Abstract
Bright lines are a characteristic feature of synthetic aperture radar (SAR) amplitude images of urban areas, and are commonly associated with man-made structures. In order to aid in the development of SAR applications using these features, an automated approach to bright line detection is proposed, based on scale-space ridge detection at a single scale, and using a naïve Bayesian classification step to select the ridge points corresponding to bright lines. The effectiveness of the technique is demonstrated by applying it to a COSMO-SkyMed image of L'Aquila, Italy.