Abstract: Traditional methods treat track points (lines) equally to extract road data, which ignores the spatial distribution disparity and restricts its application. Therefore, this paper proposes a new approach for map construction based on trajectory segmentation and layer fusion from vehicle tracks. First, track line subset is selected through the segmentation filtering method based on speed profile. Second, three road map layers are constructed by the Delaunay triangulation through adding different constraints according to the feature of track line subset. Third, buffer method is used to integrate multiple road layers into a single road map. An experiment using taxi GPS traces in Beijing is verified the novel method. The experimental results show that our method can extract road geometry and traffic semantic data considering the heterogeneity of trajectory, and the accuracy of result is improved compared with the two existing methods.

Key words: GPS trajectory, trajectory segmentation, data fusion, road map, Delaunay triangulation