Segmentation Techniques for Land Mask Estimation in SAR Imagery

Synthetic Aperture Radar systems are powerful observation tools for maritime surveillance applications such as fisheries monitoring or pirates detection and oil slick detection. It is an important problem that has not been completely solved. In order to perform those detections, a search area is needed. In this paper, several segmentation techniques are proposed and compared to give rise to the search area. These techniques are applied to different SAR images acquired by TerraSAR-X. The best results in terms of land and sea classification are obtained when the superresolution algorithm combined with the wavelet transform based edge detector is applied.