Ship Detection in SAR Images of Large-Scale Complex Scenes Based on Coastline Segmentation
Most of the traditional coastline segmentation methods on airborne and spaceborne platforms rely on artificial construction of features or morphological methods,and it is difficult to balance the relationship between model generalization ability and segmentation accuracy.In this paper,we study the ultra-high resolution Synthetic Aperture Radar(SAR)imaging and its application in ship detection on both airborne and spaceborne platforms,and propose a sea-land segmentation method based on super pixel segmentation and deep convolutional neural network.Using depth features instead of artificial features and making full use of the texture,gray and brightness features of SAR images,it has a high segmentation accuracy on the basis of ensuring the generalization ability of the model,which lays a good foundation for the ship detection task on the airborne and satellite-borne platforms.
coastal line segmentationdeep learningneural networkSynthetic Aperture Radar(SAR)
NETL
NSTL
万方数据
