Automatische Extraktion von Straßen aus digitalen Luftbildern

This paper presents an approach for the automatic extraction of roads from digital aerial imagery. The approach makes use of several versions of the aerial image having different resolutions. Roads are modeled as a network of intersections and links between these intersections. For d ifferent so-called „global contexts" , i .e. , rural, forest, and urban area, the model describes relations between background objects, e . g . , buildings or trees, and road objects, e .g. , road-segments. The relations, e .g. , the casting of a shadow from a tree on a road-segment, determine so­ called „local contexts". The automatic segmentation of the aerial imagery into d ifferent global contexts is used to focus the extraction to the most promising regions and additionally allows to predict in which parts of the image the results are most reliable. For the actual extraction of the roads edges are extracted in the original high resolu­ tion image (0.2 0.5 m) and lines in an image of reduced resolution. Using both resolution levels and explicit knowledge about roads, hypotheses for road-segments are generated. They are grouped iteratively to larger seg­ ments. In addition to grouping algorithms also knowledge about the local context and so-called „Ribbon-Snakes" are used to bridge gaps. To construct the road network, finally intersections are extracted. The given examples and the results of an evaluation based on manually plotted reference data indicate the potential of the approach.