Quality Assessment of OpenStreetMap Footpath Data

Sammanfattning: This project assesses the feasibility of reconstructing virtual cities from OpenStreetMap (OSM) road and building data, and will focus on the feasibility of including pedestrian footpaths. It will be examined how common it is that footpaths overlap with other map features such as roads and buildings. Feature overlap is a type of geometry error that arises to inaccuracies in the source data, which can lead to strange-looking mesh intersections and artifacts in the final city model. In this study, a slice of the OSM dataset in the Stockholm area has been taken, and a geometrical algorithm was applied to all road pairs to obtain the average amount of road overlap. It was found that footpaths were less problematic than car roads in terms of feature overlap, and thus footpaths should pose no more problems in 3D city reproduction than car roads. As a secondary study, the positional accuracy of the OSM dataset was calculated, since the positional accuracy determines the maximum tolerance in distance between points before it is certain that a road overlaps with another road or building. Any geographic dataset has a limit on its positional accuracy, which is determined by the theoretical precision of the measuring methods. In this project, the accuracy was obtained by comparing coordinate points of OSM building footprints to those in the Swedish University of Agricultural Sciences building dataset, whose accuracy is known.