TWO COASTLINE DETECTION METHODS BASED ON IMPROVED LEVEL SET ALGORITHM IN SYNTHETIC APERTURE RADAR IMAGES

摘要

Level Set Algorithm (LSA) is one of the efficient toolsrnof edge detection and contour extraction, which hasrnbeen applied to radar, medical, and other imagesrnprocessing successfully. However, the procedure usingrnLSA will cost a lot of time when applied tornhigh-resolution images. This paper provides two novelrnapproaches that can enhance the detection speed. Inrnthe first method, the moving step enlarges when therncontour is far from the real boundary of specificrntargets, while reduces when the contour is near the realrnboundary in the images. In the second method,rnreducing resolution make the coarse boundary bernobtained by LSA in low resolution image. Then, therncontour is processed finely by LSA in high resolutionrnimage based of the coarse boundary. Experimentalrnresults using Radarsat ScanSAR images show therneffectiveness of these new methods and verify thernimproved velocity in contour detection compared withrnthe standard LSA.