BasqueRoads: A Benchmark for Road Network Selection

Guillaume Touya (1) , Azelle Courtial (1)

1 LaSTIG - Laboratoire sciences et technologies de l'information géographique

Road network selection is one of the major issues of map generalisation, as new papers are proposed every year since the first attempts of automation in the 1990’s (Thomson & Richardson, 1995). New methods are regularly proposed because selecting roads for maps at smaller scales is a complex problem. Roads are at the same time present in maps to enable car navigation tasks, and because they are structuring elements that reveal the nature of the landscape (urban, rural, mountainous...). So road selection is not only about retaining the most important roads of the network, but the preservation of topology and connectivity is essential, as well as the preservation, or the typification of road patterns (e.g. a ring road), and the preservation of local density differences (between urban and rural areas for instance). It is rare to see comparisons of road selection techniques in the literature, because of the lack of open source in map generalisation, but also because of the lack of a common dataset to benchmark these techniques; new propositions on road selection are most of the time tied to their own dataset and use case. This is why we think that this BasqueRoads dataset could be useful to advance on this topic of road network selection.

Mots clés: Map generalisation; Benchmark; Road network; Selection

Domaines: Informatique [cs]

Guillaume Touya, Azelle Courtial. BasqueRoads: A Benchmark for Road Network Selection. 241th ICA Workshop on Map Generalisation and Multiple Representation, Dec 2021, Florence, Italy. ⟨10.5194/ica-abs-4-5-2022⟩. ⟨hal-03522557⟩