SAR Image Stitching And Rapid Extraction Of Large-scale Coastline Using Wide-swath Data

Using SAR image data for coastal monitoring and management is a convenient way.Image stitching is needed when the coverage of the study area is larger than the shooting range of a single SAR image.In addition,coastline is an important feature and an indispensable part while focusing on the coastal area.In order to monitor and manage large-scale coastal zone,this paper presents a new approach for the SAR image mosaic and rapid coastline extraction using wide-swath SAR data.Firstly,this approach extracts geographic information from the input SAR image sequence.Then,based on the gray statistics of the overlapping regions of the image sequence,the color differences between adjacent images were eliminated by Wallis transform,and the mosaic image of the studied region is generated by geographic information(longitude and latitude).The backscattering coefficients of the land and seawater in the coastal area are very nearly the same,which can easily lead to misclassification of oceans and land.So it is needed for the mosaic image to enhance the difference between land and seawater using morphological image processing method.Next,the mosaic image is used as the input image for rapid extraction of the coastline,and the Gauss image pyramid sequence of the mosaic image is established.The boundary tracing algorithm is used to extract the coastline from the low-resolution image of the Gaussian image pyramid sequence,and the resulting coastline results are resampled to the same resolution as the input image to obtain the final large-scale coastline.The experimental results show that the gray value of the joints of different input images is close,the gray variance is low,and the transition naturally does not present obvious inlaid lines in the SAR image stitching resulted form wide-swath data captured over the Yellow Sea.Moreover,under the premise of ensuring accuracy,the extracted large-scale coastline has significantly lower processing time than that of manual marking and other methods.