Image interpretability and ship target recognition in airborne SAR images of harbour scenes

This paper discusses some of the challenges with Synthetic Aperture Radar (SAR) imaging for scene analysis and target recognition in a harbour environment, and presents the results from a target recognition experiment. SAR images of Oslo harbour with a number of vessels have been collected with an airborne radar system. The images have been analyzed with the aim of assessing the extent of phenomena that could lead to reduced image interpretability. Shadows, multiple-bounce backscattering, target motion and sea-surface multipath are discussed. A simple target recognition experiment is carried out, in which SAR images of ship targets are correlated with a reference set of previously collected images. One of the purposes with the experiment is to get a better understanding of how the potential for scene analysis and target recognition varies with the imaging geometry, and to find favourable geometries for the analysis of harbour scenes.