A Non-Supervised Method For Shoreline Extraction Using High Resolution SAR Image

       2012年遥感计算器视觉国际会议——This paper presents a non-supervised method for shoreline extraction from high resolution synthetic aperture radar (SAR) image. The proposed technique is based on level set with all the parameters optimized in order to be used with different kinds of SAR data and to find the desired boundary with a minimize number of iterations so as to be fast enough. The preprocessing including Roberts Operator and Histogram adjusting is used to enhance the contrast of the boundary. Then we initial the curve to cover the whole image, and an improved level set method is used to do the segmentation. After that, several post-processing steps are utilized to remove any remaining spurious segments. It’s completely non-supervised and can be applied to different kinds of SAR data. The results confirm that the proposed method can provide a stable and fast solution to the shoreline extraction using SAR data.