Coastline extraction from remote sensing image based on improved minimum filter

In this paper, coastline has been extracted from remote sensing image using the supervised classification method through the research of remote sensing image features. The coastline extraction can be very accurate when the seawater features are obviously consistent in image. For the calm sea image, the coastline can be extracted accurately based on supervised classification. But for the images that have large variation and obvious reflective waves, it will be under the influence of sea surface texture obviously. In order to solve this problem, it put forward an improved minimum filter method to avoid the impact of sea surface texture. The method added two thresholds, which were the PSNR (peak signal-to-noise ratio) and the correlation coefficient. It limited the filter range in sea area. At last, semiautomatic extracted the accurate coastline in ArcGIS. This method can extract most of the coastline in the high-resolution image.