This paper presents an automatic methodology for road network extraction from medium- and high-resolution images. It is based on two steps. In the first step road seeds are extracted using a set of four road objects and connection rules among them. Each road object is a local representation of an approximately straight road segment. In the second step, two strategies for road network completion are applied in order to generate the complete road network. The first strategy is based on two basic perceptual grouping rules, i.e., proximity and collinearity rules, which allow the sequential reconstruction of gaps between every pair of disconnected road seeds. The second strategy for road network completion aims at reconstructing road crossings. Basically, polygons representing the road crossings are extracted. This paper presents the proposed methodology and experimental results.
road seeds; perceptual grouping; road crossing
