Coastline detection in SAR images using multi-feature and SVM

Coastline detection in synthetic aperture radar (SAR) images is difficult and important. As we all know, water areas in SAR images are much more homogeneous in grey levels than land regions. Therefore, features reflecting the texture of an image can be very useful for water-land separation. Traditional methods mainly view gray-scale as a critical rule to detect coastline from background. In this paper, we pay more attention to the characteristics of the different texture between water and non-water objects. A novel method based on circular-window gray feature and gray level co-occurrence matrix (GLCM) is proposed. Eighteen features of water and non-water areas depicted by circular-window gray feature and GLCM are fed into a support vector machine (SVM) classifier to extract coastline. The experimental results demonstrate that the proposed approach has better performance compared with other ones.