The present invention relates to a
satellite image based coastline automatic extraction and classification method. The method is characterized technically by comprising the following steps: step 1, using a multi-
spectral image of a
coastal zone after precise registration, and by combining a normalized difference
water index (NDWI) and a
level set model, obtaining a precise contour of a water
boundary line; step 2, performing object-oriented segmentation on a land portion region, and extracting coastline coastal feature basic units by means of a spatial position topological relationship between the basic units; step 3, based on a
decision tree, performing determination and classification on each coastline coastal feature basic unit; step 4, determining a coastline position according to a coastline interpretation rule; and step 5, on the basis of a result of the coastline position based on the multi-
spectral image in the step 4 and property information, outputting a coastline vector based on a high-resolution panchromatic image. According to the method provided by the present invention, the coastline position can be rapidly, accurately and comprehensively surveyed and the change of the coastline position can be dynamically monitored.