Identification of airfield runways in synthetic aperture radar images

Synthetic Aperture Radar (SAR) is a microwave-based remote sensing technique whereby images can be captured when optical images cannot, at night or when there is cloud cover. However, its very low signal-to-noise ratio (1:1) means that conventional image analysis techniques are unsuitable for SAR imagery. This paper presents a novel approach to the detection of very small objects in SAR imagery, which combines a pre-processing stage with a Hough transform analysis stage using contextual information to identify suitable 'signatures'. The result of this procedure is the fast and accurate identification of airfield runways. The method is based on the fact that the only reliable characteristic of airfield runways visible in SAR images is the location of lights along their sides.