrading of the maps by the means of airborne or in situ data. But in some regions the annual coastline movement can reach 1 km. In such cases, there is a need for a cheaper and accurate data, that can ensure a good spatial covering at short intervals. Spaceborne remote sensing data partly fulfills these criteria. The authors present in this work a new method using SAR images for coastline detection. Discrimination between sea and land is achieved by the use of the speckle textural information. The results are compared with other ones found with more classical methods. The coastline retrieved with the authors´ method is shown to contain more detail and to be more reliable