SAR image segmentation by mathematical morphology and texture analysis

Presents a promising segmentation method for synthetic aperture radar (SAR) imagery designed in the context of agricultural remote sensing using a morphological region based image analysis. The region based approach is really appropriate for land use applications given that land cover is naturally built-up from regions. The method is divided into two principal steps: a morphological image partitioning resulting in a severe oversegmentation and a further region growing process exploiting information from a textural edge detector.