Coastline Detection By Using Sar Data

The coastal zone has close relationship with humankind.The coastline divides the regions of sea and land which is key area of coastal zone research.Remote sensing has the advantage of frequet observation view of the large geographics area,so it can be an important tool to help us on research of the coastline. SAR(Synthetic Aperture Radar) has special advantage in satellite remote sensing with capacity of full-time and all-weather observations. The SAR images are used to detect the coastline in this thesis which includes the work as following:1) The SAR data are pre-processed, including general processing of remote sensing imagery (mosaic,georeference, registration). Different speckle noise reduction methods (including the filters such as Lee, Enhanced Lee, Frost, Enhanced Frost and Kuan, Gamma) and filter sizes are tested and analysied. The Gamma filter with a 5×5 box size is used in this paper.2) Basing on previous studies, a number of methods on edge detections are applied to detect the coastline on SAR images. First, traditional operators like Roberts, Sobel, Laplace, Prewitt, Canny, SUSAN are used and analysied, then two sophisticated methods like level set and Chan-Vese model as well as their improved model are introduced and applied. The Level-Set method requires an initial contour which is relative close to the coastline, and can't get good result when the initial contour isbad. The Chan-Vese model can detect the edge in the image with any given initial contour in theory, but in experiment we can't get satisfactory result with a bad initial contour though the result is better than that obtained from level-set method with the same initial contour. The initial contour by using threshold segmentation and morphology method is used in the method based on Chan-Vese model. These experiments show that Chan-Vese model is better than Level-Set method in the accuracy of result and requirement of initial contour.3) The coastlines of some local areas of Jiao Zhou Bay are extracted by using Chan-Vese model on two SAR image, which were acquired in different time. The results are discussed by considering different definations of coastline, the characters of the coasts of different types and the effects of tide.