Task-specific segmentation of remote sensing images

Presents a task-specific segmentation method that incorporates semantic knowledge into data-driven segmentation process through different region merge scores. Starting from a simple region growing algorithm which results in over-segmented regions, the authors apply a region merging method designed specifically for each task such as road extraction or vegetation area identification. Further, edge information is integrated to verify and correct region boundaries. The experimental results show that this method can reliably extract areas of interest such as roads and vegetation areas in Landsat images.