Unsupervised Coastline Detection And Tracing In Sar Images

Images of coastlines generated by Synthetic Aperture Radars (SAR) suffer from a number of deficiencies which arise from the presence of the speckle ' effect and the strong signal return from a wind roughened, wave modulated sea. The frequent lack of contrast caused by these effects makes coastline detection difficult by most conventional procedures such as, gray level thresholding. or segmentation by edge detection. This paper describes an algorithm for the global detection of coastlines based on a sequence of basic image processing procedures and a new edge tracing algorithm. The application of the proposed procedure to SEASAT SAR and SIR-B (Shuttle Imaging Radar B) images demonstrates that it achieves good performance with only a modest computational burden. The computational burden can potentially be reduced to a level permitting a near real time performance in the sense of completing computations within a few frame times of typical image displays.