An improved road extraction method based on MRFs in rural areas for SAR images

In the proposed general processing framework for road network extraction in synthetic aperture radar (SAR) images, this paper deals with automatic extraction of the road network. The general framework is based on road candidate detection, followed by optimization using a Markov random field description of road networks. The first part introduces the segment extraction technique in the road detection step. After introducing segment junction rules in the latter step, the major modification is the clique potentials of Markov random field that extracts the intersections of road networks, especially for T shape junctions. Advantage over previous extraction is proved by comparison of the experimental results.